Elucidating Rheological Properties of Cementitious Materials Containing Fly Ash and Nanosilica by Machine Learning

Area of Science:

• Materials Science
• Civil Engineering
• Computational Science

Background:

• Rheological properties of concrete are crucial for enhancing mechanical performance and material sustainability.
• Predicting these properties is challenging due to numerous influencing factors, necessitating advanced computational approaches.
• Machine learning offers a powerful solution for predicting construction material properties in the context of big data.

Purpose of the Study:

• To develop predictive models for the rheological behavior of cementitious materials incorporating fly ash and nanosilica.
• To evaluate the efficacy of various machine learning algorithms in forecasting shear stress and apparent viscosity curves.
• To identify the most accurate model for reliable prediction of cementitious material viscosity.

Main Methods:

• Four machine learning models were utilized: Random Forest, XGBoost, Artificial Neural Network (ANN), and Recurrent Neural Network (RNN) with Stacked Long Short-Term Memory (LSTM) layers.
• Models were trained and validated to predict shear rate versus shear stress and shear rate versus apparent viscosity relationships.
• Hyperparameter tuning was performed to optimize model performance.

Main Results:

• The RNN (Stacked LSTM) model demonstrated superior predictive accuracy for both shear rate versus shear stress (R² = 0.9582) and shear rate versus apparent viscosity (R² = 0.9257) curves.
• The Stacked LSTM model exhibited excellent statistical parameters and fitting effects.
• Comparative analysis indicated the RNN (Stacked LSTM) possesses better generalization capabilities compared to other models.

Conclusions:

• The RNN (Stacked LSTM) model is highly effective for predicting the rheological properties of cementitious materials containing fly ash and nanosilica.
• This advanced machine learning approach offers a reliable tool for future predictions in cementitious material viscosity.
• The findings contribute to the development of more sustainable and high-performance concrete through accurate rheological characterization.