Quantification of Total Porosity from CT Images by Segmenting Unhydrated Cement: A Model-Informed Framework Integrating POWERS’ Volume Model

Abstract

Quantification of total porosity, including the nano-scale fraction, is critical for predicting the performance of cement-based materials but remains a significant challenge. While X-ray computed tomography (CT) is a powerful non-destructive tool, a fundamental trade-off between resolution and representative sample volume prevents the direct segmentation of nano-scale pores in macroscopically relevant specimens. Herein, we propose and validate a novel model-informed framework that overcomes this limitation by integrating the classical Powers’ hydration model with micro-CT analysis. The method circumvents the need for nano-scale resolution by deriving the total porosity from the volume fraction of the easily segmentable, micron-scale unhydrated cement phase. The framework’s validity was demonstrated by showing a strong correlation between the CT-derived total porosity and established porosity–strength relationships. Quantitative analysis indicated that the total porosity of the cement pastes ranged from 36.5% to 60.3% as the w/c ratio increased from 0.4 to 0.7. Laboratory strength data show good correlation (R2 > 0.98) with four porosity–strength prediction models, demonstrating the feasibility of applying the Powers’ volume model in CT-based analyses of cement pastes. This work transforms micro-CT from a qualitative imaging tool into a comprehensive technique for the quantitative microstructural characterization of cementitious materials.