AI Technique Accurately Predicts Compressive Strength in Waste Marble Concrete

A recent study in Scientific Reports proposes using artificial intelligence (AI) methods instead of traditional experimental approaches to predict the compressive strength (CS) of waste marble (WM) concrete (WMC). In this study, researchers developed three models using artificial neural networks (ANN) combined with hybrid optimization techniques—ant colony optimization (ACO) and biogeography-based optimization (BBO).

Modeling Compressive Strength of Sustainable Concrete Using AI
Study: An AI-driven approach for modeling the compressive strength of sustainable concrete incorporating waste marble as an industrial by-product. Image Credit: ungvar/Shutterstock.com

Background

The concrete industry is increasingly adopting sustainable practices, such as substituting cement with agro-industrial by-products and wastes. WM, a by-product of marble cutting and polishing, has fine particles and high lime content that may improve concrete’s mechanical properties and durability.

However, large-scale use of WMC requires an understanding of how WM impacts CS. Traditional laboratory methods are often costly, time-intensive, and less adaptable to changing material proportions or testing conditions.

In contrast, AI-driven methods can address these limitations by providing reliable predictions without the need for repetitive laboratory tests.

Methods

As part of this study, an extensive dataset including 1135 different mixtures was collected from 53 literature sources to predict the CS of WMC. In the constructed models, the amounts of cement (C), water (W), WM, coarse aggregate (CA), superplasticizer (SP),  fine aggregate (FA), and age of specimens (A) were considered as inputs (independent variables), and CS was the output (dependent variable).

Three AI models—ANN, ANN-ACO, and ANN-BBO—were developed using this dataset, which was divided into three phases: training (70 % or 795 samples), validation (15 % or 170 samples), and testing (15 % or 170 samples).

The Pearson correlation coefficient method was applied to examine relationships among the characteristic variables, while a sensitivity analysis based on linear correlations between inputs and CS was also conducted. The Levenberg-Marquardt algorithm demonstrated superior performance compared to other learning algorithms and was selected for the study.

Model accuracy was evaluated using various parameters, including mean absolute error, root mean squared error, determination coefficient, Nash-Sutcliffe efficiency, performance index, and A-10 index. Additionally, K-fold cross-validation was implemented by partitioning the training and validation data into ten folds, with 90 % of data in each fold used for training and 10 % for validation. Errors were calculated across folds to obtain an average. Finally, the predictive models were tested on separate data to assess overall performance and conduct re-validation.

Results and Discussion

The ANN model achieved R2 values of 0.9540, 0.9353, and 0.9392 in the training, validation, and testing phases, respectively. In comparison, the ANN-ACO model achieved R2 values of 0.9721, 0.9710, and 0.9655, while the ANN-BBO model reached 0.9955, 0.9882, and 0.9867 across these phases. These results indicate that the ANN-BBO model showed the strongest correlation between observed and predicted values.

Examining error ranges, 98 %, 97 %, and 94 % of predictions in the ANN-BBO model’s training, validation, and testing phases, respectively, fell within an error margin of -10 % to 10 %. In comparison, the ANN-ACO model achieved 85 %, 83 %, and 82 % within this range, while the ANN model recorded 79 %, 71 %, and 80%, further underscoring the ANN-BBO model’s superior accuracy in predicting CS of WMC within a narrower error range.

The influence of different input variables on the CS of WMC varied. Specimen age emerged as the most significant factor, accounting for 24 % of the influence on CS, followed by cement (20 %), water (18 %), and WM (12 %). WM’s contribution underscores its potential as a sustainable alternative to cement in concrete mixtures.

The remaining inputs, fine aggregate, coarse aggregate, and superplasticizer, contributed 10 %, 8 %, and 8 % to CS, respectively. SHapley Additive exPlanations (SHAP) analysis further revealed that higher values of age and cement positively impacted CS, whereas increases in water and WM showed a negative trend. Additionally, the optimum level of WM substitution was identified at up to 15 % of the total cement content, balancing performance with sustainability.

Conclusion and Future Prospects

This study presents an accurate and reliable AI-driven approach for predicting the CS of WMC. The hybrid optimization techniques, ACO and BBO, demonstrated superior performance over the single ANN model, with the ANN-BBO model emerging as the most effective. Compared to models in the literature, the ANN-BBO model achieved improved accuracy, likely due to the use of an extensive dataset.

For future research, incorporating additional variables and assessing the durability and strength of WMC under diverse environmental conditions are recommended to further refine the model’s applicability and robustness.

Journal Reference

Kazemi, R. & Mirjalili, S. (2024). An AI-driven approach for modeling the compressive strength of sustainable concrete incorporating waste marble as an industrial by-product. Scientific Reports14(1). DOI: 10.1038/s41598-024-77908-3, https://www.nature.com/articles/s41598-024-77908-3

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. (2024, November 12). AI Technique Accurately Predicts Compressive Strength in Waste Marble Concrete. AZoBuild. Retrieved on November 23, 2024 from https://www.azobuild.com/news.aspx?newsID=23644.

  • MLA

    Dhull, Nidhi. "AI Technique Accurately Predicts Compressive Strength in Waste Marble Concrete". AZoBuild. 23 November 2024. <https://www.azobuild.com/news.aspx?newsID=23644>.

  • Chicago

    Dhull, Nidhi. "AI Technique Accurately Predicts Compressive Strength in Waste Marble Concrete". AZoBuild. https://www.azobuild.com/news.aspx?newsID=23644. (accessed November 23, 2024).

  • Harvard

    Dhull, Nidhi. 2024. AI Technique Accurately Predicts Compressive Strength in Waste Marble Concrete. AZoBuild, viewed 23 November 2024, https://www.azobuild.com/news.aspx?newsID=23644.

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.