pith. sign in

arxiv: 2606.06765 · v1 · pith:V6XRRG7Knew · submitted 2026-06-04 · ❄️ cond-mat.mtrl-sci · cs.LG

Reactivity-Informed Machine Learning for Performance Prediction and Design Space Exploration of Alkali-Activated Slag

Qiyao He , Zhanzhao Li , Kai Gong This is my paper

Pith reviewed 2026-06-28 00:06 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci cs.LG
keywords alkali-activated slagmachine learningcompressive strength predictionreactivity descriptordesign space explorationembodied CO2ground granulated blast-furnace slag
0
0 comments X

The pith

A reactivity descriptor called AMODE lets machine learning predict alkali-activated slag strength from heterogeneous data while mapping low-emission high-strength designs.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper assembles over 3100 compressive strength records from 155 distinct ground granulated blast-furnace slag sources to train and benchmark machine learning models. It demonstrates that adding precursor chemistry, fineness, curing conditions, and specimen geometry improves prediction accuracy across several algorithms. Replacing detailed oxide compositions with the average metal oxide dissociation energy maintains comparable performance while yielding a more compact and physically interpretable input. Model interpretations recover expected physical trends such as strength drops with higher water content or larger specimens. Statistically constrained exploration then locates design regions that deliver high strength together with lower embodied CO2 emissions than ordinary Portland cement references at similar cost.

Core claim

Integrating the average metal oxide dissociation energy as a reactivity descriptor enables machine learning models trained on heterogeneous literature data to achieve predictive performance comparable to models that use explicit oxide compositions, to extract physically consistent trends from that data, and to reveal reactivity-dependent trade-offs that identify high-strength alkali-activated slag regions with substantially lower CO2 emissions than OPC-based references at similar cost.

What carries the argument

The average metal oxide dissociation energy (AMODE), a single physically interpretable value that condenses precursor reactivity from metal oxide properties and serves as a compact substitute for full multi-oxide composition inputs in the machine learning feature sets.

If this is right

  • Machine learning models achieve comparable predictive performance when the average metal oxide dissociation energy replaces explicit oxide composition inputs.
  • Model interpretation recovers physically consistent trends such as non-monotonic activator effects, reduced strength at higher water content, and reduced strength at larger specimen sizes.
  • Statistically constrained design space exploration reveals reactivity-dependent trade-offs among strength, embodied CO2 emissions, and cost.
  • Design maps locate high-strength regions that achieve substantially lower embodied CO2 emissions than ordinary Portland cement references at comparable cost.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The public dataset release could support independent retraining or extension of the models by other groups working on similar materials.
  • If AMODE generalizes beyond the current slag sources, it could reduce the experimental burden of full compositional analysis when screening new precursors.
  • The same reactivity-informed workflow might be tested on other alkali-activated binder systems to check whether the observed design trade-offs hold more broadly.

Load-bearing premise

The literature-derived collection of more than 3100 records from 155 chemically distinct slag sources can be treated as a consistent, unbiased sample of real material behavior despite differences in experimental protocols and reporting across the source studies.

What would settle it

New laboratory compressive strength measurements on alkali-activated slag mixes prepared with ground granulated blast-furnace slag sources absent from the original 155-source set, checked against model predictions within the reported error bounds.

read the original abstract

Establishing quantitative relationships among mix design, raw material properties, curing conditions, and performance remains a long-standing challenge in cementitious materials, particularly for alkali-activated materials with variable precursor and activator chemistry. Here, we curated the largest literature-derived alkali-activated slag (AAS) dataset to date, comprising over 3100 compressive strength records, 155 chemically distinct ground granulated blast-furnace slags (GGBSs), and 24 attributes incorporating precursor chemistry, fineness, and reactivity. Multiple machine learning (ML) algorithms were benchmarked across progressively enriched feature scenarios, demonstrating that integrating GGBS compositions, fineness, curing conditions, and specimen geometry improves predictive performance. The average metal oxide dissociation energy (AMODE), a physically interpretable representation of precursor reactivity, provides a compact alternative descriptor to explicit oxide compositions while enabling comparable predictive performance. Model interpretation revealed physically consistent trends from heterogeneous data, including non-monotonic effects of Na2O dosage and silicate modulus, reduced predicted strength at higher water content and larger specimen size, and coupled oxide-level effects more coherently represented by AMODE than by individual oxide contents. Statistically constrained design space exploration reveals reactivity-dependent trade-offs among strength, embodied CO2 emissions, and cost. The design maps identify high-strength regions with substantially lower CO2 emissions than OPC-based references at similar cost. Overall, this work demonstrates how reactivity-informed ML can extract physically meaningful trends from heterogeneous AAS data and guide source-dependent binder design. The curated dataset is publicly accessible to support advances in cement and concrete research.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript curates the largest literature-derived AAS dataset to date (3100+ compressive strength records from 155 chemically distinct GGBS sources with 24 attributes) and benchmarks multiple ML algorithms across feature scenarios that progressively add GGBS composition, fineness, curing conditions, and geometry. It introduces the average metal oxide dissociation energy (AMODE) as a compact, physically interpretable reactivity descriptor that achieves comparable predictive performance to explicit oxide compositions, extracts physically consistent trends (non-monotonic Na2O and silicate modulus effects, water and size dependencies), and performs statistically constrained design-space exploration to map reactivity-dependent trade-offs among strength, embodied CO2, and cost, identifying high-strength regions with lower CO2 than OPC references at comparable cost. The curated dataset is released publicly.

Significance. If the central claims hold, the work supplies a valuable public dataset at unprecedented scale for cementitious materials research and demonstrates how a physically motivated descriptor (AMODE) can support interpretable ML that extracts trends from heterogeneous data while guiding low-emission binder design. The public data release and the attempt to link reactivity to multi-objective optimization are clear strengths.

major comments (3)
  1. [ML benchmarking results (abstract and § on model performance)] The abstract and results on ML benchmarking report improved predictive performance across feature scenarios and comparable performance with AMODE versus explicit oxides, yet supply no quantitative error metrics (RMSE, MAE, R²), cross-validation protocol, data exclusion criteria, or held-out experimental validation; this absence is load-bearing for the claims of benchmarking success and AMODE utility.
  2. [Dataset curation and interpretation sections] The central premise that the 3100+ records from 155 GGBS sources constitute a consistent, unbiased sample is required for the interpretation of physically consistent trends and the reactivity-dependent design maps, but the curation pipeline provides no quantification of inter-study protocol variance, sensitivity checks, or bias assessment; without these, the non-monotonic Na2O effects and AMODE coherence claims rest on an untested uniformity assumption.
  3. [Design space exploration section] The statistically constrained design-space maps identify high-strength, low-CO2 regions relative to OPC, but these maps inherit the same unverified data-consistency assumption; no sensitivity analysis to curation choices or external validation is described, making the trade-off conclusions load-bearing on the weakest assumption identified in the review.
minor comments (2)
  1. [Abstract] The abstract is information-dense; consider separating the dataset description, benchmarking claims, and design results into clearer sentences.
  2. [Methods or feature definition section] Notation for AMODE and the 24 attributes should be defined at first use with an explicit equation or table reference.

Simulated Author's Rebuttal

3 responses · 1 unresolved

Thank you for the thorough review. We respond to each major comment below and indicate where revisions will be made to address the concerns about quantitative metrics, data consistency, and sensitivity analyses.

read point-by-point responses

  1. Referee: The abstract and results on ML benchmarking report improved predictive performance across feature scenarios and comparable performance with AMODE versus explicit oxides, yet supply no quantitative error metrics (RMSE, MAE, R²), cross-validation protocol, data exclusion criteria, or held-out experimental validation; this absence is load-bearing for the claims of benchmarking success and AMODE utility.

    Authors: We will revise the abstract to include key performance metrics such as R², RMSE, and MAE, and add explicit details on the cross-validation protocol (e.g., k-fold) and data exclusion criteria in the methods section. Held-out experimental validation is not included as the study relies on literature data; we will add a limitations section noting this as a future direction. revision: partial

  2. Referee: The central premise that the 3100+ records from 155 GGBS sources constitute a consistent, unbiased sample is required for the interpretation of physically consistent trends and the reactivity-dependent design maps, but the curation pipeline provides no quantification of inter-study protocol variance, sensitivity checks, or bias assessment; without these, the non-monotonic Na2O effects and AMODE coherence claims rest on an untested uniformity assumption.

    Authors: We will expand the dataset curation section to include a discussion of inter-study variances and add sensitivity checks by evaluating model performance on data subsets stratified by publication source or year. This will help validate the robustness of the extracted trends and AMODE utility. revision: yes

  3. Referee: The statistically constrained design-space maps identify high-strength, low-CO2 regions relative to OPC, but these maps inherit the same unverified data-consistency assumption; no sensitivity analysis to curation choices or external validation is described, making the trade-off conclusions load-bearing on the weakest assumption identified in the review.

    Authors: We will include sensitivity analyses in the design space exploration section, testing variations in data curation parameters and their impact on the identified optimal regions for strength, CO2, and cost. This will strengthen the reliability of the trade-off maps. revision: yes

standing simulated objections not resolved
  • Provision of held-out experimental validation, as this would require conducting new experiments outside the scope of the current literature-based study.

Circularity Check

0 steps flagged

No significant circularity; AMODE is physically defined from external data and central claims rest on literature curation plus standard ML.

full rationale

The paper curates an external literature-derived dataset (3100+ records from 155 GGBS sources) and defines AMODE explicitly as the average metal oxide dissociation energy, a physical quantity computed from precursor chemistry rather than fitted to compressive strength targets. ML benchmarking, interpretation of trends (e.g., non-monotonic Na2O effects), and statistically constrained design-space exploration all operate on this independent input; no equations or steps reduce the claimed predictions or design maps to the fitted outputs by construction. No self-citation load-bearing steps, uniqueness theorems, or ansatz smuggling appear in the provided text. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim depends on the representativeness of heterogeneous literature data and on the physical validity of AMODE as a reactivity proxy; no new entities are postulated and no parameters are fitted directly to strength in the reported workflow.

free parameters (1)
  • ML algorithm hyperparameters and feature-selection thresholds
    Standard in any ML pipeline; not enumerated in the abstract but required to reproduce the reported performance gains.
axioms (2)
  • domain assumption Literature-reported compressive strength values are comparable across studies despite differences in specimen geometry, curing protocols, and testing standards.
    Invoked by the decision to pool 3100+ records into a single training set.
  • standard math Average metal oxide dissociation energy can be computed from tabulated oxide properties without additional fitting to the AAS strength data.
    Required for AMODE to function as an independent, physically interpretable descriptor.

pith-pipeline@v0.9.1-grok · 5820 in / 1556 out tokens · 32639 ms · 2026-06-28T00:06:11.208094+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

171 extracted references · 147 canonical work pages · 1 internal anchor

  1. [1]

    Monteiro, S.A

    P.J.M. Monteiro, S.A. Miller, A. Horvath, Towards sustainable concrete, Nat. Mater. 16 (2017) 698–699. https://doi.org/10.1038/nmat4930

    work page doi:10.1038/nmat4930 2017

  2. [2]

    Miller, V .M

    S.A. Miller, V .M. John, S.A. Pacca, A. Horvath, Carbon dioxide reduction potential in the global cement industry by 2050, Cem. Concr. Res. 114 (2018) 115–124. https://doi.org/10.1016/j.cemconres.2017.08.026

    work page doi:10.1016/j.cemconres.2017.08.026 2050

  3. [3]

    DeRousseau, J.R

    M.A. DeRousseau, J.R. Kasprzyk, W.V . Srubar, Computational design optimization of concrete mixtures: A review, Cem. Concr. Res. 109 (2018) 42–53. https://doi.org/10.1016/j.cemconres.2018.04.007

    work page doi:10.1016/j.cemconres.2018.04.007 2018

  4. [4]

    Pfeiffer, K

    O.P. Pfeiffer, K. Gong, K.A. Severson, J. Chen, J.R. Gregory, S. Ghosh, R.T. Goodwin, E.A. Olivetti, Bayesian design of concrete with amortized Gaussian processes and multi - objective optimization, Cem. Concr. Res. 177 (2024) 107406. https://doi.org/10.1016/j.cemconres.2023.107406

    work page doi:10.1016/j.cemconres.2023.107406 2024

  5. [5]

    Lothenbach, K

    B. Lothenbach, K. Scrivener, R.D. Hooton, Supplementary cementitious materials, Cem. Concr. Res. 41 (2011) 1244–1256. https://doi.org/10.1016/j.cemconres.2010.12.001

    work page doi:10.1016/j.cemconres.2010.12.001 2011

  6. [6]

    Juenger, R

    M.C.G. Juenger, R. Snellings, S.A. Bernal, Supplementary cementitious materials: New sources, characterization, and performance insights, Cem. Concr. Res. 122 (2019) 257–273. https://doi.org/10.1016/j.cemconres.2019.05.008

    work page doi:10.1016/j.cemconres.2019.05.008 2019

  7. [7]

    Shi, A.F

    C. Shi, A.F. Jiménez, A. Palomo, New cements for the 21st century: The pursuit of an alternative to Portland cement, Cem. Concr. Res. 41 (2011) 750–763. https://doi.org/10.1016/j.cemconres.2011.03.016

    work page doi:10.1016/j.cemconres.2011.03.016 2011

  8. [8]

    Juenger, F

    M.C.G. Juenger, F. Winnefeld, J.L. Provis, J.H. Ideker, Advances in alternative cementitious binders, Cem. Concr. Res. 41 (2011) 1232–1243. https://doi.org/10.1016/j.cemconres.2010.11.012

    work page doi:10.1016/j.cemconres.2010.11.012 2011

  9. [9]

    Provis, Alkali -activated materials, Cem

    J.L. Provis, Alkali -activated materials, Cem. Concr. Res. 114 (2018) 40–48. https://doi.org/10.1016/j.cemconres.2017.02.009

    work page doi:10.1016/j.cemconres.2017.02.009 2018

  10. [10]

    Kriven, C

    W.M. Kriven, C. Leonelli, J.L. Provis, A.R. Boccaccini, C. Attwell, V .S. Ducman, C. Ferone, S. Rossignol, T. Luukkonen, J.S.J. Van Deventer, J.V . Emiliano, J.E. Lombardi, Why geopolymers and alkali‐activated materials are key components of a sustainable w orld: A perspective contribution, J. Am. Ceram. Soc. 107 (2024) 5159–5177. https://doi.org/10.1111/...

    work page doi:10.1111/jace.19828 2024

  11. [11]

    Provis, S.A

    J.L. Provis, S.A. Bernal, Geopolymers and related alkali -activated materials, Annu. Rev. Mater. Res. 44 (2014) 299–327. https://doi.org/10.1146/annurev-matsci-070813-113515

    work page doi:10.1146/annurev-matsci-070813-113515 2014

  12. [12]

    Asadi Shamsabadi, M

    E. Asadi Shamsabadi, M. Salehpour, P. Zandifaez, D. Dias -da-Costa, Data- driven multicollinearity-aware multi-objective optimisation of green concrete mixes, J. Clean. Prod. 390 (2023) 136103. https://doi.org/10.1016/j.jclepro.2023.136103. 57

    work page doi:10.1016/j.jclepro.2023.136103 2023

  13. [13]

    Mahjoubi, R

    S. Mahjoubi, R. Barhemat, W. Meng, Y . Bao, AI-guided auto-discovery of low-carbon cost- effective ultra-high performance concrete (UHPC), Resour. Conserv. Recycl. 189 (2023) 106741. https://doi.org/10.1016/j.resconrec.2022.106741

    work page doi:10.1016/j.resconrec.2022.106741 2023

  14. [14]

    Taffese, B

    W.Z. Taffese, B. Hilloulin, Y . Villagran Zaccardi, A. Marani, M.L. Nehdi, M.U. Hanif, M. Kamath, S. Nunes, S. V on Greve-Dierfeld, A. Kanellopoulos, Machine learning in concrete durability: challenges and pathways identified by RILEM TC 315- DCS towards enhanced predictive models, Mater. Struct. 58 (2025) 145. https://doi.org/10.1617/s11527-025-02664- 3

    work page doi:10.1617/s11527-025-02664- 2025

  15. [15]

    Nunez, A

    I. Nunez, A. Marani, M. Flah, M.L. Nehdi, Estimating compressive strength of modern concrete mixtures using computational intelligence: A systematic review, Constr. Build. Mater. 310 (2021) 125279. https://doi.org/10.1016/j.conbuildmat.2021.125279

    work page doi:10.1016/j.conbuildmat.2021.125279 2021

  16. [16]

    D. Luo, K. Wang, D. Wang, A. Sharma, W. Li, I.H. Choi, Artificial intelligence in the design, optimization, and performance prediction of concrete materials: A comprehensive review, Npj Mater. Sustain. 3 (2025) 14. https://doi.org/10.1038/s44296-025-00058-8

    work page doi:10.1038/s44296-025-00058-8 2025

  17. [17]

    Ben Chaabene, M

    W. Ben Chaabene, M. Flah, M.L. Nehdi, Machine learning prediction of mechanical properties of concrete: Critical review, Constr. Build. Mater. 260 (2020) 119889. https://doi.org/10.1016/j.conbuildmat.2020.119889

    work page doi:10.1016/j.conbuildmat.2020.119889 2020

  18. [18]

    Z. Li, J. Yoon, R. Zhang, F. Rajabipour, W.V . Srubar Iii, I. Dabo, A. Radlińska, Machine learning in concrete science: applications, challenges, and best practices, Npj Comput. Mater. 8 (2022) 127. https://doi.org/10.1038/s41524-022-00810-x

    work page doi:10.1038/s41524-022-00810-x 2022

  19. [19]

    Imran, R.A

    M. Imran, R.A. Khushnood, M. Fawad, A hybrid data-driven and metaheuristic optimization approach for the compressive strength prediction of high-performance concrete, Case Stud. Constr. Mater. 18 (2023) e01890. https://doi.org/10.1016/j.cscm.2023.e01890

    work page doi:10.1016/j.cscm.2023.e01890 2023

  20. [20]

    1998 , issn =

    I.-C. Yeh, Modeling of strength of high-performance concrete using artificial neural networks, Cem. Concr. Res. 28 (1998) 1797–1808. https://doi.org/10.1016/S0008-8846(98)00165-3

    work page doi:10.1016/s0008-8846(98)00165-3 1998

  21. [21]

    Young, A

    B.A. Young, A. Hall, L. Pilon, P. Gupta, G. Sant, Can the compressive strength of concrete be estimated from knowledge of the mixture proportions?: New insights from statistical analysis and machine learning methods, Cem. Concr. Res. 115 (2019) 379–388. https://doi.org/10.1016/j.cemconres.2018.09.006

    work page doi:10.1016/j.cemconres.2018.09.006 2019

  22. [22]

    DeRousseau, E

    M.A. DeRousseau, E. Laftchiev, J.R. Kasprzyk, B. Rajagopalan, W.V . Srubar, A comparison of machine learning methods for predicting the compressive strength of field -placed concrete, Constr. Build. Mater. 228 (2019) 116661. https://doi.org/10.1016/j.conbuildmat.2019.08.042

    work page doi:10.1016/j.conbuildmat.2019.08.042 2019

  23. [23]

    J. Yu, Y . Weng, J. Yu, W. Chen, S. Lu, K. Yu, Generative AI for performance-based design of engineered cementitious composite, Compos. Part B Eng. 266 (2023) 110993. https://doi.org/10.1016/j.compositesb.2023.110993

    work page doi:10.1016/j.compositesb.2023.110993 2023

  24. [24]

    Asteris, A.D

    P.G. Asteris, A.D. Skentou, A. Bardhan, P. Samui, K. Pilakoutas, Predicting concrete compressive strength using hybrid ensembling of surrogate machine learning models, Cem. Concr. Res. 145 (2021) 106449. https://doi.org/10.1016/j.cemconres.2021.106449. 58

    work page doi:10.1016/j.cemconres.2021.106449 2021

  25. [25]

    Zhang, D

    T. Zhang, D. Wang, Y . Lu, RheologyNet: A physics -informed neural network solution to evaluate the thixotropic properties of cementitious materials, Cem. Concr. Res. 168 (2023) 107157. https://doi.org/10.1016/j.cemconres.2023.107157

    work page doi:10.1016/j.cemconres.2023.107157 2023

  26. [26]

    T. Oey, S. Jones, J.W. Bullard, G. Sant, Machine learning can predict setting behavior and strength evolution of hydrating cement systems, J. Am. Ceram. Soc. 103 (2020) 480 –490. https://doi.org/10.1111/jace.16706

    work page doi:10.1111/jace.16706 2020

  27. [27]

    Canbek, Q

    O. Canbek, Q. Xu, Y . Mei, N.R. Washburn, K.E. Kurtis, Predicting the rheology of limestone calcined clay cements (LC3): Linking composition and hydration kinetics to yield stress through machine learning, Cem. Concr. Res. 160 (2022) 106925. https://doi.org/10.1016/j.cemconres.2022.106925

    work page doi:10.1016/j.cemconres.2022.106925 2022

  28. [28]

    Marani, T

    A. Marani, T. Oyinkanola, D.K. Panesar, Probabilistic deep learning prediction of natural carbonation of low-carbon concrete incorporating SCMs, Cem. Concr. Compos. 152 (2024) 105635. https://doi.org/10.1016/j.cemconcomp.2024.105635

    work page doi:10.1016/j.cemconcomp.2024.105635 2024

  29. [29]

    R. Cai, T. Han, W. Liao, J. Huang, D. Li, A. Kumar, H. Ma, Prediction of surface chloride concentration of marine concrete using ensemble machine learning, Cem. Concr. Res. 136 (2020) 106164. https://doi.org/10.1016/j.cemconres.2020.106164

    work page doi:10.1016/j.cemconres.2020.106164 2020

  30. [30]

    Allahyari, A

    H. Allahyari, A. Heidarpour, A. Shayan, V .P. Nguyen, A robust time -dependent model of alkali-silica reaction at different temperatures, Cem. Concr. Compos. 106 (2020) 103460. https://doi.org/10.1016/j.cemconcomp.2019.103460

    work page doi:10.1016/j.cemconcomp.2019.103460 2020

  31. [31]

    K.-B. Park, T. Noguchi, J. Plawsky, Modeling of hydration reactions using neural networks to predict the average properties of cement paste, Cem. Concr. Res. 35 (2005) 1676–1684. https://doi.org/10.1016/j.cemconres.2004.08.004

    work page doi:10.1016/j.cemconres.2004.08.004 2005

  32. [32]

    T. Han, R. Bhat, S.A. Ponduru, A. Sarkar, J. Huang, G. Sant, H. Ma, N. Neithalath, A. Kumar, Deep learning to predict the hydration and performance of fly ash -containing cementitious binders, Cem. Concr. Res. 165 (2023) 107093. https://doi.org/10.1016/j.cemconres.2023.107093

    work page doi:10.1016/j.cemconres.2023.107093 2023

  33. [33]

    Sheiati, H

    S. Sheiati, H. Nguyen, P. Kinnunen, N. Ranjbar, Cementitious phase quantification using deep learning, Cem. Concr. Res. 172 (2023) 107231. https://doi.org/10.1016/j.cemconres.2023.107231

    work page doi:10.1016/j.cemconres.2023.107231 2023

  34. [34]

    K. Xu, Q. Jin, J. Li, D.M. Ushizima, V .C. Li, K.E. Kurtis, P.J.M. Monteiro, In- situ microtomography image segmentation for characterizing strain -hardening cementitious composites under tension using machine learning, Cem. Concr. Res. 169 (2023) 107164. https://doi.org/10.1016/j.cemconres.2023.107164

    work page doi:10.1016/j.cemconres.2023.107164 2023

  35. [35]

    Z. Li, K. Yang, Q. He, K. Gong, Large language model -enabled automated data extraction for concrete materials informatics, (2026). https://doi.org/10.48550/arXiv.2604.22938

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2604.22938 2026

  36. [36]

    Gomaa, T

    E. Gomaa, T. Han, M. ElGawady, J. Huang, A. Kumar, Machine learning to predict properties of fresh and hardened alkali-activated concrete, Cem. Concr. Compos. 115 (2021) 103863. https://doi.org/10.1016/j.cemconcomp.2020.103863. 59

    work page doi:10.1016/j.cemconcomp.2020.103863 2021

  37. [37]

    Zhang, P

    M. Zhang, P. Guo, X. Tan, J. Du, W. Meng, Y . Bao, Cradle -to-gate assessment and optimization of sustainable geopolymer concrete, J. Clean. Prod. 538 (2026) 147387. https://doi.org/10.1016/j.jclepro.2025.147387

    work page doi:10.1016/j.jclepro.2025.147387 2026

  38. [38]

    Zhang, A

    L.V . Zhang, A. Marani, M.L. Nehdi, Chemistry -informed machine learning prediction of compressive strength for alkali -activated materials, Constr. Build. Mater. 316 (2022) 126103. https://doi.org/10.1016/j.conbuildmat.2021.126103

    work page doi:10.1016/j.conbuildmat.2021.126103 2022

  39. [39]

    Völker, B

    C. Völker, B. Moreno Torres, T. Rug, R. Firdous, G.A. Jan Zia, S. Lüders, H.L. Scaffino, M. Höpler, F. Böhmer, M. Pfaff, D. Stephan, S. Kruschwitz, Data driven design of alkali - activated concrete using sequential learning, J. Clean. Prod. 418 (2023) 138221. https://doi.org/10.1016/j.jclepro.2023.138221

    work page doi:10.1016/j.jclepro.2023.138221 2023

  40. [40]

    J. Shen, Y . Li, H. Lin, Y . Li, Development of autogenous shrinkage prediction model of alkali- activated slag -fly ash geopolymer based on machine learning, J. Build. Eng. 71 (2023) 106538. https://doi.org/10.1016/j.jobe.2023.106538

    work page doi:10.1016/j.jobe.2023.106538 2023

  41. [41]

    X. Ke, Y . Duan, Coupling machine learning with thermodynamic modelling to develop a composition-property model for alkali-activated materials, Compos. Part B Eng. 216 (2021) 108801. https://doi.org/10.1016/j.compositesb.2021.108801

    work page doi:10.1016/j.compositesb.2021.108801 2021

  42. [42]

    Y . Han, W. Dai, L. Zhou, L. Guo, M. Liu, D. Wang, Y . Ju, Predicting the adsorption capacity of geopolymers for heavy metals in solution based on machine learning, J. Environ. Chem. Eng. 13 (2025) 115978. https://doi.org/10.1016/j.jece.2025.115978

    work page doi:10.1016/j.jece.2025.115978 2025

  43. [43]

    T. Han, E. Gomaa, A. Gheni, J. Huang, M. ElGawady, A. Kumar, Machine learning enabled closed‐form models to predict strength of alkali‐activated systems, J. Am. Ceram. Soc. 105 (2022) 4414–4425. https://doi.org/10.1111/jace.18399

    work page doi:10.1111/jace.18399 2022

  44. [44]

    Philip, N

    S. Philip, N. Marakkath, An interpretable machine learning approach with graphical user interface for strength prediction of sustainable slag -based geopolymer concrete, Structures 81 (2025) 110281. https://doi.org/10.1016/j.istruc.2025.110281

    work page doi:10.1016/j.istruc.2025.110281 2025

  45. [45]

    Ziamiavaghi, V

    B. Ziamiavaghi, V . Toufigh, Fracture toughness evaluation of ground granulated blast furnace slag concrete using experimental study and machine learning techniques, Eng. Fract. Mech. 291 (2023) 109577. https://doi.org/10.1016/j.engfracmech.2023.109577

    work page doi:10.1016/j.engfracmech.2023.109577 2023

  46. [46]

    G. Xue, E. Yilmaz, W. Song, S. Cao, Compressive strength characteristics of cemented tailings backfill with alkali -activated slag, Appl. Sci. 8 (2018) 1537. https://doi.org/10.3390/app8091537

    work page doi:10.3390/app8091537 2018

  47. [47]

    Tang, Y .H

    Y .X. Tang, Y .H. Lee, M. Amran, R. Fediuk, N. Vatin, A.B.H. Kueh, Y .Y . Lee, Artificial neural network-forecasted compression strength of alkaline-activated slag concretes, Sustainability 14 (2022) 5214. https://doi.org/10.3390/su14095214

    work page doi:10.3390/su14095214 2022

  48. [48]

    Sheshadri, S

    A. Sheshadri, S. Marathe, A.P. Rodrigues, M. Nieświec, Predictive modelling of pavement quality fibre-reinforced alkali-activated nano-concrete mixes through artificial intelligence, Stud. Geotech. Mech. 46 (2024) 389–416. https://doi.org/10.2478/sgem-2025-0007

    work page doi:10.2478/sgem-2025-0007 2024

  49. [49]

    Sheshadri, S

    A. Sheshadri, S. Marathe, A.P. Rodrigues, R. Fernandes, Ł. Sadowski, Synergy of industrial wastes in eco-friendly, air-cured alkali activated pavement concrete composites: Properties, 60 embodied carbon and energy assessment and modelling, Road Mater. Pavement Des. (2025) 1–30. https://doi.org/10.1080/14680629.2025.2495705

    work page doi:10.1080/14680629.2025.2495705 2025

  50. [50]

    Philip, M

    S. Philip, M. Nidhi, H.U. Ahmed, A comparative analysis of tree -based machine learning algorithms for predicting the mechanical properties of fibre -reinforced GGBS geopolymer concrete, Multiscale Multidiscip. Model. Exp. Des. 7 (2024) 2555–2583. https://doi.org/10.1007/s41939-023-00355-6

    work page doi:10.1007/s41939-023-00355-6 2024

  51. [51]

    Philip, N

    S. Philip, N. Marakkath, Compressive strength prediction and feature analysis for GGBS - based geopolymer concrete using optimized XGBoost and SHAP: A comparative study of optimization algorithms and experimental validation, J. Build. Eng. 108 (2025) 112879. https://doi.org/10.1016/j.jobe.2025.112879

    work page doi:10.1016/j.jobe.2025.112879 2025

  52. [52]

    Parhi, A

    S.K. Parhi, A. Nanda, S.K. Panigrahi, Multi-objective optimization and prediction of strength along with durability in acid -resistant self -compacting alkali -activated concrete, Constr. Build. Mater. 456 (2024) 139235. https://doi.org/10.1016/j.conbuildmat.2024.139235

    work page doi:10.1016/j.conbuildmat.2024.139235 2024

  53. [53]

    Parhi, S

    S.K. Parhi, S. Panda, S. Dwibedy, S.K. Panigrahi, Metaheuristic optimization of machine learning models for strength prediction of high -performance self -compacting alkali - activated slag concrete, Multiscale Multidiscip. Model. Exp. Des. 7 (2024) 2901–2928. https://doi.org/10.1007/s41939-023-00349-4

    work page doi:10.1007/s41939-023-00349-4 2024

  54. [54]

    Kumar Dash, S

    P. Kumar Dash, S. Kumar Parhi, S. Kumar Patro, R. Panigrahi, Efficient machine learning algorithm with enhanced cat swarm optimization for prediction of compressive strength of GGBS-based geopolymer concrete at elevated temperature, Constr. Build. Mater. 400 (2023) 132814. https://doi.org/10.1016/j.conbuildmat.2023.132814

    work page doi:10.1016/j.conbuildmat.2023.132814 2023

  55. [55]

    Zhang, O

    A. Jafari, V . Toufigh, Developing a comprehensive prediction model for the compressive strength of slag -based alkali-activated concrete, J. Sustain. Cem. -Based Mater. 13 (2024) 256–273. https://doi.org/10.1080/21650373.2023.2266442

    work page doi:10.1080/21650373.2023.2266442 2024

  56. [56]

    Chen, W.-H

    Y.-C. Chen, W.-H. Lee, T.-W. Cheng, W. Chen, Y .-F. Li, The length change ratio of ground granulated blast furnace slag -based geopolymer blended with magnesium oxide cured in various environments, Polymers 14 (2022) 3386. https://doi.org/10.3390/polym14163386

    work page doi:10.3390/polym14163386 2022

  57. [57]

    Al-Sughayer, H

    R. Al-Sughayer, H. Alkhateb, H. Yasarer, Y . Najjar, A. Al-Ostaz, Evaluation and prediction of slag -based geopolymer’s compressive strength using design of experiment (DOE) approach and artificial neural network (ANN) algorithms, Constr. Build. Mater. 440 (2024) 137322. https://doi.org/10.1016/j.conbuildmat.2024.137322

    work page doi:10.1016/j.conbuildmat.2024.137322 2024

  58. [58]

    Al-Fakih, E

    A. Al-Fakih, E. Al-wajih, R.A.A. Saleh, I.B. Muhit, Ensemble machine learning models for predicting the CO 2 footprint of GGBFS -based geopolymer concrete, J. Clean. Prod. 472 (2024) 143463. https://doi.org/10.1016/j.jclepro.2024.143463

    work page doi:10.1016/j.jclepro.2024.143463 2024

  59. [59]

    Abdellatief, A

    M. Abdellatief, A. Baktheer, M. Shahin, A.A. Abadel, A.M. Heniegal, Production and optimization of affordable artificial geopolymer aggregates containing crumb rubber, plastic waste, and granulated cork based on machine learning algorithms, Case Stud. Constr. Mater. 22 (2025) e04725. https://doi.org/10.1016/j.cscm.2025.e04725

    work page doi:10.1016/j.cscm.2025.e04725 2025

  60. [60]

    Stel’makh, A.N

    S.A. Stel’makh, A.N. Beskopylny, E.M. Shcherban’, I. Razveeva, S. Oganesyan, D.M. Shakhalieva, A. Chernil’nik, G. Onore, Compressive strength of geopolymer concrete 61 prediction using machine learning methods, Algorithms 18 (2025) 744. https://doi.org/10.3390/a18120744

    work page doi:10.3390/a18120744 2025

  61. [61]

    Shi, P.V

    C. Shi, P.V . Krivenko, D. Roy, Alkali-activated cements and concretes, Taylor & Francis, London, 2006

    2006

  62. [62]

    Gong, C.E

    K. Gong, C.E. White, Impact of chemical variability of ground granulated blast-furnace slag on the phase formation in alkali-activated slag pastes, Cem. Concr. Res. 89 (2016) 310–319. https://doi.org/10.1016/j.cemconres.2016.09.003

    work page doi:10.1016/j.cemconres.2016.09.003 2016

  63. [63]

    Jiang, Y

    T. Jiang, Y . Jin, H. Ye, Correlating slag chemistry to setting and mechanical behaviors of alkali-activated slag, Constr. Build. Mater. 338 (2022) 127661. https://doi.org/10.1016/j.conbuildmat.2022.127661

    work page doi:10.1016/j.conbuildmat.2022.127661 2022

  64. [64]

    X. Zhu, M. Zhang, K. Yang, L. Yu, C. Yang, Setting behaviours and early-age microstructures of alkali-activated ground granulated blast furnace slag (GGBS) from different regions in China, Cem. Concr. Compos. 114 (2020) 103782. https://doi.org/10.1016/j.cemconcomp.2020.103782

    work page doi:10.1016/j.cemconcomp.2020.103782 2020

  65. [65]

    Tänzer, Y

    R. Tänzer, Y . Jin, D. Stephan, Alkali activated slag binder: Effect of cations from silicate activators, Mater. Struct. 50 (2017) 91. https://doi.org/10.1617/s11527-016-0961-y

    work page doi:10.1617/s11527-016-0961-y 2017

  66. [66]

    Tänzer, A

    R. Tänzer, A. Buchwald, D. Stephan, Effect of slag chemistry on the hydration of alkali - activated blast -furnace slag, Mater. Struct. 48 (2015) 629–641. https://doi.org/10.1617/s11527-014-0461-x

    work page doi:10.1617/s11527-014-0461-x 2015

  67. [67]

    M.B. Haha, B. Lothenbach, G. Le Saout, F. Winnefeld, Influence of slag chemistry on the hydration of alkali-activated blast-furnace slag — Part I: Effect of MgO, Cem. Concr. Res. 41 (2011) 955–963. https://doi.org/10.1016/j.cemconres.2011.05.002

    work page doi:10.1016/j.cemconres.2011.05.002 2011

  68. [68]

    M.B. Haha, G. Le Saout, F. Winnefeld, B. Lothenbach, Influence of activator type on hydration kinetics, hydrate assemblage and microstructural development of alkali activated blast-furnace slags, Cem. Concr. Res. 41 (2011) 301–310. https://doi.org/10.1016/j.cemconres.2010.11.016

    work page doi:10.1016/j.cemconres.2010.11.016 2011

  69. [69]

    M.B. Haha, B. Lothenbach, G. Le Saout, F. Winnefeld, Influence of slag chemistry on the hydration of alkali-activated blast-furnace slag — Part II: Effect of Al2O3, Cem. Concr. Res. 42 (2012) 74–83. https://doi.org/10.1016/j.cemconres.2011.08.005

    work page doi:10.1016/j.cemconres.2011.08.005 2012

  70. [70]

    K. Chen, C. Miao, G. Lyu, K. Wu, J. Chen, J. Xia, Developing an indexing methodology for estimating the reactivity of slag from different sources use in alkali -activated materials, J. Non-Cryst. Solids 647 (2025) 123278. https://doi.org/10.1016/j.jnoncrysol.2024.123278

    work page doi:10.1016/j.jnoncrysol.2024.123278 2025

  71. [71]

    W. Wang, S. Zhang, Y . Zhang, T. Noguchi, I. Maruyama, Understanding the influence of slag fineness and water -to-binder ratio on the alkali -silica reaction in alkali -activated slag mortars, Cem. Concr. Compos. 157 (2025) 105907. https://doi.org/10.1016/j.cemconcomp.2024.105907

    work page doi:10.1016/j.cemconcomp.2024.105907 2025

  72. [72]

    Nasr, A.H

    D. Nasr, A.H. Pakshir, H. Ghayour, The influence of curing conditions and alkaline activator concentration on elevated temperature behavior of alkali activated slag (AAS) mortars, Constr. Build. Mater. 190 (2018) 108–119. https://doi.org/10.1016/j.conbuildmat.2018.09.099. 62

    work page doi:10.1016/j.conbuildmat.2018.09.099 2018

  73. [73]

    Hojati, F

    M. Hojati, F. Rajabipour, A. Radlińska, Drying shrinkage of alkali-activated cements: Effect of humidity and curing temperature, Mater. Struct. 52 (2019) 118. https://doi.org/10.1617/s11527-019-1430-1

    work page doi:10.1617/s11527-019-1430-1 2019

  74. [74]

    Yi, E.-I

    S.-T. Yi, E.-I. Yang, J.-C. Choi, Effect of specimen sizes, specimen shapes, and placement directions on compressive strength of concrete, Nucl. Eng. Des. 236 (2006) 115–127. https://doi.org/10.1016/j.nucengdes.2005.08.004

    work page doi:10.1016/j.nucengdes.2005.08.004 2006

  75. [75]

    Del Viso, J.R

    J.R. Del Viso, J.R. Carmona, G. Ruiz, Shape and size effects on the compressive strength of high-strength concrete, Cem. Concr. Res. 38 (2008) 386–395. https://doi.org/10.1016/j.cemconres.2007.09.020

    work page doi:10.1016/j.cemconres.2007.09.020 2008

  76. [76]

    Z. Pan, X. Li, Y . Wang, S. Hossain, K. Gong, Physics -informed glass-structure descriptors for assessing the intrinsic reactivity of mixed amorphous -crystalline precursors in alkali - activated materials, (2026). https://doi.org/10.48550/arXiv.2602.13101

    work page doi:10.48550/arxiv.2602.13101 2026

  77. [77]

    Gong, E.A

    K. Gong, E.A. Olivetti, Development of structural descriptors to predict dissolution rate of volcanic glasses: Molecular dynamic simulations, J. Am. Ceram. Soc. 105 (2022) 2575–

    2022

  78. [78]

    https://doi.org/10.1111/jace.18265

    work page doi:10.1111/jace.18265

  79. [79]

    Gong, C.E

    K. Gong, C.E. White, Development of physics-based compositional parameters for predicting the reactivity of amorphous aluminosilicates in alkaline environments, Cem. Concr. Res. 174 (2023) 107296. https://doi.org/10.1016/j.cemconres.2023.107296

    work page doi:10.1016/j.cemconres.2023.107296 2023

  80. [80]

    Gong, C.E

    K. Gong, C.E. White, Predicting CaO -(MgO)-Al2O3-SiO2 glass reactivity in alkaline environments from force field molecular dynamics simulations, Cem. Concr. Res. 150 (2021) 106588. https://doi.org/10.1016/j.cemconres.2021.106588

    work page doi:10.1016/j.cemconres.2021.106588 2021

Showing first 80 references.