Irregular plan geometries and soil-abutment-structure interaction, as well as the torsional components of ground motions (TGMs) are significant contributing factors that result in excessive torsional demand during seismic events that may lead to premature and asymmetric failure of shear keys. Therefore, a design method that ensures the effectiveness of shear keys in mitigating the seismic response of highway bridges is proposed. The proposed method follows the conventional approach to capacity-protect the substructure components, however, formalizes a practical procedure to specify desired deformation limits associated with i) gap size between the superstructure and shear keys, and ii) ultimate deformation capacity of shear keys. The efficacy of the method is demonstrated through nonlinear response history analyses of a series of example bridges. It is demonstrated that excessive inplane deck rotations and the extent of damage can be limited when the effectiveness of shear keys is maintained throughout the duration of seismic excitation. Furthermore, probable deck unseating may be prevented when the deck displacements are restrained due to shear keys that are designed to remain intact.

A careful evaluation has been carried out to reveal advantages and disadvantages of linear and nonlinear modelling in dynamic analysis. 4- and 7- story building models representing characteristics of about 500 existing buildings models in Turkey was used in analyses. In the study, displacement demand parameters such as roof drift ratio and interstory drift ratio obtained from linear and nonlinear analyses were compared using a total of 24 ground motion records including forward directivity effects (Set 2) as well as records (Set 1) recorded in type B and C soils. Although the seismic demands for Set 2 are obtained extremely high in the nonlinear models, the demand differences between Set 1 and Set 2 are not excessive for the linear models. In the region where the T/Tp ratio is close to one, the linear analysis predicts unrealistically high demands compared to the nonlinear analysis. Linear analysis results mostly show an increase or decrease depending on dynamic amplification effects. The effects of ground motion intensity and damage mechanism cannot be observed in linear analysis method. For all these reasons, it is recommended not to prefer linear modeling method when using time- history analysis.