Work

Advancing microstructural insights in 3D-Printed cementitious materials via X-ray micro-computed tomography

Year: 2025

Type: article

Abstract: This work aims to provide a comprehensive review of utilising X-ray micro-computed tomography (μCT) to analyse 3D-printed cementitious composites and porous concrete (foamed concrete and pervious concrete), followed by recent machine learning applications in μCT analysis of cementitious materials. The X-ray μCT analysis on 3D-printed concrete primarily reveals the porosity distribution along the layer height and pore morphology, which correlate with the mechanical and durability properties of 3D-printed concrete. On the other hand, a proper phase segmentation method is considered critical to accurately differentiate the air void phase from the solid phase in porous concrete. Machine learning has been predominantly applied to develop reliable and automated models that can efficiently segment different phases in the X-ray μCT sliced images of concrete, including air voids, aggregates, cement paste and lightweight polymer fibres. Meanwhile, an in-depth understanding of the concrete failure mechanism could be achieved by segmenting crack phases from air void phases with the assistance of machine learning. X-ray μCT should be regarded as a powerful tool for visualising and characterising the pore structures of both 3D-printed and porous concrete. Furthermore, the integration of machine learning shows the promising future of moving the X-ray μCT analysis on cementitious materials towards higher efficiency and reliability. •X-ray μCT analysis of cementitious materials aids in visualising porosity and structural integrity. •Machine learning improves phase segmentation in X-ray μCT images of 3D-printed and porous concrete. •Challenges in segmenting lightweight polymer fibres and recycled waste materials require advanced models. •Multi-scale scans using μCT and nano-CT can reveal finer pore structures in cementitious materials. •Further research on scanning parameters and reconstruction settings can optimize model size and quality.

Source: Journal of Building Engineering

DOI: 10.1016/j.jobe.2025.112675

Authors: Junli Liu, Hai Hoang Khieu, Mien Van Tran, Phuong Tran

Institutions: RMIT University, Vietnam National University Ho Chi Minh City

Language: English

Topic: Innovations in Concrete and Construction Materials

Subfield: Building and Construction

Field: Engineering

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: hybrid

Estimated APC paid: $3,940