Mechanistic insights into the rheology of recycled concrete powder based geopolymer using SHAP-integrated machine learning

Dongsheng Zhang

doi:10.1016/j.conbuildmat.2026.147669

Mechanistic insights into the rheology of recycled concrete powder based geopolymer using SHAP-integrated machine learning

2026-08-07

0

PRE

AI

D

Dongsheng Zhang

T

Tao Zhu

M

Mingjie Mao

P

Ping Chen

李

李嘉宾 (Jiabin Li) *

DOI：10.1016/j.conbuildmat.2026.147669

Original

Original request for help

Save

Abstract

En 中文

• A multi-scale closed-loop framework links mix design, microstructure, and time-dependent rheology of RCP based geopolymers. • Yield stress and plastic viscosity show distinct evolution mechanisms under varying L/S, Ms, and W/B conditions. • XGBoost and Bagging models achieved high accuracy for τ₀ and η prediction, verified by SHAP-based interpretability. • SHAP revealed Ms dominates structural build-up (τ₀), while L/S and W/B mainly govern viscous transport (η).

Mechanistic insights into the rheology of recycled concrete powder based geopolymer using SHAP-integrated machine learning

Abstract

Journal

Organization