Iterative Optimization of Concrete Mix Composition with a Target Hardening Rate Using Machine Learning
DOI:
https://doi.org/10.36910/6775-2410-6208-2025-14(24)-27Keywords:
early hardening, concrete technology, machine learning, Random Forest, mix optimization, Monte Carlo methodAbstract
In the modern construction industry, one of the key factors enhancing the economic efficiency of monolithic construction is the formwork turnover rate, which is determined by the kinetics of concrete strength gain. Traditional approaches to concrete mix design primarily focus on ensuring design strength after 28 days and often fail to rapidly determine optimal recipes for accelerated hardening without lengthy physical experiments. This paper proposes and implements an approach to solving the inverse technological problem: the automated design of concrete mix composition with a specified hardening rate. The methodological basis of the study combines predictive modeling and stochastic optimization. A Random Forest ensemble regression model, trained on an open dataset, was used as the analytical component. The optimization problem is implemented through a two-loop algorithm: the outer loop determines the time intervals for achieving the design strength, while the inner loop performs Monte Carlo generation of pseudo-random mix designs (100,000 iterations) followed by their ranking. Within the computational experiment for concrete class C32/40 (40 MPa), a technologically feasible limit for early strength achievement was established. The modeling identified a cluster of 236 mix designs capable of ensuring the design strength on the third day of hardening. Analysis of the component composition of the obtained mix designs demonstrated consistency with established technological principles for High Early Strength Concrete. Specifically, to achieve target performance indicators within an ultra-short timeframe, the model identified a reduction of the water-cement ratio to 0.24–0.26 and a cement content exceeding 450 kg/m³ as optimal parameters, which correlates with Abrams' law and the principles of High Performance Concrete production. It was established that reducing the curing time to 3 days increases the cost of the mix by approximately 82% relative to the standard 28-day cycle. The proposed approach can serve as a support tool for engineering and technological decisions, as it provides an automated search for a rational concrete mix composition considering cost constraints and required construction schedules.
Downloads
References
1. Qi C., Huang B., Wu M., Wang K., Yang S., Li G. Concrete Strength Prediction Using Different Machine Learning Processes: Effect of Slag, Fly Ash and Superplasticizer // Materials. – 2022. – Vol. 15, № 15. – P. 5369. – DOI: 10.3390/ma15155369.
2. Sun C., Wang K., Liu Q., Wang P., Pan F. Machine-Learning-Based Comprehensive Properties Prediction and Mixture Design Optimization of Ultra-High-Performance Concrete // Sustainability. – 2023. – Vol. 15, № 21. – P. 15338. – DOI: 10.3390/su152115338.
3. Beskopylny A.N., Stel’makh S.A., Shcherban’ E.M., Mailyan L.R., Meskhi B., Razveeva I., Chernil’nik A., Beskopylny N. Concrete Strength Prediction Using Machine Learning Methods: CatBoost, k-Nearest Neighbors, Support Vector Regression // Applied Sciences. – 2022. – Vol. 12, № 21. – P. 10864. – DOI: 10.3390/app122110864.
4. Forsdyke J.C., Zviazhynski B., Lees J.M., Conduit G.J. Probabilistic Selection and Design of Concrete Using Machine Learning // Data-Centric Engineering. – 2023. – Vol. 4. – e9. – DOI: 10.1017/dce.2023.5
5. Nguyen H., Vu T., Vo T.P., Thai H.T. Efficient machine learning models for prediction of concrete strengths // Construction and Building Materials. – 2021. – Vol. 266, Part B. – Article 120950. – DOI: 10.1016/j.conbuildmat.2020.120950
6. Pfeiffer O.P., Gong K., Severson K.A., Chen J., Gregory J.R., Ghosh S., Goodwin R.T., Olivetti E.A. Bayesian design of concrete with amortized Gaussian processes and multi-objective optimization // Cement and Concrete Research. – 2024. – Vol. 177. – 107406. – DOI: 10.1016/j.cemconres.2023.107406
7. Li P., Zhang Y., Gu J., et al. Prediction of compressive strength of concrete based on improved artificial bee colony-multilayer perceptron algorithm // Sci Rep. – 2024. – Vol. 14. – 6414. – DOI: 10.1038/s41598-024-57131-w
8. Ament S., Witte A., Garg N., Kusuma J. Sustainable concrete via Bayesian optimization // arXiv. – 2023. – URL: https://arxiv.org/abs/2310.18288
9. Vargas J.F., Oviedo A.I., Ortega N.A., Orozco E., Gómez A., Londoño J.M. Machine-Learning-Based Predictive Models for Compressive Strength, Flexural Strength, and Slump of Concrete // Applied Sciences. – 2024. – Vol. 14, № 11. – 4426. – DOI: 10.3390/app14114426
10. Yeh I.-C. Concrete Compressive Strength [Dataset] // UCI Machine Learning Repository. – 1998. – DOI: 10.24432/C5PK67
11. Ananyachandran, Polipati & Vasugi, V.. Review on high early strength concrete. International Journal of Civil Engineering and Technology. 8. 697-705c – 2017
