Page 365 - Forensic Structural Engineering Handbook

10.52 CAUSES OF FAILURES

The girder: The girder that failed during removal was the Outside North Girder in Span
8, cantilevering into Span 9. Prior to its removal, the west end of the 192-foot girder
was supported on Pier 7; it spanned 160 feet across Span 8 to Pier 8, then continued
east as a 32-foot long cantilever beyond Pier 8 into Span 9. It was a built-up plate
7 3
girder of carbon steel. The web plate was 91-in. tall by / 16 in. thick with 6 × 4 × / 8
3
vertical stiffeners on both sides at approximately 6'-5'' spacing, and a pair of 8 × 8 × / 8
flange angles running the length of the girder along its top and bottom edges. All top
and bottom flange cover plates were 24 in. wide by / 8 in. thick, varying in number from
5
one to three layers along the length of the girder—appearing to be consistent with the
design moment diagram. The girder had a field splice shown at 70 feet from the can-
tilever (east) end and 122 feet from Pier 7 (west end) on erection drawings.
Prior to disassembly of the bridge, transverse floor beams and horizontal braces had
framed into the interior (south) side of the north girder transmitting vertical loads and
providing lateral bracing. After cutting away the floor beams and braces, 1- to 2-foot-long
stubs of these members appear to have remained attached to the interior (south) side of the
girder. Being off-center, these became eccentric loads that created torsion in the girder—
twisting and laterally bending it towards the south.
Before the pick, the 192-foot girder was a propped cantilever with a 160-foot main span
(between Piers 7 and 8) and a 32-foot cantilever (from Pier 8 toward Pier 9). After the pick,
when the cantilever (east) end of the girder was lifted up from Pier 8, the girder became a
propped cantilever with a longer 181.5-foot main span (192'– 10.5' = 181.5') between pick
points and a shorter 10.5-foot cantilever. The entire moment diagram and the location of
the maximum moment changed, resulting in larger moments and higher bending stresses at
weaker sections of the variable cross section girder.

Structural analysis. A simple structural analysis of the girder in its position when the
cantilever end of it was lifted was performed. The purpose of the analysis was to
examine the approximate available ultimate strength of the girder in bending as limited
by lateral-torsional stability when lifted up off Pier 8 and supported on Pier 7 at its west
end and on the crane line near its cantilever east end, and to compare that strength to
what was required of the girder.
The available ultimate strengths, i.e., ultimate moments controlled by lateral torsional
buckling at selected cross sections of the girder, were calculated by hand in accordance with
the AISC and AASHTO LRFD design specifications. The required strengths (bending
moments due to the self-weight) of the variable cross section girder were calculated with the
aid of a computer using the ANSYS structural analysis software. For simplicity, all vertical
loads were assumed to act in the plane of the web. The results of the simple analysis indicated
that the girder had inadequate strength and stability when lifted. This was the result even with-
out considering the twisting and destabilizing effect of the eccentric weights on the south side
of the girder. More accurate stability analyses could have been performed using a sophisti-
cated nonlinear finite element computer software, but in light of the convincing results of the
simple analysis, the time and expense of greater sophistication were not warranted.
Cause(s) of the failure: It was the author’s opinion with reasonable engineering certainty
that the girder failed in a lateral-torsional buckling mode. The reasons for the failure were
the girder’s instability on account of its slenderness (due to the narrowness of its top
flange relative to the long length between the pick-locations). The eccentric weights of
the floor beam and brace stubs that remained attached to the south side of the web plate
created twisting moments and consequent lateral bending in the girder immediately
upon lifting, which further compromised its stability. A secondary contributing factor
to the failure was the nature of the suspended support at the end of the crane line which
allowed lateral displacement (swaying) of the east end when lifted off the pier. Other