Machine Learning Enhances Predictive Modeling for Eco-Friendly Concrete

A recent study published in Scientific Reports highlights the potential of PyCaret, an automated machine learning (ML) platform, in developing and evaluating predictive models for the compressive strength of compressed earth blocks (CEB). By leveraging ML, researchers aim to better understand the complex nonlinear relationships among various CEB components, such as cement, clay, sand, silt, and fibers.

Study: Assessment of compressive strength of eco-concrete reinforced using machine learning tools. Image Credit: Tonis Valing/Shutterstock.com

Background

Machine learning, a subset of artificial intelligence (AI), has been widely applied in predicting the mechanical properties of different types of concrete, including high-performance, self-healing, and recycled coarse aggregate concrete. This is largely due to ML models’ ability to process complex datasets and generate highly accurate predictions with minimal deviation compared to traditional laboratory experiments.

Among various ML techniques, ensemble modeling has demonstrated superior accuracy and robustness compared to individual models.

Other approaches, such as support vector machines (SVM) and artificial neural networks (ANN), have also proven effective in predicting concrete compressive strength using parameters like ultrasonic pulse velocity. These techniques additionally facilitate the evaluation of concrete durability characteristics, such as resistance to carbonation and environmental degradation.

Building on AI advancements, this study explores the application of ML tools—including neural networks and ensemble models—to predict the mechanical properties of earthen concrete, focusing on speed, accuracy, and cost-effectiveness.

Methods

Researchers employed PyCaret, an open-source, low-code ML framework, to predict the compressive strength of earth concrete. The process involved several key stages: data collection, preprocessing, feature selection, model training, validation, and justification for the input variables.

PyCaret automates feature selection by assessing feature importance using methods like mutual information and correlation analysis. The Pearson correlation coefficient was calculated between input features to identify highly correlated variables, such as cement content and aggregate type.

Depending on the correlation strength, these features were either merged or excluded to prevent redundancy. Features with low PyCaret scores, indicating minimal contribution to compressive strength, were discarded to improve model generalization and avoid overfitting.

Multiple ML models were trained using PyCaret, including:

  • Linear models (linear and lasso regression)
  • Tree-based models (random forests, decision trees, and gradient boosting machines)
  • SVM (for capturing nonlinear relationships in data)
  • ANN (for complex, high-dimensional data)

To validate the models, the researchers employed K-fold cross-validation and evaluated performance using metrics such as mean absolute error (MAE), root mean squared error (RMSE), and R-squared (R2). The best-performing model was retrained on the full dataset before making final predictions.

Results and Discussion

The ML analysis revealed that cement plays a crucial role in enhancing the compressive strength of CEB. For optimal results, cement content should be maintained between 10–15 % of the total CEB mass. Additionally, fibers significantly contribute to material cohesion and exhibit a strong positive correlation with cement content, highlighting their influence on compressive strength.

The study found that artificial fibers outperform natural fibers due to their superior durability and lifespan. However, natural fibers remain widely used due to their availability and lower cost. The optimal fiber content for reinforcing CEB is between 1–2 %; exceeding this threshold can negatively impact performance.

Other key components in CEB were found to have optimal ranges as well:

  • Clay: 5–25 %
  • Sand: 50–75 %
  • Silt: 5–15 %

Among the ML models tested using PyCaret, the Extra Trees Regressor emerged as the most accurate and precise, with:

  • RMSE: 0.4909 (indicating low variability in prediction errors)
  • R2: 0.9444 (signifying highly accurate predictions)
  • MAE: 0.1899 (suggesting minimal average prediction error)

The results demonstrate that the Extra Trees Regressor effectively handles highly nonlinear and multivariate datasets, making it a reliable tool for predictive modeling in material science.

Conclusion and Future Prospects

This study underscores the value of machine learning, particularly the Extra Trees Regressor, in accurately predicting the compressive strength of CEB. By utilizing PyCaret, researchers streamlined the ML model development and evaluation process, highlighting the importance of high-quality data and thorough preprocessing for reliable predictions.

ML models offer a practical way to optimize material composition by simulating different formulations, reducing the need for extensive trial-and-error experiments. This approach not only saves time and resources but also improves efficiency in material development. The high accuracy of the Extra Trees Regressor suggests it could serve as a viable alternative to conventional laboratory testing.

However, one limitation of this study is the specific dataset used, which may not fully capture variations in soil composition, environmental conditions, and curing practices across different regions. Expanding the dataset to incorporate diverse real-world conditions will be crucial in enhancing the generalizability of the ML models.

Journal Reference

Bentegri, H. et al. (2025). Assessment of compressive strength of eco-concrete reinforced using machine learning tools. Scientific Reports15(1). DOI: 10.1038/s41598-025-89530-y, https://www.nature.com/articles/s41598-025-89530-y

Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.

Nidhi Dhull

Written by

Nidhi Dhull

Nidhi Dhull is a freelance scientific writer, editor, and reviewer with a PhD in Physics. Nidhi has an extensive research experience in material sciences. Her research has been mainly focused on biosensing applications of thin films. During her Ph.D., she developed a noninvasive immunosensor for cortisol hormone and a paper-based biosensor for E. coli bacteria. Her works have been published in reputed journals of publishers like Elsevier and Taylor & Francis. She has also made a significant contribution to some pending patents.  

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Dhull, Nidhi. (2025, February 21). Machine Learning Enhances Predictive Modeling for Eco-Friendly Concrete. AZoBuild. Retrieved on February 21, 2025 from https://www.azobuild.com/news.aspx?newsID=23720.

  • MLA

    Dhull, Nidhi. "Machine Learning Enhances Predictive Modeling for Eco-Friendly Concrete". AZoBuild. 21 February 2025. <https://www.azobuild.com/news.aspx?newsID=23720>.

  • Chicago

    Dhull, Nidhi. "Machine Learning Enhances Predictive Modeling for Eco-Friendly Concrete". AZoBuild. https://www.azobuild.com/news.aspx?newsID=23720. (accessed February 21, 2025).

  • Harvard

    Dhull, Nidhi. 2025. Machine Learning Enhances Predictive Modeling for Eco-Friendly Concrete. AZoBuild, viewed 21 February 2025, https://www.azobuild.com/news.aspx?newsID=23720.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type
Submit

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.