脱墨紙スラッジとメタカオリンを用いた3Dプリンティング建設向け低炭素廃棄物価値化電気機械セメント複合材料：ハイブリッドML–FEMデジタルツイン

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:廃棄物由来バイオ炭とML–FEMデジタルツインを組み合わせた自己感知3Dプリント材料の実証データと再現可能なコードを提供する。

🏢実務担当者:建設・素材企業にとって、セメント置換と廃棄物活用による低炭素材料の物性・耐久性トレードオフを把握する参考になる。

This study develops a low-carbon electromechanical self-sensing composite incorporating deinking paper sludge (DPS)-derived biochar for extrusion-based three-dimensional concrete printing (3DCP). It further demonstrates a proof-of-concept physics-based hybrid machine-learning–finite-element (ML–FEM) digital twin for engineering-scale structural assessment. Ordinary Portland cement was partially replaced by 40 wt.% metakaolin, while 18 wt.% DPS-derived biochar (electrical percolation threshold) was used as the conductive filler. Fresh-state (rheology and printability), hardened-state physical performance under compressive and flexural loading (⊥ and ∥ to the printed layers), microstructural/chemical characterisation (SEM, XCT, MIP, and TGA), and electromechanical behaviour under four internal relative humidity values (44.6–96.7%) using elastic cyclic, progressive cyclic, and monotonic loading were experimentally investigated. The Herschel–Bulkley yield stress increased from 82.3 to 158.6 Pa (+93%) while maintaining stable shear-thinning behaviour suitable for extrusion. Incorporating 18 wt.% DPS-derived biochar reduced the compressive strength from 39.0 to 23.9 MPa. XCT revealed the highest total porosity (5.08%) and connected porosity (2.14%), confirming the formation of an interconnected conductive network. Internal humidity strongly governed the electromechanical response. Under elastic cyclic loading, the gauge factor increased from 34 to 114 as internal relative humidity decreased from 96.7% to 44.6%. Under progressive cyclic loading, the terminal fractional change in resistance (FCR) increased from approximately 5% to 32%, enabling clear discrimination between reversible deformation and progressive damage. The environment-aware ML model accurately predicted tensile strain and flexural stress from RH and FCR with a median R 2 ≈ 0.96. The hybrid ML–FEM digital twin successfully reconstructed full-field stress, strain, and displacement distributions while quantifying the remaining load-bearing capacity. The proposed framework provides a scalable pathway towards environmentally adaptive physics-based digital twins and next-generation structural health monitoring of 3DCP infrastructure. The complete experimental database and Python source codes are provided in Appendix A to ensure full reproducibility.

• openalex https://doi.org/10.1016/j.jpcs.2026.114154first seen 2026-09-29 05:00:09