Assembly construction is extensively employed in bridge construction due to its ability to accelerate construction and improve quality. To speed the recovery of bridges after major earthquakes, this study proposes an assembled connection for precast piers and footings based on assembly construction. The precast piers are connected to the footings using ultra-high-performance concrete (UHPC) post-cast cupped sockets. Two specimens are tested with a 1:4 scale, namely, the cast-in-place (CIP) specimen and, the UHPC cupped socket pier specimen. Finite element models (FEM) of a continuous girder bridge with cupped socket connections are developed and verified by experimental results. The seismic fragility analysis is conducted to investigate the difference between the cupped socket connection and the CIP connection. The experimental results showed that the plastic hinge was formed on the precast piers and there was little damage to the UHPC sockets. The results of FEA indicate that UHPC cupped socket piers have slightly higher seismic fragility than the seismic fragility of cast-in-place piers. Then, some methods were proposed to reduce the seismic fragility of UHPC cupped socket piers, and their availability was confirmed by comparing them with the seismic fragility of CIP piers. Finally, an example bridge with this connection is introduced to illustrate replacing prefabricated piers after an earthquake.

The uncertainties of earthquake ground motions have the most important effects on seismic responses of bridge structures, especially long-span bridges because of complex and special dynamic properties. The nonlinear dynamic time-history analysis is conducted for a two-pylon long-span cable-stayed highway bridge by using real earthquake ground motions rationally selected. The correlation between the important characteristics of earthquake ground motions and the probabilistic seismic demand assessment of the cable-stayed bridge reveals that the geometric means and dispersions of response spectra from selected ground motions have very significant effects on mean values, dispersions and probabilistic distributions of seismic demands of long-span bridges. The spectral shape of geometric mean spectra in the period ranges with large cumulative modal mass participation factors should be well matched to the target spectrum for improving the precision and computational efficiency of probabilistic seismic demand assessment. If earthquake ground motions are rationally selected, response spectral values at the periods with comparatively large modal participation mass ratios or PGA can be used as intensity measures and even provide more precise probabilistic seismic demand assessment than response spectral values at the fundamental periods.