Machine learning-based and experimental analysis of natural carbonation depth prediction in reinforced concrete with Shapley Additive Explanations

Abstract

The durability of reinforced concrete (RC) structures is significantly affected by carbonation, a process driven by material properties and environmental conditions. In this study, machine learning (ML) techniques, random forest (RF), support vector regression (SVR), and artificial neural networks (ANN), were applied to predict carbonation depth, integrating Shapley additive explanations (SHAP) to enhance model interpretability. A synthetic dataset with 20000 instances was used to evaluate ML techniques unbiasedly across diverse scenarios. A short-term experimental program under natural exposure was also conducted to validate model predictions against real carbonation data, serving as a proof-of-concept validation given the limited exposure period. The results demonstrated that compressive strength and exposure time were the most influential factors. In contrast, exposure conditions and relative humidity played an intermediate role, with indoor environments and relative humidity values between 50%–70% favoring carbonation. The type of cement and CO₂ concentration had a lesser impact, with high-pozzolanic cements accelerating carbonation, and increased CO₂ levels intensifying it. SHAP-based analyses, including beeswarm, dependence, and waterfall plots, revealed critical feature interactions and non-linear behaviors, such as Relative humidity’s parabolic trend and compressive strength’s moderating effect on CO₂ concentration. RF demonstrated greater sensitivity to feature interactions, whereas SVR and ANN captured smoother non-linear relationships, particularly for relative humidity and type of cement. By evaluating low, moderate, and high carbonation depth scenarios and confirming the consistency of predictions with experimental findings, this study demonstrated ML models’ practical applicability and ability to reflect real-world carbonation patterns. The integration of SHAP facilitated model transparency, providing a structured approach to quantify feature contributions and methodological variations. These insights reinforce the potential of ML for durability assessments, supporting data-driven strategies for carbonation modeling and structural maintenance planning in RC structures.

Graphical abstract

Keywords

artificial intelligence / service life prediction; durability / RC / explainable models / ML

References

[1] Paulpandian M K S , Neves R . Residual service life of reinforced fly ash recycled aggregate concrete structures subject to carbonation. Journal of Building Engineering, 2025, 99: 111553

[2] Hou Y L , Turcry P , Mahieux P Y , Cazacliu B , Lux J , Aït-Mokhtar A . A comparative study of CO₂ uptake quantification methods: A case study on recycled concrete aggregates under natural carbonation. Journal of Building Engineering, 2025, 101: 111845

[3] Chen L J , Su R K L . Service life modelling of carbonated reinforced concrete with supplementary cementitious materials considering early corrosion propagation. Construction and Building Materials, 2024, 413: 134861

[4] Al Fuhaid A F , Niaz A . Carbonation and corrosion problems in reinforced concrete structures. Buildings, 2022, 12(5): 586

[5] Ehsani M , Ostovari M , Mansouri S , Naseri H , Jahanbakhsh H , Nejad F M . Machine learning for predicting concrete carbonation depth: A comparative analysis and a novel feature selection. Construction and Building Materials, 2024, 417: 135331

[6] Ekolu S O . Model for practical prediction of natural carbonation in reinforced concrete: Part 1—formulation. Cement and Concrete Composites, 2018, 86: 40–56

[7] Tong L Y , Cai Y X , Liu Q F . Carbonation modelling of hardened cementitious materials considering pore structure characteristics: A review. Journal of Building Engineering, 2024, 96: 110547

[8] Jia H X , Qiao G F , Han P . Machine learning algorithms in the environmental corrosion evaluation of reinforced concrete structures—A review. Cement and Concrete Composites, 2022, 133: 104725

[9] Wang X H , Yang Q W , Peng X , Qin F J . A review of concrete carbonation depth evaluation models. Coatings, 2024, 14(4): 386

[10] Fu B , Lei H , Ullah I , El-Meligy M , El Hindi K , Javed M F , Ahmad F . Predictive modeling for durability characteristics of blended cement concrete utilizing machine learning algorithms. Case Studies in Construction Materials, 2025, 22: e04209

[11] Hachaj T , Piekarczyk M . On explainability of reinforcement learning-based machine learning agents trained with proximal policy optimization that utilizes visual sensor data. Applied Sciences, 2025, 15(2): 538

[12] Liu X Y , Sun G Z , Ju R Q , Li J , Li Z L , Jiang Y L , Zhao K , Zhang Y , Jing Y C , Yang G T . Prediction of load-bearing capacity of FRP-steel composite tubed concrete columns: Using explainable machine learning model with limited data. Structures, 2025, 71: 107890

[13] Peng Y M , Unluer C . Interpretable machine learning-based analysis of hydration and carbonation of carbonated reactive magnesia cement mixes. Journal of Cleaner Production, 2024, 434: 140054

[14] Kulasooriya W K V J B , Ranasinghe R S S , Perera U S , Thisovithan P , Ekanayake I U , Meddage D P P . Modeling strength characteristics of basalt fiber reinforced concrete using multiple explainable machine learning with a graphical user interface. Scientific Reports, 2023, 13(1): 13138

[15] Guidotti R , Monreale A , Ruggieri S , Turini F , Giannotti F , Pedreschi D . A survey of methods for explaining black box models. ACM Computing Surveys (CSUR), 2019, 51(5): 93

[16] Doshi-Velez F, Kim B. Towards a rigorous science of interpretable machine learning. 2017, arXiv: 1702.08608

[17] Du M N , Liu N H , Hu X . Techniques for interpretable machine learning. Communications of the ACM, 2019, 63(1): 68–77

[18] Lundberg S M, Lee S I. A unified approach to interpreting model predictions. In: Proceedings of the 31st International Conference on Neural Information Processing Systems. Long Beach: Curran Associates Inc., 2017, 4765–4774

[19] Mangalathu S , Hwang S H , Jeon J S . Failure mode and effects analysis of RC members based on machine-learning-based SHapley Additive exPlanations (SHAP) approach. Engineering Structures, 2020, 219: 110927

[20] Harrison M . Machine Learning Pocket Reference: Working with Structured Data in Python. North Sebastopol: O’Reilly Media, 2019,

[21] de Oliveira Lopes B G C. Explain-ML: A Human-Centered Holistic and Interactive Visual Tool for Explainable Machine Learning. Thesis for the Master’s Degree. Belo Horizonte: Universidade Federal de Minas Gerais, 2020

[22] Ukwaththa J , Herath S , Meddage D P P . A review of machine learning (ML) and explainable artificial intelligence (XAI) methods in additive manufacturing (3D Printing). Materials Today Communications, 2024, 41: 110294

[23] Meddage D P P , Ekanayake I U , Weerasuriya A U , Lewangamage C S , Tse K T , Miyanawala T P , Ramanayaka C D E . Explainable Machine Learning (XML) to predict external wind pressure of a low-rise building in urban-like settings. Journal of Wind Engineering and Industrial Aerodynamics, 2022, 226: 105027

[24] Huo Z H , Wang L , Huang Y M . Predicting carbonation depth of concrete using a hybrid ensemble model. Journal of Building Engineering, 2023, 76: 107320

[25] Wang J , Zhang Z Q , Liu X M , Shao Y , Liu X Y , Wang H M . Prediction and interpretation of concrete corrosion induced by carbon dioxide using machine learning. Corrosion Science, 2024, 233: 112100

[26] Chen X Y , Liu X , Cheng S K , Bian X Y , Bai X X , Zheng X , Xu X , Xu Z F . Machine learning-based modelling and analysis of carbonation depth of recycled aggregate concrete. Case Studies in Construction Materials, 2025, 22: e04162

[27] Wang D C , Tan Q H , Wang Y R , Liu G Y , Lu Z , Zhu C Q , Sun B C . Carbonation depth prediction and parameter influential analysis of recycled concrete buildings. Journal of CO₂ Utilization, 2024, 85: 102877

[28] Golafshani E M , Behnood A , Kim T , Ngo T , Kashani A . Metaheuristic optimization based-ensemble learners for the carbonation assessment of recycled aggregate concrete. Applied Soft Computing, 2024, 159: 111661

[29] Golafshani E M , Behnood A , Kim T , Ngo T , Kashani A . A framework for low-carbon mix design of recycled aggregate concrete with supplementary cementitious materials using machine learning and optimization algorithms. Structures, 2024, 61: 106143

[30] Khan M , Javed M F . Towards sustainable construction: Machine learning based predictive models for strength and durability characteristics of blended cement concrete. Materials Today Communications, 2023, 37: 107428

[31] Ekanayake I U , Meddage D P P , Rathnayake U . A novel approach to explain the black-box nature of machine learning in compressive strength predictions of concrete using Shapley additive explanations (SHAP). Case Studies in Construction Materials, 2022, 16: e01059

[32] Alyousef R , Rehman M F , Khan M , Fawad M , Khan A U , Hassan A M , Ghamry N A . Machine learning-driven predictive models for compressive strength of steel fiber reinforced concrete subjected to high temperatures. Case Studies in Construction Materials, 2023, 19: e02418

[33] Meng S Q , Shi Z M , Xia C Z , Zhou C S , Zhao Y Z . Exploring LightGBM-SHAP: Interpretable predictive modeling for concrete strength under high temperature conditions. Structures, 2025, 71: 108134

[34] Zheng X L , Xie Y , Yang X J , Amin M N , Nazar S , Khan S A , Althoey F , Deifalla A F . A data-driven approach to predict the compressive strength of alkali-activated materials and correlation of influencing parameters using SHapley Additive exPlanations (SHAP) analysis. Journal of Materials Research and Technology, 2023, 25: 4074–4093

[35] Meddage D P P , Fonseka I , Mohotti D , Wijesooriya K , Lee C K . An explainable machine learning approach to predict the compressive strength of graphene oxide-based concrete. Construction and Building Materials, 2024, 449: 138346

[36] Taffese W Z , Espinosa-Leal L . Unveiling non-steady chloride migration insights through explainable machine learning. Journal of Building Engineering, 2024, 82: 108370

[37] Khan M , Nassar R U D , Khan A U , Houda M , El Hachem C , Rasheed M , Anwar W . Optimizing durability assessment: Machine learning models for depth of wear of environmentally-friendly concrete. Results in Engineering, 2023, 20: 101625

[38] Martínez-Muñoz D , García J , Martí J V , Yepes V . Deep learning classifier for life cycle optimization of steel−concrete composite bridges. Structures, 2023, 57: 105347

[39] Possan E . Modelagem da Carbonatação e Previsão de Vida Útil de Estruturas de Concreto em Ambiente Urbano. Dissertation for the Doctoral Degree, 2010,

[40] Couto R A , Reis E D , Campos I A G , Dalip D H , Poggiali F S J , Ludvig P . A systematic review of machine learning applications in carbonation depth prediction for reinforced concrete structures. Journal of Building Pathology and Rehabilitation, 2026, 11(1): 28

[41] George K , Ziska L H , Bunce J A , Quebedeaux B . Elevated atmospheric CO₂ concentration and temperature across an urban-rural transect. Atmospheric Environment, 2007, 41: 7654–7665

[42] Park H , Jeong S , Koo J H , Sim S , Bae Y , Kim Y , Park C , Bang J . Lessons from COVID-19 and Seoul: Effects of reduced human activity from social distancing on urban CO₂ concentration and air quality. Aerosol and Air Quality Research, 2021, 21(1): 200376

[43] Metya A , Datye A , Chakraborty S , Tiwari Y K , Sarma D , Bora A , Gogoi N . Diurnal and seasonal variability of CO₂ and CH₄ concentration in a semi-urban environment of western India. Scientific Reports, 2021, 11(1): 2931

[44] Qi B , Hu C , Yu Y , Pang Y T , Wu F Q , Yang X J , Liu H L , Zhang J Q , Xiao Q T , Liu C . et al. Using urban-suburban difference of atmospheric CO₂ to evaluate carbon neutrality capacity in Hangzhou, China. Journal of Environmental Sciences, 2025, 157: 908–923

[45] Torres-Acosta A A , Castro-Borges P , de Rincón O T , Martín-Pérez B , Martínez-Molina W , Alonso-Guzmán E , Bernabé-Reyes C , Pérez-López T , Arista-Perrusquía C , Crespo-Sánchez S E . et al. Concrete carbonation in tropical urban and urban/marine environments after 20 years of natural exposure. Construction and Building Materials, 2024, 450: 138511

[46] Neville A M . Properties of Concrete. London: Pitnam, 1963,

[47] Brasil-ABNT NBR . Concreto Para Fins Estruturais— Classificação Pela Massa Específica, por Grupos de Resistência e Consistência, ABNT NBR 8953:2015. Rio de Janeiro: Associação Brasileira de Normas Técnicas, 2015,

[48] Elsalamawy M , Mohamed A R , Kamal E M . The role of relative humidity and cement type on carbonation resistance of concrete. Alexandria Engineering Journal, 2019, 58(4): 1257–1264

[49] Brasil-ABNT NBR . Cimento Portland—Requisitos, ABNT NBR 16697:2018. Rio de Janeiro: Associação Brasileira de Normas Técnicas, 2018,

[50] Brasil-ABNT NBR . Projeto de Estruturas de Concreto— Procedimento, ABNT NBR 6118:2023. Rio de Janeiro: Associação Brasileira de Normas Técnicas, 2023,

[51] Biau G , Scornet E . A random forest guided tour. TEST, 2016, 25(2): 197–227

[52] Zhang F , O’Donnell L J . Support vector regression. In: Mechelli A, Vieira S, eds. Machine Learning: Methods and Applications to Brain Disorders, 2020, 123–140

[53] Bengio Y. Practical recommendations for gradient-based training of deep architectures. In: Montavon G, Orr G B, Müller K R, eds. Neural Networks: Tricks of the Trade. 2nd ed. Berlin: Springer, 2012, 437–478

[54] Belgium-EN . Cement—Part 1: Composition, Specifications and Conformity Criteria for Common Cements, EN 197–1:2011. Brussels: European Committee for Standardization, 2011,

[55] Brasil-NBR NM . Agregados—Determinação da Composição Granulométrica, NBR NM 248:2003. , 2003,

[56] Brasil-ABNT NBR . Agregados—Determinação da Massa Específica de Agregados Miúdos por Meio do Frasco Chapman, ABNT NBR 9776:1987. Rio de Janeiro: Associação Brasileira de Normas Técnicas, 1987,

[57] Brasil-ABNT NBR . Concreto—Procedimento Para Moldagem e Cura de Corpos de Prova, ABNT NBR 5738: 2015. Rio de Janeiro: Associação Brasileira de Normas Técnicas, 2015,

[58] US-ASTM. Standard Test Method for Density, Absorption, and Voids in Hardened Concrete, ASTM C642-21. West Conshohocken: ASTM, 2022

[59] Belgium-EN. Testing Hardened Concrete—Part 3: Compressive Strength of Test Specimens, EN 12390-3:2019. Brussels: European Committee for Standardization, 2019

[60] Instituto Nacional de Meteorologia (INMET). Dados Históricos Anuais. 2025

[61] Massana J , Reyes E , Bernal J , León N , Sánchez-Espinosa E . Influence of nano- and micro-silica additions on the durability of a high-performance self-compacting concrete. Construction and Building Materials, 2018, 165: 93–103

[62] Cantero B , Bravo M , de Brito J , del Bosque I F S , Medina C . Assessment of the permeability to aggressive agents of concrete with recycled cement and mixed recycled aggregate. Applied Sciences, 2021, 11(9): 3856

[63] Majlesi A , Koodiani H K , de Rincón O T , Montoya A , Millano V , Torres-Acosta A A , Troconis B C R . Artificial neural network model to estimate the long-term carbonation depth of concrete exposed to natural environments. Journal of Building Engineering, 2023, 74: 106545

[64] Parrott L J , Killoh D C . Carbonation in a 36 year old, in-situ concrete. Cement and Concrete Research, 1989, 19(4): 649–656

[65] Li D W , Chen B , Sun H F , Memon S A , Deng X B , Wang Y C , Xing F . Evaluating the effect of external and internal factors on carbonation of existing concrete building structures. Construction and Building Materials, 2018, 167: 73–81

[66] Mahmood A , Dehn F , Thissen P . Carbonation under varying humidity conditions: A 3D micro-scale kinetic Monte Carlo approach. Langmuir, 2025, 41(4): 2259–2268

[67] Chen Y , Liu P , Yu Z W . Effects of environmental factors on concrete carbonation depth and compressive strength. Materials, 2018, 11(11): 2167

[68] Falcão J , Melo R , Masuero A , Dal Molin D . Influence of dregs waste on the alkali–silica reaction: A comparative analysis among different types of cement. Sustainability, 2024, 16(19): 8610

[69] Witkowski H , Koniorczyk M . The influence of pozzolanic additives on the carbonation rate and life cycle inventory of concrete. Construction and Building Materials, 2020, 254: 119301

[70] Cui H Z , Tang W C , Wei L , Dong Z J , Xing F . Experimental study on effects of CO₂ concentrations on concrete carbonation and diffusion mechanisms. Construction and Building Materials, 2015, 93: 522–527

[71] Han S H , Park W S , Yang E I . Evaluation of concrete durability due to carbonation in harbor concrete structures. Construction and Building Materials, 2013, 48: 1045–1049

[72] Lollini F , Redaelli E . Carbonation of blended cement concretes after 12 years of natural exposure. Construction and Building Materials, 2021, 276: 122122

[73] von Greve-Dierfeld S , Lothenbach B , Vollpracht A , Wu B , Huet B , Andrade C , Medina C , Thiel C , Gruyaert E , Vanoutrive H . et al. Understanding the carbonation of concrete with supplementary cementitious materials: A critical review by RILEM TC 281-CCC. Materials and Structures, 2020, 53(6): 136

[74] Tuutti K. Corrosion of Steel in Concrete. Dissertation for the Doctoral Degree. Stockholm: Swedish Cement and Concrete Research Institute, 1982

[75] Papadakis V G , Vayenas C G , Fardis M N . A reaction engineering approach to the problem of concrete carbonation. AIChE Journal, 1989, 35(10): 1639–1650

[76] Papadakis V G , Vayenas C G , Fardis M N . Fundamental modeling and experimental investigation of concrete carbonation. Materials Journal, 1991, 88(4): 363–373

[77] Couto R A , Campos I A G , Reis E D , Dalip D H , Poggiali F S J , Ludvig P . Machine learning models for carbonation depth prediction in reinforced concrete structures: A comparative study. Modelling, 2025, 6(2): 46

[78] Tang S W , Wang Y , Geng Z C , Xu X F , Yu W Z , A H , Chen J T . Structure, fractality, mechanics and durability of calcium silicate hydrates. Fractal and Fractional, 2021, 5(2): 47

[79] Xi X Y , Zheng Y X , Du C W , Zhang P , Sun M . Study on the hydration characteristics, mechanical properties, and microstructure of thermally activated low-carbon recycled cement. Construction and Building Materials, 2024, 447: 138042

[80] Kim J , Na S , Hama Y . Effect of blast-furnace slag replacement ratio and curing method on pore structure change after carbonation on cement paste. Materials, 2020, 13(21): 4787

[81] Shah V , Bishnoi S . Understanding the process of carbonation in concrete using numerical modeling. Journal of Advanced Concrete Technology, 2021, 19(11): 1148–1161

[82] Reis E D , Resende H F , Christoforo A L , Costa R M , Gatuingt F , Poggiali F S J , Bezerra A C S . Assessment of physical and mechanical properties of concrete with carbon nanotubes pre-dispersed in cement. Journal of Building Engineering, 2024, 89: 109255

[83] Reis E D , Gatuingt F , Poggiali F S J , Bezerra A C S . Influence of carbon nanotubes and cement reduction on the bond strength of steel bars in concrete: An experimental investigation. Construction and Building Materials, 2025, 461: 139940

[84] Wu B , Ye G . Development of porosity of cement paste blended with supplementary cementitious materials after carbonation. Construction and Building Materials, 2017, 145: 52–61

[85] Shen Q Z , Pan G H , Bao B F . Influence of CSH carbonation on the porosity of cement paste. Magazine of Concrete Research, 2016, 68(10): 504–514

[86] Ngala V T , Page C L . Effects of carbonation on pore structure and diffusional properties of hydrated cement pastes. Cement and Concrete Research, 1997, 27(7): 995–1007

[87] Shi P , Falliano D , Yang Z X , Marano G C , Briseghella B . Investigation on the anti-carbonation properties of alkali-activated slag concrete: Effect of activator types and dosages. Journal of Building Engineering, 2024, 91: 109552

[88] Ding Z Z , Quy N X , Kim J , Hama Y . Evaluations of frost and scaling resistance of fly ash concrete in terms of changes in water absorption and pore structure under the accelerated carbonation conditions. Construction and Building Materials, 2022, 345: 128273

[89] Shah V , Mackechnie J , Scott A . Determination of carbonation resistance of concrete through a combination of cement content and tortuosity. Journal of Building Engineering, 2022, 60: 105176

[90] Chen C H , Liu R G , Zhu P H , Liu H , Wang X J . Carbonization durability of two generations of recycled coarse aggregate concrete with effect of chloride ion corrosion. Sustainability, 2020, 12(24): 10544

[91] Bretti G , Ceseri M , Natalini R , Ciacchella M C , Santarelli M L , Tiracorrendo G . A forecasting model for the porosity variation during the carbonation process. GEM-International Journal on Geomathematics, 2022, 13(1): 13

Rights & permissions

Higher Education Press