Scaling factors quantifying seismic load uncertainty with soil nonlinearity effect in displacement-based design of bridge abutments

Abstract

The displacement-based design methods for bridge abutments, essentially cantilever retaining walls, require an input of expected ground motion on-site. For major projects, these motions are obtained based on site-specific response studies. In other projects, it is merely a guess based on the PGA values prescribed in the design codes. Further, the seismic displacements in retaining walls are usually estimated using simplified rigid-plastic analytical models. These models do not account for the effect of soil nonlinearity, alteration of input signal due to soil-structure interaction, amplification effects, or material damping. The currently followed seismic load factor of 1 on the prescribed PGA does not account for the induced uncertainties for the reasons given here, which can often be unconservative and lead to failure. This study statistically quantifies the effect of these uncertainties and proposes scaling factors on the design PGA of the input motion to estimate more reliable seismic displacements accounting for the assumptions in the simplified rigid-plastic model. The nonlinear finite-element analysis of retaining wall in OpenSees has been compared with the analytical double-wedge model considering a realistic V-shaped mechanism in backfill. It includes analysis for 83 different cases of earthquake motions scaled to four different PGAs. Finally, the scaling factors are proposed to estimate residual as well as peak sliding and in-plane rotational displacements of cantilever retaining walls with and without shear key.