Engineering-informed machine learning for bond-strength prediction, mechanism discovery, and reliability assessment of FRP-reinforced seawater sea-sand concrete

Abstract

The bond behaviour between fiber-reinforced polymer (FRP) bars and seawater sea-sand concrete (SWSSC) governs load transfer, anchorage performance, crack development, and the structural reliability of marine concrete members. Although recent machine-learning (ML) models have demonstrated high predictive accuracy, they provide limited insight into the governing bond-transfer mechanisms and rarely quantify the prediction uncertainty required for reliability-informed structural design. This study develops an uncertainty-aware and explainable ML framework to predict the bond strength of FRP–SWSSC systems while identifying the engineering variables and behavioural mechanisms governing bond transfer. A database comprising 286 experimental bond tests collected from 17 independent studies was analysed using five ML algorithms within a leakage-free validation framework. The optimized Random Forest model achieved the highest predictive accuracy on the independent testing dataset (R² = 0.820, RMSE = 2.52 MPa, and MAE = 1.77 MPa). Engineering interpretation revealed that the bonded-length-to-diameter ratio, FRP elastic modulus, concrete compressive strength, and rib-height-to-diameter ratio are the primary parameters governing bond performance. The analyses further demonstrated that bond behaviour is controlled predominantly by nonlinear threshold responses and progressive mechanical interlocking rather than simple linear additive effects, highlighting the coupled influence of anchorage geometry, reinforcement stiffness, and concrete confinement on load-transfer efficiency. Bootstrap resampling confirmed stable predictive performance (R² = 0.777 ± 0.038), while distribution-free split-conformal prediction achieved 91.23% empirical coverage for a nominal 95% confidence level, providing calibrated uncertainty estimates for unseen engineering configurations. The proposed framework bridges data-driven prediction and structural mechanics by integrating accurate bond-strength prediction, physically interpretable mechanism discovery, and quantified predictive reliability, providing a practical basis for reliability-informed assessment and design of FRP-reinforced seawater sea-sand concrete structures.

Keywords

Fiber-reinforced polymer (FRP) / Seawater sea-sand concrete (SWSSC) / Bond strength / Machine learning / Explainable artificial intelligence / Uncertainty quantification / Structural reliability

References

[1] Fayed S, Mansour W, Elkafrawy M. Flexural response of coated steel-reinforced seawater-sea sand concrete beams incorporating black sand from Egypt. Constr Build Mater, 2025, 489: 142371.

[2] Fuqaha S, Zaki A, Nursetiawan N (2025) Performance of concrete mixed with non- potable water sources: a compressive strength investigation. In: 2nd International Graduate Conference on Smart Agriculture and Green Renewable Energy (SAGE-Grace 2025), vol 5005. pp 1–8. https://doi.org/10.1051/bioconf/202519905005

[3] Gu Y, Chu H, Zhang W (2025) Sustainable seawater-sea sand cement mortar with municipal solid waste incineration ash: mechanical performance, microstructure, and environmental benefits. Results Eng 27. https://doi.org/10.1016/j.rineng.2025.106839

[4] Huang T, Jia J, Zhang L, Liu Q. Study on the corrosion resistance of seawater sea sand concrete under the coupling action of freeze-thaw and salt corrosion with a new composite corrosion inhibitor. Constr Build Mater, 2025, 482: 141627.

[5] Guo Q, Zhang T, Liu J (2024) Study of endogenous chloride ion content on the behavior of steel bar depassivation in sea sand concrete. J Building Eng 97. https://doi.org/10.1016/j.jobe.2024.110710

[6] Zhang J, Yang J, Yu T. Durability of FRP bars and FRP-reinforced concrete beams: a critical review of accelerated aging tests and performance insights. Compos B Eng, 2026, 308: 112965.

[7] Louro AS, Cabral-Fonseca S, Firmo JP, Filipe J (2026) FRP bars for reinforced concrete: main challenges of technical assessment. LNCE 777. https://doi.org/10.1007/978-3-032-09399-8_164

[8] Zhou C, Yang B, Zhang Z, Yang Z, Zhang Z. Bond-slip behavior of the FRP bar-sea sand concrete interface and its effect on the finite element analysis of RC beam. Constr Build Mater, 2024, 436: 136917.

[9] Lin M, Weng C, Xiao H, Zeng D, Zhang B, Chen X, et al. Bond performance between fiber-wrapped ribbed basalt fiber-reinforced polymer bars and seawater sea-sand concrete. Buildings, 2024, 14138.

[10] Baena M, Barris C, Perera R, Torres L. Influence of bond characterization on load-mean strain and tension stiffening behavior of concrete elements reinforced with embedded FRP reinforcement. Materials, 2022, 153799.

[11] Wang X, Shang H, Lu Z, Xiao Z, Xu S (2025) A review on the bonding performance of FRP bars to concrete: mechanisms, testing methods, and influencing factors. Mater Rep 39. https://doi.org/10.11896/cldb.24070179

[12] Wang Z, Qi S, Li X, Zhao J. Experimental study on the bonding behavior between FRP bars and sulphoaluminate cement concrete under cyclic loading. Fuhe Cailiao Xuebao Acta Materiae Compositae Sin, 2026, 43: 978-995.

[13] Wang D, Gong Q, Yuan Q, Luo S (2021) Review of the properties of fiber-reinforced polymer-reinforced seawater-sea sand concrete. J Mater Civ Eng 33(10). https://doi.org/10.1061/(ASCE)MT.1943-5533.0003894

[14] Wei W, Liu F, Xiong Z, Lu Z, Li L. Bond performance between fibre-reinforced polymer bars and concrete under pull-out tests. Constr Build Mater, 2019, 227: 116803.

[15] Pichler N, Wang W, Li L, Ghafoori E, Motavalli M. Mixed-mode failure of adhesively bonded CFRP and iron-based shape memory alloy joints: experiments and modeling. Eng Struct, 2025, 326: 119486.

[16] Zhang R, Liu K, Jin L, Yu W, Du X (2023) Bond-Slip behavior between deformed rebar and concrete at cryogenic temperatures. J Mater Civ Eng 35(10). https://doi.org/10.1061/JMCEE7.MTENG-15616

[17] Li H, Hu G, Wang L, Wei Y, Zong Z. Bond strength between FRP bars and concrete under seawater corrosion environments: data-driven machine learning analysis. Polym Compos, 2026, 47328542865.

[18] Katlav M, Turk K. Explainable stacked machine learning models for predicting fiber-reinforced polymer (FRP) bars-concrete interfacial bond strength retention in marine environments. Eng Struct, 2026, 365: 123237.

[19] Zhang PF, Zhang D, Zhao Q, Zhao X, Tuerxunmaimaiti Y. A semantic-aware transfer learning framework via natural language processing to predict the cross-material bond strength of composite bars with limited data. Eng Appl Artif Intell, 2025, 160: 112011.

[20] Zhou Y, Wu G, Li L, Guan Z, Guo M, Yang L, et al. Experimental investigations on Bond behavior between FRP bars and advanced sustainable concrete. Polym (Basel), 2022, 1461132.

[21] Bai Z, Hong Z, Zhiqiang D, Qiang W. Enhancement of bond performance of FRP bars with seawater coral aggregate concrete by utilizing Ecoefficient slag-based alkali-activated materials. J Compos Constr, 2022, 2614021059.

[22] Yang S, Yang C, Huang M, Liu Y, Jiang J, Fan G. Study on bond performance between FRP bars and seawater coral aggregate concrete. Constr Build Mater, 2018, 173: 272-288.

[23] Gao J, Xu P, Fan L, Pietro TG. Study on Bond-Slip behavior between seawater sea-sand concrete and carbon Fiber-reinforced polymer (CFRP) bars with different surface shapes. Polym (Basel), 2022, 14132689.

[24] Shan B, Tong GQ, Liu QY. Experiment on bond performance of CFRP bars in seawater and sea sand concrete. J Archit Civ Eng, 2020, 37: 113-123

[25] Tong G. Experimental research on the bond performance of FRP bars in seawater and sea sand concrete, 2020Hunan University

[26] Yu ZL, Chang YF, Wang ZZ, Zhou Z, Ou JP (2022) Bond durability of FRP bars and seawater sea-sand concrete under real seawater immersion. Nat Sci J Hainan Univ (Nat Sci) 40. https://doi.org/10.15886/j.cnki.hdxbzkb.2022.0045

[27] Xiong Z, Mai G, Qiao S, He S, Zhang B, Wang H, et al. Fatigue bond behaviour between basalt fibre-reinforced polymer bars and seawater sea-sand concrete. Ocean Coast Manag, 2022, 218: 106038.

[28] Mai G. Investigation on fatigue bond behaviour between FRP bars and seawater sea-sand concrete, 2021Guangdong University of Technology

[29] Zhang B, Zhu H, Cao R, Ding J, Chen X. Feasibility of using geopolymers to investigate the bond behavior of FRP bars in seawater sea-sand concrete. Constr Build Mater, 2021, 282: 122636.

[30] Ruiming C, Bai Z, Luming W, Jianming D, Xianhua C. Utilizing alkali-activated materials as ordinary Portland cement replacement to study the bond performance of fiber-reinforced polymer bars in seawater sea-sand concrete. Adv Struct Eng, 2022, 25511031113.

[31] Pan Y, Yu Y, Yu J, Lu Z, Chen Y. Effects of simulated seawater on static and fatigue performance of GFRP bar-concrete bond. J Building Eng, 2023, 68: 105985.

[32] Dong ZQ, Wu G, Zhao XL, Lian JL (2018) Long-term bond durability of fiber-reinforced polymer bars embedded in seawater sea-sand concrete under ocean environments. J Compos Constr 22(5). https://doi.org/10.1061/(ASCE)CC.1943-5614.0000876

[33] Chang Y, Wang Y, Wang M, Zhou Z, Ou J. Bond durability and degradation mechanism of GFRP bars in seawater sea-sand concrete under the coupling effect of seawater immersion and sustained load. Constr Build Mater, 2021, 307: 124878.

[34] Yang S, Yang T, Sun Z, Ren Z (2022) A predictive model for determining shear strength and shear fracture energy of FRP bars in alkali-activated slag seawater coral aggregate concrete. J Building Eng 59. https://doi.org/10.1016/j.jobe.2022.105085

[35] Zhou L, Wan J, Zheng Y, Luo Y, Wang X, Wu F (2022) Study on bond behavior of GFRP bars and self-compacting concrete mixed with seawater sea-sand and high-volume fly ash. Xi’Jianzhu Keji Daxue Xuebao/J Xi’Univ Archit Technol 54. https://doi.org/10.15986/j.1006-7930.2022.02.007

[36] Xiong Z, Zeng Y, Li LG, Kwan AKH, He SH. Experimental study on the effects of glass fibres and expansive agent on the bond behaviour of glass/basalt FRP bars in seawater sea-sand concrete. Constr Build Mater, 2021, 274: 122100.

[37] Habibzadeh F. On using standard deviation or standard error of the mean. Iran J Med Sci, 2025, 50: 274-277.

[38] Fuqaha S. Seismic shear strength prediction of RC joints using shallow neural networks. Eng Perspect, 2026, 202611932.

[39] Ray DE Regression and statistical methods. 2019. https://doi.org/10.4324/9780429042959-9

[40] Kalaivani B, Ranichitra A (2024) Unveiling the impact of outliers: an improved feature engineering technique for heart disease prediction. LNNS 789. https://doi.org/10.1007/978-981-99-6586-1_32

[41] Mahaki B, Jafari S, Rezaei M, Solouki L. Types of correlation coefficients in medical science research using SPSS and R software: a review article. J Isfahan Med Sch, 2024, 42: 560-573.

[42] Kim JH. Multicollinearity and misleading statistical results. Korean J Anesthesiol, 2019, 72: 558-569.

[43] Truong O, Zhang T, Marchareddy A, Lee R, Busold J, Socas M et al. (2025) Leakagedetector 2.0: analyzing data leakage in Jupyter-Driven machine learning pipelines. In Proceedings 2025 IEEE International Conference on Software Maintenance and Evolution Icsme 2025, pp 895–899). https://doi.org/10.1109/ICSME64153.2025.00102

[44] Fuqaha S, Zaki A, Nugroho G. MACHINE LEARNING AND RSM FOR STRENGTH FORECASTING IN SUSTAINABLE SCGC. IIUM Eng J, 2025, 2635388.

[45] Sarika S, Raman A, Pillai KH (2026) Adversarial validation for identifying hidden data leakage in machine learning. CCIS 2852. https://doi.org/10.1007/978-981-95-7289-2_23

[46] Borup D, Christensen BJ, Mühlbach NS, Nielsen MS. Targeting predictors in random forest regression. Int J Forecast, 2023, 392841868.

[47] Smith HL, Biggs PJ, French NP, Smith ANH, Marshall JC. Out of (the) bag—encoding categorical predictors impacts out-of-bag samples. PeerJ Comput Sci, 2024, 10: 1-18.

[48] Freund RM, Grigas P, Mazumder R. A new perspective on boosting in linear regression via subgradient optimization and relatives. Ann Stat, 2017, 45623282364.

[49] Ikhsan J, Fuqaha S, Rahardjo AP, Suharyanto S. Machine learning for rainfall-driven debris flow prediction in data-scarce volcanic watersheds. Int J Saf Secur Eng, 2025, 151123792391.

[50] Kaneko H (2021) Support vector regression that takes into consideration the importance of explanatory variables. J Chemom 35(4). https://doi.org/10.1002/cem.3327

[51] Kocijan J, Murray-Smith R. Nonlinear predictive control with a Gaussian process model. Lect Notes Comput Sci, 2005, 3355: 185-200.

[52] Chen Z, Wang B. How priors of initial hyperparameters affect Gaussian process regression models. Neurocomputing, 2018, 275: 1702-1710.

[53] Hodson TO. Root-mean-square error (RMSE) or mean absolute error (MAE): when to use them or not. Geosci Model Dev, 2022, 151454815487.

[54] Clavière A, Kirov D, Cofer D (2026) How to verify generalization capability of a neural network with formal methods. LNCS 15947. https://doi.org/10.1007/978-3-031-99991-8_7

[55] Fuqaha S, Zaki A, Riyadi S. Compressive strength prediction of sustainable concrete incorporating non- potable water via advanced machine learning. Sustain Struct, 2025, 54124.

[56] Renaud O, Victoria-Feser M-P. A robust coefficient of determination for regression. J Stat Plan Inference, 2010, 140718521862.

[57] Fuqaha S, Nugroho G, Zaki A. Interpretable AI-Based prediction of elastic modulus in Bamboo- reinforced polypropylene using mori – Tanaka and neural networks. Diyala J Eng Sci, 2025, 8716: 104-123.

[58] Lee H, Chen S. Systematic bias of machine learning regression models and correction. IEEE Trans Pattern Anal Mach Intell, 2025, 47649744983.

[59] Karbalaie A, Abtahi F, Häger CK (2026) Participant-aware Model validation for repeated-measures data: comparative cross-validation study. Jmir AI 5. https://doi.org/10.2196/87728

[60] Fuqaha S. An explainable machine learning framework for predicting the compressive strength of sustainable metakaolin mortars. Next Mater, 2026, 13: 102723.

[61] Jan S-L, Shieh G. An improved nonparametric Test and sample size procedures for the randomized complete block designs. Sankhya B, 2025, 87: 686-711.

[62] Rayner JCW, Livingston G. Relating the Friedman test adjusted for ties, the cochran-mantel-haenszel mean score test and the ANOVA F test. Commun Stat Theory Methods, 2023, 521243694378.

[63] Kaneko H. Interpretation of machine learning models for data sets with many features using feature importance. ACS Omega, 2023, 8: 23218-23225.

[64] Eryilmaz SE, Triepels R (2025) Identification of key feature interactions via PDP decomposition. CCIS 2627. https://doi.org/10.1007/978-3-032-04339-9_10

[65] Berrar D (2025) Introduction to the Non-Parametric Bootstrap. https://doi.org/10.1016/B978-0-323-95502-7.00124-X

[66] Caro E, Juan J (2026) Probabilistic analysis of electricity load forecasting errors. Electric Power Syst Res 255. https://doi.org/10.1016/j.epsr.2026.112807

[67] Mehdiyev N, Majlatow M, Fettke P. Integrating permutation feature importance with conformal prediction for robust explainable artificial intelligence in predictive process monitoring. Eng Appl Artif Intell, 2025, 149: 110363.

[68] Olea RA, Pardo-Igúzquiza E, Dowd PA. Robust and resistant semivariogram modelling using a generalized bootstrap. J South Afr Inst Min Metall, 2015, 11513744.

[69] Barman S, R V, Ray M. Improved sieve bootstrap based prediction intervals for time series. Commun Stat Simul Comput, 2024, 531266126632.

[70] Fuqaha S. Artificial intelligence for sustainable self-healing concrete design through evolutionary algorithms. An-Najah Univ J Res - (Nat Sci), 2026, 404367380.

[71] Basnayake BRPM, Chandrasekara NV. Application of random data splitting technique for time series modeling using feedforward neural network and support vector regression models. J Appl Data Sci, 2026, 7214961515.

[72] Sauerbrei W, Boulesteix AL. Piantadosi S, Meinert CL. Use of resampling procedures to investigate issues of model Building and its stability. Principles and practice of clinical trials, 2022, Cham: Springer

[73] Degée H, Das R (2023) Load introduction and force-transfer mechanisms at steel profile-concrete interface. https://doi.org/10.1201/9781003149811-2

[74] Xu F, Lin X, Zhou A, Liu Q. Effects of recycled ceramic aggregates on internal curing of high performance concrete. Constr Build Mater, 2022, 322: 126484.

[75] Xu F, Lin X, Zhou A. Effect of recycled ceramic aggregate on hydration heat and permeability of high performance concrete. Cem Concr Compos, 2023, 137: 104930.

[76] Liao JJ, Zeng JJ, Bai YL, Zhang L. Bond strength of GFRP bars to high strength and ultra-high strength fiber reinforced seawater sea-sand concrete (SSC). Compos Struct, 2022, 281: 115013.

Rights & permissions

The Author(s)