Tensile constitutive law identification for ultra-high-performance concrete: interchangeability of test methods

This study presents a multi-method framework for the tensile characterization of Ultra-High Performance Concrete (UHPC) with enhanced autogenous healing capacity, developed to reflect realistic service-life conditions. The primary objective is to assess the reliability and compatibility of different test methods for identifying tensile constitutive behaviour in self-healing UHPC. To this end, a UHPC mix containing crystalline admixtures was cast on-site, then cut into thin beams and double-edge wedge splitting (DEWS) specimens. These were pre-cracked, subjected to sustained deformation using external bracing, and exposed to various environmental conditions. Tensile performance was evaluated through DEWS, direct tension (DT), and four-point bending (4PB) tests, with simplified inverse analysis used to extract tensile laws from 4PB data. Results showed a consistent, though modest, increase in tensile strength over time across all methods except DEWS specimens with narrow initial cracks, where early healing limited further evolution. DEWS and inverse analysis provided comparable results, while DT tests were more affected by residual deformation and exhibited lower values and greater scatter. Statistical comparisons using Bland–Altman analysis and Deming regression confirmed the robustness of DEWS and 4PB methods and identified systematic deviations in DT measurements. While the methodology shows strong potential for application in structural-scale UHPC assessment, further validation with a broader dataset and different UHPC formulations is necessary to ensure generalizability and refine its predictive capabilities.