Constitutive model selection for unreinforced masonry cross sections based on best-fit analytical moment–curvature diagrams

• Nonlinear moment–curvature analysis of URM cross sections was carried out.
• The influence of constitutive models and ultimate limit state assumptions was studied.
• Theoretical moment–curvature diagrams were compared to experimental data.
• Constitutive models allowing the best moment–curvature response simulation were identified.
• The model error of moment–curvature diagrams was statistically characterised.

A significant number of macroscopic constitutive models is available to simulate the behaviour of unreinforced masonry (URM) in compression, but they produce different predictions of sectional response to eccentric compressive loading. In this paper, nonlinear moment–curvature analysis of rectangular URM cross sections is presented to explore the influence of alternative constitutive models and ultimate limit state assumptions in terms of post-peak strength degradation and strain ductility. Theoretical moment–curvature diagrams are quantitatively compared to experimental data at multiple levels of load eccentricity through goodness-of-fit measures. Different constitutive models are found to produce a good simulation of experimental moment–curvature behaviour, depending on whether experimental data are disaggregated in terms of masonry type or not. Moment–curvature diagrams are derived by keeping constant either the magnitude or the eccentricity of axial load. Finally, the effects of tensile properties and model error of theoretical moment–curvature diagrams are assessed.