Physics-Based As-Damaged Modeling of Reinforced Concrete Shear Walls Using Graph-Based Quantified Visual Crack Patterns

This research proposes a comprehensive approach to address the assessment challenges of reinforced concrete (RC) shear walls subjected to multi-hazard cyclic scenarios. Traditional methods focus on the assessment of single-hazard conditions and often ignore the patterns of cracking when modeling damaged concrete, which limits their ability to fully represent the compromised state of structures subjected to successive hazards. Furthermore, the inherently subjective and manual nature of current structural assessment procedures undermines the reliability of such practices. Therefore, developing objective and autonomous methods that facilitate robust and reliable infrastructure assessment is a necessity. This study presents a novel approach to fill this critical gap by embedding observed mosaic crack patterns from an initial hazard (e.g., cyclic loading) into structural models, providing a more realistic approach. Experimental data of two concrete shear walls with aspect ratios of 2.00 and 0.94 are used to investigate the proposed approach. In this regard, the crack patterns of the walls under cyclic loading are converted to their representative graph through the developed novel crack-to-graph conversion method, transforming the visual damage into quantifiable indicators of reductions in structural properties (e.g., stiffness) suggested by FEMA 306. Afterward, lumped-plasticity models are developed to model the as-designed (initial) and as-damaged state of the walls. The results from the quantified crack patterns are used to model the as-damaged state of the walls without prior information on the loading and stress–strain history of the walls. To simulate a multi-hazard cyclic scenario, this paper takes the first few cycles of the cyclic loading as the initial hazard (damage) and the rest of the cycles as the subsequent hazard. The results showed that the extracted graph features predict the stiffness reduction factors of the walls with a low 7.0% root mean squared error (RSME) and a high coefficient of determination of 0.94. Building on this quantification, the stiffness reduction factors are employed in the crack-pattern-based lump-plasticity models. Results show that all models compare well with the experimental data obtained after the corresponding as-damaged cycle, capturing the loading path, peak strength, and unloading path. Crack-pattern-based lump-plasticity models offer a computationally efficient alternative for rapid post-hazard assessments. This physics-based approach is compatible with existing software, guidelines, and engineering codes, thereby supporting more accurate, condition-based evaluations of damaged RC shear walls and enhancing current assessment and retrofitting practices for improved post-hazard resilience.