Impact of modeling decisions on seismic loss and fragility assessment of steel buildings

Seismic loss estimation is an essential tool for assessing the performance and resilience of buildings exposed to seismic hazards. Numerous loss methodologies with varying fidelities exist, with each comprising different steps and requiring analysts to adopt certain modeling decisions and assumptions. This study compares three seismic loss estimation methods, namely FEMA P-58 component-, story loss function (SLF)-, and HAZUS assembly-based approaches, using a database of 621 steel special moment-resisting frame buildings with diverse designs and geometries. In addition, a set of sensitivity analyses are performed to evaluate the influence of key modeling decisions in fragility and loss assessment stages on different loss methodologies. These decisions include (1) intensity measure-engineering demand parameter (IM-EDP) formulation to derive fragility functions, (2) EDP proxy for fragility models within the assembly-based method, (3) characterization of demolition fragility, and (4) uncertainties in nonstructural component quantities. The results show that the assembly-based approach estimates lower median loss for short- to medium-rise buildings but higher estimates for taller ones, with significantly higher variability in losses for multi-story buildings compared to the component-based approach. In contrast, the SLF-based method yields slightly higher median loss for 1-story buildings but consistently lower estimates for multi-story buildings than the component-based approach. The sensitivity analysis highlights that the critical modeling decisions vary by the selected method. The assembly-based approach shows greater sensitivity to the choice of EDP proxy impacting the dispersion of losses. In contrast, uncertainties in nonstructural component quantities predominantly govern both the median and dispersion of losses for the component-based approach, and the selected IM-EDP formulation has the highest effect on SLF-based estimates.