# Assessing the Porosity-Binder Ratio and Machine Learning Models for Predicting the Strength and Durability of Soil-Cement-Glass Powder Geomaterial

## TL;DR

This study uses porosity-binder ratios and machine learning to predict the strength and durability of soil-cement-glass powder mixtures.

## Contribution

The study introduces a validated porosity-binder index and identifies optimal machine learning models for predicting geocomposite properties.

## Key findings

- The porosity–cement index (η/Civ) strongly predicts compressive and tensile strength with R2 > 0.98.
- Adding 30% glass powder reduced mass loss to below 0.5% after 90 days of curing.
- Matern 5/2 Gaussian Process Regression and trilayered neural networks achieved R2 > 0.987 for strength and durability predictions.

## Abstract

This study evaluates the mechanical behavior and durability of a silty soil stabilized with Portland cement and recycled ground glass powder (GGP). The porosity–cement index (η/Civ) was applied to predict unconfined compressive strength (qu), splitting tensile strength (qt), and accumulated mass loss (ALM) under wetting–drying cycles. Mixtures were prepared with cement contents of 3%, 6%, and 9%, GGP contents of 5%, 15%, and 30%, and dry unit weights of 13.5, 14.5, and 15.5 kN/m3, and were cured for 7, 28, and 90 days. The experimental program consisted of a large dataset, comprising 486 mechanical tests (unconfined compressive and splitting tensile strength) and 81 durability tests, providing a robust basis for both empirical modeling and machine learning analysis. The results confirmed a strong power-law relationship between η/Civ and both qu and qt, achieving high coefficients of determination (R2 > 0.98). The strength coefficient (A) increased consistently with curing time and GGP addition, indicating enhanced pozzolanic reactivity and matrix densification. After 90 days, qu increased by over 250% and qt by nearly 700%. Durability tests revealed exponential reductions in ALM with higher density and binder content, achieving values below 0.5% for the densest mixtures, which contained 30% GGP. These findings validate the η/Civ index as an effective predictor of strength and durability in soil–cement–GGP geomaterials, establishing a solid basis for future integration with machine learning models. The implementation of twenty-eight machine learning presets for predicting qu, qt, and ALM demonstrated that the Matern 5/2 Gaussian Process Regression and the trilayered neural network are the most suitable algorithms, achieving R2 values higher than 0.987 in both the validation and testing stages.