The reliability of numerical models for masonry-infilled reinforced concrete frames is limited by significant parameter uncertainty. Existing sensitivity analyses often neglect crucial aspects like parameter interactions, the effect of structural scale on results and the generalisability of findings from simple to complex systems. This paper aims to address these fundamental gaps to provide a more robust understanding of model uncertainty.
A global sensitivity analysis using the Sobol method is applied to a nonlinear numerical model based on the Crisafulli modelling technique. The study assesses parameter influences on four key outputs (Y) from the capacity curve (stiffness, strength, drift and energy). The robustness of the findings is then investigated against changes in structural scale (1:2 vs 1:1) and complexity (single-frame vs multi-story, multi-bay systems).
The analysis identifies distinct parameter groups governing each output () that can catch 90% of the total variance. A key finding is that these influence hierarchies are remarkably robust to changes in structural scale. Conversely, they cannot be extrapolated to complex multi-story, multi-bay systems, although the identified still explain over 81% of the total variance in these configurations.
The originality of this work lies in its approach. It provides a quantitative assessment of parameter interactions (synergies) and, crucially, investigates the robustness of sensitivity results against both structural scale and complexity – a rarely explored aspect. Based on these findings, this study proposes an evidence-based, hierarchical methodology for calibrating the nonlinear parameters of infilled frames.
