A Comprehensive Review of Numerical and Machine Learning Approaches for Predicting Concrete Properties: From Fresh to...

Area of Science:

• Materials Science
• Civil Engineering
• Computational Modeling

Background:

• Cementitious composites require predictive models due to increasing material diversity and innovation.
• Existing numerical, code-based, and machine learning (ML) models predict concrete properties but are sensitive to input variables.
• Accurate prediction of concrete properties necessitates models that integrate mix design, composition, intrinsic properties, and external conditions.

Purpose of the Study:

• To conduct a comprehensive review of numerical, code-based, and ML modeling techniques for predicting concrete properties.
• To evaluate the impact of data gathering, preprocessing, and handling on model performance.
• To compare the predictive accuracy of ML models against traditional numerical approaches.

Main Methods:

• Review of existing literature on numerical, code-based, and ML modeling for concrete properties.
• Analysis of factors influencing model accuracy, including data variability and dataset size.
• Comparison of performance metrics for various ML algorithms and traditional numerical models.

Main Results:

• Machine learning models demonstrate superior predictive performance over traditional numerical methods for concrete properties.
• Data preprocessing significantly impacts the accuracy of both numerical and ML models.
• ML-based approaches, including ensemble and non-ensemble models, show strong predictive capabilities.
• Modified numerical models with additional parameters show improved accuracy.
• Optimization algorithms and interpretability tools enhance model reliability and transparency.

Conclusions:

• ML models are highly effective for predicting a wide range of fresh and long-term concrete properties.
• Robust data handling and preprocessing are critical for reliable concrete property prediction.
• Integrating optimization and interpretability tools is essential for advancing concrete modeling.