Straight and curved steel I-girder bridges - Computational Analysis and Design of Bridge Structures - page 206

that are employed by the software. For instance, a given element will
have a unique formulation, interpolation, integration, and software
implementation, all of which will affect results. However, if properly
modeled, in the forensic or load test cases, such refined analysis is com-
monly adopted due to its refinement and accuracy. 3D FEA models are
described here:

a. In-plane shell-beam model . Hays Jr. et al. (1986) and Mabsout
et al. (1997) modeled the deck slab using quadrilateral shell ele-
ments in plane with five DOFs per node and the steel girders
using 3D beam elements with six DOFs per node (Figure 7.10).
The bridge deck slab and steel girders shared nodes where the
steel girder is present. This model is essentially a 2D FEA, and
it is not capable of capturing the effect of the offset between the
center of gravity of the steel girder and that of the deck slab.
Furthermore, it cannot capture the system's actual boundary con-
ditions, that is, the supports in the actual system are located at
the bottom of the steel girder rather than at the center of gravity
of the deck slab.

b. 3D brick-shell model . Tarhini and Frederick (1992), Eamon and
Nowak (2001), Baskar et al. (2002), and Queiroz et al. (2007)
used eight-node linear solid brick elements with three displacement
DOFs in each node to model the concrete deck. The girders were
modeled using quadrilateral shell elements, which contain three
displacement and two rotational DOFs per node (Figure 7.11).
The cross frames were modeled using 3D two-node truss elements
with three displacement DOFs per node. Tarhini and Frederick

Shell element Shared node

Beam element

Figure 7.10 In-plane shell-beam model.