Explainable Machine Learning Framework for Predicting Bond Strength of FRP Bars Embedded in UHPC

The interfacial performance of advanced composites bars embedded in Ultra-High Performance Concrete (UHPC) is an important factor that controls load transfer and the performance of structural elements. Predicting bond strength is still difficult because it is affected by several factors, such as rebar type, bar profile, bar diameter, bonded length, cover depth, fiber content, UHPC compressive strength, and FRP tensile strength. Therefore, this study uses machine-learning models to estimate the the bonding capacity of FRP bars placed in UHPC Using a collected experimental database of 183 specimens from previous studies. Four machine-learning models were developed and compared, including Linear Regression, Random Trees, Multi-Layer Perceptron, and Locally Weighted Learning. The MLP model gave the best prediction performance, with a correlation coefficient of 0.9466, MAE of 2.3083 MPa, and RMSE of 3.0631 MPa. SHAP analysis showed that embedment length was the most influential variable, followed by bar surface condition, FRP tensile strength, and concrete cover. This confirms that FRP–UHPC bond behavior is controlled by the interaction between bonded length, surface condition, mechanical interlock, and confinement provided by UHPC. Overall, the developed explainable ML framework provides a useful tool for predicting FRP–UHPC bond strength and supporting future UHPC-specific bond models.

[1]       H. Fergani, M. Di Benedetti, C. Miàs Oller, C. Lynsdale, and M. Guadagnini, “Durability and degradation mechanisms of GFRP reinforcement subjected to severe environments and sustained stress,” Constr. Build. Mater., vol. 170, pp. 637–648, May 2018, doi: 10.1016/j.conbuildmat.2018.03.092.

[2]         L. Le Gué, P. Davies, M. Arhant, B. Vincent, and W. Verbouwe, “Basalt fibre degradation in seawater and consequences for long term composite reinforcement,” Compos. Part A Appl. Sci. Manuf., vol. 179, p. 108027, Apr. 2024, doi: 10.1016/j.compositesa.2024.108027.

[3]         S. Abdal et al., “Application of Ultra-High-Performance Concrete in Bridge Engineering: Current Status, Limitations, Challenges, and Future Prospects,” Buildings, vol. 13, no. 1, p. 185, Jan. 2023, doi: 10.3390/buildings13010185.

[4]         A. Qasem, Y. S. Sallam, H. Hossam Eldien, and B. H. Ahangarn, “Bond-slip behavior between ultra-high-performance concrete and carbon fiber reinforced polymer bars using a pull-out test and numerical modelling,” Constr. Build. Mater., vol. 260, p. 119857, Nov. 2020, doi: 10.1016/j.conbuildmat.2020.119857.

[5]         C. S. Sun, N. Farzana, D. Kuruppuarachchi, M. Azimi, and H. Zhong, “Experimental and Analytical Studies of an Ultra-high Performance Concrete Beam Using Carbon Fiber Reinforced Polymer Bars,” Recent Prog. Mater., vol. 05, no. 02, pp. 1–17, May 2023, doi: 10.21926/rpm.2302019.

[6]         X. Cao, D. He, K. Qian, F. Fu, X. Deng, and L. Wang, “Shear behavior of glass FRP bars?reinforced ultra?high performance concrete I?shaped beams,” Structural Concrete, vol. 24, no. 1, pp. 1503–1520, Feb. 2023, doi: 10.1002/suco.202100801.

[7]         Y. Luo, P. Liao, R. Pan, J. Zou, and X. Zhou, “Effect of bar diameter on bond performance of helically ribbed GFRP bar to UHPC,” Journal of Building Engineering, vol. 91, p. 109577, Aug. 2024, doi: 10.1016/j.jobe.2024.109577.

[8]         R. Kumar Tipu, V. Batra, Suman, K. S. Pandya, and V. R. Panchal, “Shear capacity prediction for FRCM-strengthened RC beams using Hybrid ReLU-Activated BPNN model,” Structures, vol. 58, p. 105432, Dec. 2023, doi: 10.1016/j.istruc.2023.105432.

[9]         M. Alhusban, M. Alhusban, and A. A. Alkhawaldeh, “The Efficiency of Using Machine Learning Techniques in Fiber-Reinforced-Polymer Applications in Structural Engineering,” Sustainability, vol. 16, no. 1, p. 11, Dec. 2023, doi: 10.3390/su16010011.

[10]       M. Kova?evi?, M. Hadzima-Nyarko, P. Petronijevi?, T. Vasiljevi?, and M. Radomirovi?, “Comparative Analysis of Machine Learning Models for Predicting Interfacial Bond Strength of Fiber-Reinforced Polymer-Concrete,” Computation, vol. 13, no. 1, p. 17, Jan. 2025, doi: 10.3390/computation13010017.

[11]       C. Arndt, C. Crusenberry, B. Heng, R. Butler, and S. TerMaath, “Reduced-Dimension Surrogate Modeling to Characterize the Damage Tolerance of Composite/Metal Structures,” Modelling, vol. 4, no. 4, pp. 485–514, Nov. 2023, doi: 10.3390/modelling4040028.

[12]       D.-Y. Yoo and Y.-S. Yoon, “Bond behavior of GFRP and steel bars in ultra-high-performance fiber-reinforced concrete,” Advanced Composite Materials, vol. 26, no. 6, pp. 493–510, Nov. 2017, doi: 10.1080/09243046.2016.1197493.

[13]       Y. Luo, P. Liao, R. Pan, J. Zou, and X. Zhou, “Effect of bar diameter on bond performance of helically ribbed GFRP bar to UHPC,” Journal of Building Engineering, vol. 91, p. 109577, Aug. 2024, doi: 10.1016/j.jobe.2024.109577.

[14]       C. Wang, Y. Jiao, J. Xing, and Y. Chen, “Bond Behavior of High-Strength Steel Rebar in Ultra-High-Performance Manufactured Sand Concrete: Experiment and Modelling,” Buildings, vol. 14, no. 8, p. 2292, Jul. 2024, doi: 10.3390/buildings14082292.

[15]       Y.-W. Chan and S.-H. Chu, “Effect of silica fume on steel fiber bond characteristics in reactive powder concrete,” Cem. Concr. Res., vol. 34, no. 7, pp. 1167–1172, Jul. 2004, doi: 10.1016/j.cemconres.2003.12.023.

[16]       Z. Wu, C. Shi, and K. H. Khayat, “Influence of silica fume content on microstructure development and bond to steel fiber in ultra-high strength cement-based materials (UHSC),” Cem. Concr. Compos., vol. 71, pp. 97–109, Aug. 2016, doi: 10.1016/j.cemconcomp.2016.05.005.

[17]       L. Ke, K. Ma, Z. Chen, Y. Li, G. Chen, and Z. Feng, “Bond-slip and bond strength models for FRP bars embedded in ultra-high-performance concrete: A critical review,” Structures, vol. 64, p. 106551, Jun. 2024, doi: 10.1016/j.istruc.2024.106551.

[18]       Y. J. , & A. A. Kim, “Interface Shear between Ordinary Concrete and Ultra-High-Performance Concrete with Glass Fiber-Reinforced Polymer Reinforcing Bars,” ACI Struct. J., vol. 123, no. 3, May 2026, doi: 10.14359/51749317.

[19]       Z. Shan, K. Liang, and L. Chen, “Bond behavior of helically wound FRP bars with different surface characteristics in fiber-reinforced concrete,” Journal of Building Engineering, vol. 65, p. 105504, Apr. 2023, doi: 10.1016/j.jobe.2022.105504.

[20]       K. Liang, L. Chen, Z. Shan, and R. K. L. Su, “Experimental and theoretical study on bond behavior of helically wound FRP bars with different rib geometry embedded in ultra-high-performance concrete,” Eng. Struct., vol. 281, p. 115769, Apr. 2023, doi: 10.1016/j.engstruct.2023.115769.

[21]       R. Pan, P. Liao, J. Zou, X. Zhou, and C. Li, “Experimental investigation on bond performance of helically ribbed GFRP rebar embedded in Ultra-high performance concrete,” Constr. Build. Mater., vol. 357, Nov. 2022, doi: 10.1016/j.conbuildmat.2022.129295.

[22]       R. Pan, J. Zou, P. Liao, S. Dong, and J. Deng, “Effects of fiber content and concrete cover on the local bond behavior of helically ribbed GFRP bar and UHPC,” Journal of Building Engineering, vol. 80, p. 107939, Dec. 2023, doi: 10.1016/j.jobe.2023.107939.

[23]       F. Peng, H. Guan, J. Deng, and W. Xue, “Bond-slip behavior and model of sand-coated deformed GFRP bars in ultra-high-performance concrete,” Advances in Structural Engineering, vol. 28, no. 8, pp. 1385–1402, Jun. 2025, doi: 10.1177/13694332241310113.

[24]       F. AL-mahmoud, A. Castel, R. François, and C. Tourneur, “Effect of surface pre-conditioning on bond of carbon fibre reinforced polymer rods to concrete,” Cem. Concr. Compos., vol. 29, no. 9, pp. 677–689, Oct. 2007, doi: 10.1016/j.cemconcomp.2007.04.010.

[25]       Y. J. Kim and J. Wang, “Interface of Ultra-High-Performance Concrete with Steel, Glass Fiber-Reinforced Polymer, and Basalt Fiber-Reinforced Polymer Reinforcing Bars,” ACI Struct. J., vol. 119, no. 3, pp. 3–15, May 2022, doi: 10.14359/51732640.

[26]       F. Sayed Ahmad, G. Foret, and R. Le Roy, “Bond between carbon fibre-reinforced polymer (CFRP) bars and ultra high performance fibre reinforced concrete (UHPFRC): Experimental study,” Constr. Build. Mater., vol. 25, no. 2, pp. 479–485, Feb. 2011, doi: 10.1016/j.conbuildmat.2010.02.006.

[27]       K. M. A. Hossain, D. Ametrano, and M. Lachemi, “Bond strength of GFRP bars in ultra-high strength concrete using RILEM beam tests,” Journal of Building Engineering, vol. 10, pp. 69–79, Mar. 2017, doi: 10.1016/j.jobe.2017.02.005.

[28]       N. Khalil and J. J. Assaad, “Bond properties between smooth carbon fibre-reinforced polymer bars and ultra-high performance concrete modified with polymeric latexes and fibres,” European Journal of Environmental and Civil Engineering, vol. 26, no. 13, pp. 6211–6228, Oct. 2022, doi: 10.1080/19648189.2021.1934554.

[29]       D. Y. Yoo, K. Y. Kwon, J. J. Park, and Y. S. Yoon, “Local bond-slip response of GFRP rebar in ultra-high-performance fiber-reinforced concrete,” Compos. Struct., vol. 120, pp. 53–64, Feb. 2015, doi: 10.1016/j.compstruct.2014.09.055.

[30]       D.-Y. Yoo, J.-J. Park, S.-W. Kim, and Y.-S. Yoon, “Influence of reinforcing bar type on autogenous shrinkage stress and bond behavior of ultra high performance fiber reinforced concrete,” Cem. Concr. Compos., vol. 48, pp. 150–161, Apr. 2014, doi: 10.1016/j.cemconcomp.2013.11.014.

[31]       J. Chen, Z. Fang, X. Chen, and R. Jiang, “Experimental study on lap behavior of CFRP indented bars in UHPC,” Constr. Build. Mater., vol. 344, p. 127959, Aug. 2022, doi: 10.1016/j.conbuildmat.2022.127959.

[32]       A. Alatify and Y. J. Kim, “SP-363-8: Artificial Neural Networks for Prediction of Bond Strength between UHPC and FRP Reinforcing Bars.”

[33]       S. Abdal et al., “Application of Ultra-High-Performance Concrete in Bridge Engineering: Current Status, Limitations, Challenges, and Future Prospects,” Buildings, vol. 13, no. 1, p. 185, Jan. 2023, doi: 10.3390/buildings13010185.

[34]       C. G. Atkeson, A. W. Moore, and S. Schaal, “Locally Weighted Learning,” Artif. Intell. Rev., vol. 11, no. 1–5, pp. 11–73, Feb. 1997, doi: 10.1023/A:1006559212014.

[35]       “The WEKA workbench,” in Data Mining, Elsevier, 2017, pp. 553–571. doi: 10.1016/B978-0-12-804291-5.00024-6.