Skip links

The Westminster Street Bridge Was A Bad Idea

I came across the bridge above while looking at “high bridges” two weeks ago. That was the Westminster Street Bridge in Saxtons Falls, Vermont, across Saxtons River, a small tributary of the Connecticut River. The village has a population of under 600, which is small even by Vermont standards, so the bridge is usually referred to by the neighboring towns of Bellows Falls and Westminster. Circa 1907, the falls had obviously been blocked by a wood grillage weir, but that’s long gone. The steel truss bridge shown was constructed in 1891 to replace a wood covered bridge and was in turn replaced by a concrete bridge (still extant) in 1926. (My thanks to the Rockingham Library.)

As a matter of general philosophy, I try not to directly criticize engineering design decisions in this blog, even when I disagree with them. Almost all arguments in design have multiple options, and I don’t know everything (or sometimes anything) that influenced the design. When talking about old engineering designs, I try to remember that people in the past were just as smart as we are, but lacked some of our tools. They had fewer material options, more limited analysis methods, more limited tools, fewer methods of making connections between structural members, less research data on the behavior of structural elements under load, and so on. Given what they accomplished with the tools they had, I sometimes think that they were smarter than us. That said, I hate this bridge design with a burning passion and I think it’s important to explain why. In discussing structures here, I often treat design decisions as if they were simply a matter of preference, but that’s not true. Some design are more efficient than others, and some are more resilient than others.

The basic structure of the bridge is a pair of warren trusses with eyebars for the lower chord and diagonals, built-up H sections with laced webs for the verticals, and built-up H sections for the upper chord. Given the date, the metal is almost certainly all steel. So far, so good, and pretty standard stuff. The wood deck is carried on carried on I-beam stringers, which are carried by built-up plate girders running between the trusses in line with the verticals. Again, standard. Then we deviate badly from ordinary practice…

There are three usual configurations of the deck in a truss bridge. The deck can be at (or near) the elevation of the bottom chords, with lateral bracing between the upper chords, which is a through truss; the deck can be at the elevation of the top chords, with lateral bracing between the lower chords, which is a deck truss; or the deck can be at the elevation of the bottom chords without bracing at the top, which is a pony truss. Under ordinary circumstances, pony trusses were only used when the truss depth needed for the span was too small to allow for a through truss and the site made it difficult to build a deck truss. Pony trusses have a bracing problem: in a simply-spanning truss, the top chord is in compression and therefore has a tendency to buckle sideways. If it is not braced, the stress must be kept low to prevent buckling.

The Westminster Street bridge is none of the three regular options. The deck has been dropped a few feet below the top chord creating a hybrid deck/pony configuration. There’s lateral bracing at the lower chord, where it helps against wind pressure but does nothing else. The upper chord is restrained against buckling by the verticals acting in weak-axis bending. In theory this works, but it’s an inefficient use of the verticals on top of an inefficient use of the top chord. They’ve put cables between the verticals, to allow the top chord to act as a primitive safety rail, but this emphasizes the worst part of the design, which is the vulnerability of the top chord. This bridge was in use long after motorized traffic became common, and the lateral push from a car hitting the top chord could trigger an overall buckling failure. In a deck truss, traffic can’t hit the trusses; in a through warren truss, traffic can only hit the verticals (unlikely to cause an overall failure) or the diagonals (in tension and therefore not subject to bucking); in a pony truss, the top chord is usually (not always) above the elevation of a projected impact. Using the top chord as a safety rail is as close to a self-defeating design as I’ve ever seen.

Also, the connection from the cross-deck girders to the verticals will tend to put some weak-axis bending into the verticals, which is a similarly terrible idea.

For a cute little bridge, this one has a design that should never have been built.