Predicting carbonation depth in fiber-reinforced ultra-high performance concrete (FR-UHPC) using state-of-the-art...

Area of Science:

• Civil Engineering
• Materials Science
• Computational Science

Background:

• Carbonation is a major cause of concrete structure degradation.
• Existing models struggle with fiber-reinforced ultra-high-performance concrete (FR-UHPC) due to its complexity.
• Predictive models for FR-UHPC carbonation are limited.

Purpose of the Study:

• To develop a data-driven framework for predicting FR-UHPC carbonation depth.
• To apply advanced machine learning techniques for enhanced durability assessment.
• To improve the interpretability of carbonation prediction models.

Main Methods:

• Utilized a dataset of 800 experimental points.
• Employed machine learning techniques: Neural Operators for Modeling Physical Systems (NOMPS), Artificial Intelligence-based Pipeline Search for Regression (AIPSR), Quantum Machine Learning (QML), and Explainable AI using Quantum Shapley Values (EAIQSV).
• Validated models using statistical analysis and 5-fold cross-validation.

Main Results:

• Identified curing time, temperature, and silica fume content as key factors influencing carbonation depth.
• AIPSR demonstrated superior prediction accuracy (R² = 0.83) and consistency compared to other models.
• The framework offers robust and repeatable carbonation prediction for FR-UHPC.

Conclusions:

• The developed framework provides an accurate and interpretable method for predicting FR-UHPC carbonation.
• Incorporation of quantum-inspired machine learning enhances predictive capabilities.
• This approach aids in assessing and improving the durability of concrete structures.