Forecasting the rheological properties of alkali-activated concrete utilizing gene expression programming

Area of Science:

• Materials Science
• Civil Engineering
• Computational Intelligence

Background:

• Accurate prediction of fresh-state properties is crucial for alkali-activated concrete (AAC) mix uniformity and performance.
• Traditional methods for property prediction are often time-consuming and resource-intensive.
• Gene expression programming (GEP) offers a flexible and interpretable alternative for modeling complex material relationships.

Purpose of the Study:

• To develop a robust predictive model for fresh-state properties of AAC using GEP.
• To establish explicit mathematical equations for engineering design and practical application.
• To create a user-friendly graphical interface (GUI) for simplified mix design and rapid predictions.

Main Methods:

• Gene expression programming (GEP) was employed to model the relationships between AAC mixture components and fresh-state properties.
• The model predicted slump (SL), dynamic yield stress (DYS), static yield stress (SYS), and plastic viscosity (PV).
• Performance was evaluated using coefficient of determination (R²), root mean square error (RMSE), and explicit equation derivation.

Main Results:

• The GEP model achieved high accuracy for all predicted properties: SL (R²=0.979, RMSE=1.183), SYS (R²=0.960, RMSE=3.293), DYS (R²=0.962, RMSE=0.161), and PV (R²=0.962, RMSE=5.266).
• Explicit GEP-derived equations were generated for practical engineering use.
• A GUI was developed to facilitate instant predictions and simplify the mix design process.

Conclusions:

• GEP provides a powerful and accurate tool for predicting the fresh-state properties of AAC.
• The developed GEP model and GUI significantly reduce the need for extensive physical experimentation.
• This approach promotes wider adoption of AAC by enhancing mix design efficiency and control over material properties.