This study is on evaluation of two GMSM methods for estimation of bridge response parameters during seismic excitation. The main focus is on understanding the behavior and proper modeling of the structures used to conduct such evaluation.

A total of six bridges, three skewed and three curved bridges, design by the California Transportation Department, Caltrans, and located in Orange and Los Angeles counties are considered. They are modeled using OpenSees (McKenna et al. 2000) open-source structural analysis platform. The selected GMSM methods are labeled as Sa(T₁) scaling (where T₁ is the fundamental period of the bridge under consideration) and Conditional Mean Spectrum (CMS) scaling. For each method, several sets of scaled ground motions are generated. These sets of records are applied to the bridges and the respective responses are analyzed.

Curved bridges present numerical instabilities when using force-based elements to model their columns. Displacement-based elements are used instead. Pushover analysis of the bridges every 5º can be used to determine their strongest and weakest directions. For curved bridges, these are those defined by an axis tangent to the arc described by the deck in its middle point and the corresponding perpendicular axis, respectively. For skewed bridges, the transverse is the strongest direction, and the longitudinal the weakest. A method for computing the plastic hinge length of a column is proposed. Finally, the structural responses obtained using Sa(T₁) scaling and CMS scaling methods do not show any clear trend.