Need a book? Engineering books recommendations...

Return to index: [Subject] [Thread] [Date] [Author]

[SEAOC] FW: I-90 Partial Collapse / Gusset Plate Analysis

[Subject Prev][Subject Next][Thread Prev][Thread Next]
I got this interesting email today. If you choose to reply to this email, 
please make sure that you send a copy of your reply to the original author 
of the email. Enter the following email address in your Cc: field:

DvRinehart(--nospam--at)aol.com

Thanks

Shafat



----------
From: 	DvRinehart(--nospam--at)aol.com[SMTP:DvRinehart(--nospam--at)aol.com]
Sent: 	Saturday, June 15, 1996 6:42 PM
To: 	seaoc-ad(--nospam--at)seaoc.org
Subject: 	I-90 Partial Collapse / Gusset Plate Analysis

It has been suggested that you might have some thoughts or ideas concerning
the following.....

Late in the morning of 24 May 1996, four gusset plates buckled on the
westbound Interstate 90 bridge over Grand River, east of Clevland, Ohio.
Approximately four hours later the bridge was completely closed to
traffic, and remains so.

The bridge is a three-span cantilevered arched deck truss;
the approach spans are simple span girders.  The main truss span lengths
are 208'-3":297'-6":208'-3"; the six-panel suspended span is 178'-6" long.

At the first panel point of the second span (panel point 8), the lower
gusset plate joins the lower chord (L7-L8, L8-L9), two diagonals (U7-L8,
L8-U9), and a vertical (U8-L8).  The failure occurred in buckling in the
gusset plates on both trusses nearly simultaneously.  (Painters were
blasting on the suspended span, as close as we have to eyewitnesses.)  The
gusset around U8-L8 and L8-U9 virtually crumpled, dropping that part of
the joint into the pocket formed by the remaining members.  Vertical
deformation appears to be about 3 inches; longitudinal movement was about
an inch.

The bridge is on a slight (0 degrees, 28 minutes) horizontal curve, so one
would expect any transverse movement to be to the outside of the curve,
especially if caused by an overloaded truck.  The trusses both moved
south, to the inside of the curve, about five inches.

Panel point 9 is the location of the deflection joint/"fixed" end of the
suspended span.  The expansion joint at panel point 9' appears to be
closed in the bottom chord, but open in the top chord.  One possibility is
that the substructure may have been creeping toward the river for years,
slowly closing the joints in the structure.  (Our wet spring certainly
could not have helped!)  With the summer-like temperatures that week, the
big annual thermal expansion may have created its own expansion joint.

Richland Engineering Limited of Mansfield, Ohio, has been retained by ODOT
to design repairs for the buckled structure, and to determine if
preventive measures are needed elsewhere, on this structure or its twin.

If anyone has a rational method for calculating stresses in truss gusset
plates, I would be very interested in knowing it.  Our most honest
references,
the "old" books, recommend sizing gusset plate thicknesses by finding a
similar structure and using their thickness!  We have evaluated the
various gusset plates on the structures by various methods:  a) cutting
free-body diagrams at several locations and evaluating the cut edge by
conventional beam theory;  b) evaluating the "end bearing" of the member,
over a line defined by a 30 degree line from the first rivet/bolt on the
outside lines of connectors to the end row of connectors (according to one
research report in ASCE's Structural journal, this is appropriate for
tension; according to another paper in the Journal, it is terribly
unconservative in commpression, but no alternative is offered);  c) shear
in the connectors;  d) bearing of the connectors on the gusset plates.

We have settled (so far) on the stresses defined under a) and b) above,
comparing them to an allowable compressive force defined by the Euler
buckling load divided by the 2.12 factor of safety AASHTO uses in their
column allowables.  To determine the Euler buckling load, we used a column
of unit width the thickness of the gusset plate, with a length measured
along the line of action of the member in question from the last row of
connectors on the member to the first line of connectors in the chord
member.

If you have any thoughts on our method of analysis, I would appreciate
your input.

David Rinehart, PE
DvRinehart(--nospam--at)aol.com
Richland Engineering Limited
Mansfield, OH





...