Unraveling pore structure alternations in 3D-printed geopolymer concrete and corresponding impacts on macro-properties

November 01, 2022

Yuning Chen (1) (2), Yamei Zhang (1) (2), Yudong Xie (1), Zedi Zhang (1) (2), Nemkumar Banthia (3)
Additive Manufacturing. Volume 59 (November 2022). DOI: https://doi.org/10.1016/j.addma.2022.103137


3D-printed concrete, pore structure, pore elongation, mechanical anisotropy, interface


Extrusion-based 3D-printed concrete (3DPC) structures are reported to hold mechanical anisotropy behaviors and weak transport properties compared with cast concrete. Fundamental insights into the pore structure discrepancy between printed and cast concrete are essential to the performance prediction and improvement strategy for 3DPC. This study analyzes the pore structure alternations in 3D-printed geopolymer concrete (3DPGC) with cast ones as the reference. Several pore characteristics, i.e., pore volume, distribution, specific surface area (SSA), shape and connectivity are investigated via X-ray CT and MIP. The results demonstrate that a larger porosity, coarser pore size distributionand higher pore SSA exist in 3DPGC compared with CGC. The coarser pore size distribution respectively lies in large voids (>0.2 mm) and small pores (<400 nm) for printed concrete. The pulling stress applied by nozzle movements during the extrusion process contributes to the pore elongation of printed concrete. The mechanical anisotropy of printed concrete without fibers originates from two factors: (i) Oriented pore elongation induces the discrepancy in stress concentration and deformation, and (ii) The weak interlayer presence may cause sliding between layers during loading. However, the pore elongation effect decays with the pore size reduction, limiting its impact on mechanical-anisotropic behaviors. Targeted strategies are then proposed for the matrix strengthening and mechanical anisotropy mitigation in printed concrete.

How Our Software Was Used

Dragonfly was employed to visualize and analyze the grayscale and 3D-reconstructed images. These images were binarized using Dragonfly’s integrated K-means segmentation algorithm. Various parameters, including frequency, volume, and specific surface area, were obtained from the binarized images.

Author Affiliation

(1) School of Materials Science and Engineering, Jiangsu Key Laboratory of Construction Materials, Southeast University, Nanjing 211189, China
(2) Nanjing Institute for Intelligent Additive Manufacturing Co., Ltd, Nanjing 210000, China
(3) Department of Civil Engineering, University of British Columbia, Vancouver V6T 1Z4, Canada