Simple and Complex Modelling of Seat-Type Abutment-Backfill Systems

Abstract

The response of the seat-type abutment-backfill system under a dynamic excitation and its contribution to the structural system of the entire bridge is usually ignored in practice in Europe, since the designers prefer providing joint gap sizes larger than the required for the design earthquake. In the high seismic hazard areas of the US, various versions of Caltrans Guidelines prescribe a relatively simple way to account for the abutmentbackfill interaction. However, the design of Caltrans abutments is based on the 'fully sacrificial' approach, wherein the backwall 'shears off' at an early stage, while in other countries the detailing of the deck-abutment interface is such that a plastic hinge forms at the base of the backwall which is detailed for ductile behaviour. In all cases, if assessment of the bridge safety beyond the design earthquake is sought (e.g. in fragility analysis), it is essential to properly account for the response of the bridge when the end joint is closed.This paper focuses on seat-type abutments with backwall hinging. In a practical context, a 'simple' model in this case consists of a spring-gap element that models the entire abutmentbackfill system, while a 'complex' model includes explicit modelling of the abutment using beam-column elements, and of the backfill behind it using one or multiple soil springs. For dynamic response-history analysis, dashpots are also needed for modelling radiation damping. The issues of the number and the arrangement of the spring-dashpot systems and their nonlinear constitutive laws are addressed herein and several configurations are studied. SAP 2000 is used for analysing a typical overpass bridge with seat-type abutments and joints in both the longitudinal and transverse directions, for a number of spectrum compatible records. A series of pushover analyses of the 'complex' model are also carried out; their output can be used to define the (single) spring properties of the simple model. Interesting conclusions are drawn, both with regard to the spring configuration and to the difficulties in combining the various nonlinear elements in SAP 2000.

Optimization of the seismic response of bridges using variable‐width joints

“…A (relatively) complex one shown in Figure 4B, wherein the entire abutment is modelled using beam‐column elements (with the possibility of plastic hinge formation at the base of the backwall), the backfill is modelled using a number of springs and dashpots along both the stem wall and the backwall, and the shear keys are modelled through two nonlinear springs in the transverse direction, separated from the deck through gap elements; in OpenSees , the two are merged into one nonlinear element with flat initial part until gap closure, an option that prevents numerical instabilities caused by the use of separate gap elements 13 ). A more involved model including an extra spring (in series with that of the shear keys) to model the transverse stiffness of the embankment was also explored 13 , but the effect on the bridge response was small and it was dropped in the parametric analyses. The constitutive law for the nonlinear springs representing the backfill behaviour (Figure 4B) included the envelope proposed in 14 , and the compression‐only hysteresis loops of the backfill were based on the unloading/reloading stiffness values suggested in 15 for various types of backfill soil.…”

Modelling of the Full-Range Response of Abutment-Backfill Systems

“…Therefore, modelling methods that include nonlinear beamcolumn elements and soil springs to represent the structural parts of the abutment and the backfill soil, respectively, have been proposed (e.g. [10], [13]), while [14] developed a methodology that results in a simple single-spring model that is able to capture effectively the nonlinear behaviour of the entire abutment-backfill system with 'hinging' backwall.…”

“…This is a compression-only constitutive model that follows a hyperbolic backbone curve, which has been shown to properly capture the actual backfill soil behaviour found in experimental campaigns like [23]; it includes an initial zero-valued branch to represent the gap between the deck and the backwall allowing the user to avoid the implementation of separate gap elements, which are known to be prone to numerical difficulties (e.g. [14], [24]) and it is also able to capture the accumulated damage in the backfill soil. However, it has a number of drawbacks, the most important arguably being that it does not offer the option of defining a post-peak softening behaviour, although this behaviour has been observed in pertinent tests (e.g.…”