January 17, 2008
-
An email I sent to someone about the NTSC findings.
Earlier this week the National Transportation Safety Board (NTSB) found a design
flaw in the collapsed I-35W bridge. “Gusset plates” are the “hubs” that the
girders are attached to. To give a basic example, think of the roof area of a
big old full sized four door car. That’s the approximate size of these gusset
plates, think towards the large side of this. The NTSB said they should have
been at least one inch thick but were half-inch thick in areas where the bridge
started to fall.
On the local TV news they compared one-inch thick to a VHS tape (actually
15/16th”) and half inch to an 8MM “compact” video tape. Close but the standard
plastic case on an audio cassette is also close to 1/2″. Either one is huge by
the scale of everyday life. For example, that steel plates they put over open
digs in streets are usually half inch and they can support heavy truck wheels.
Bridges like this literally use millions of pounds of steel so the gusset
plates are a minor portion of the steel weight. Typically they are way over
engineered and overbuilt. The NTSB said that having them undersized came as a
surprise because this was rare.
One thing I noticed a few days after the collapse was an old inspection report
that said that gusset plates had corroded down to half an inch. Here in
Minnesota corrosion and rust are a part of life. They tend to get a “beachhead”
and work from there so a corroded/rusted surface is very uneven. This is not
something a bridge inspector in a sling would be expected to notice.
One local TV “expert” claimed that the bridge would have collapsed with the
same load even if it was new due to the thin gusset plates. I tend to disagree
with this. Forty years tended to further weaken the already thin gusset
plates. I stick by my prediction of the sticking bearings/hinges. If the
bridge cannot expand and contract with temperature something has to take the
stress. This something is likely the gusset plates. Had they been built
properly or overbuilt they could have endured this stress better. Steel is like
a prizefighter who can easily handle regular punches but is weakened by the
stronger punches. The thicker steel is like the prizefighter being in far
better condition.
Technologically, it is relatively easy to put a bridge design through a
computer design model and compare it to the original calculations. That was
recommended of all bridges but especially of this design, which is only one
percent of US bridges. This was a time when there was a strong pressure to
build bridges cheap. The easy parts of the Interstate Highway system were
finished. I recall the portion of 35W from the Twin Cities to Duluth was dubbed
“the road to nowhere” because of so many uncompleted portions. Also, there was
a confidence, perhaps an overconfidence that thing could be built “smarter”.
Aviation and “rocket science” had made great advances here because weight was so
critical but it was done with a lot of trial and error, with an emphasis on
“error”.
There are a lot of examples of “engineering disasters” but my favorite to use
as a comparison is the Citicorp building in new York City. On the skyline it
has the “lean to” roof. Structurally, it had an undetected design flaw that
would cause it to likely topple and fall from the winds that accompany the level
of hurricane that hits NYC every fifteen years. An architectural student
“playing” with the design” uncovered the design flaw and thought he had made a
mistake somewhere. It was checked and the design flaw was verified. A stealth
“fix” was done using a lot of “gusset plates”. They decided to over engineer
the Citycorp building during the fix to reassure.
