How does NH decide which bridges to fix first? Inspections, engineering, analysis - and a little bit of “we want this”

Tuesday, February 07, 2017

When writing about the annual update to the state’s lists of bridges in need of repair I discovered, much to my surprise, that the transportation officials in New Hampshire aren’t interested in my straightforward method of deciding which work to do when.

It’s simple: Just fix the bridges that I use.

Instead, they’ve adopted a big, clunky system to set priorities that involves sending trained engineers to gather actual information about the physical condition of each bridge, then factoring this data into an algorithm based on society’s needs and wants.

That seems like a lot of unnecessary work to me, but I thought I’d check it out anyway in the name of feeding the insatiable maw of reader curiosity. (Plus, my other column idea fell through, but you don’t need to know that.)

Nicholas Goulas, chief of the Existing Bridge Section within the New Hampshire Department of Transportation, walked me through the system, which has been tightened in recent years by federal and state governments, partly in response to the 2007 collapse of a bridge carrying Interstate 35 over the Mississippi River in Minneapolis.

“They wanted less human influence and more data, basically putting them in a hopper to suggest bridge repairs based on numeric scores. . . . They want to make sure that bridge projects aren’t arbitrarily inflated or ignored,” he said.

New Hampshire has eight engineers with the DOT who inspect our 3,846 bridges, each of them at least once every two years. Bridges previously detected as being in rough shape are inspected more often, sometimes every few months: “We don’t want to wait two years between inspections if we think conditions will change rapidly,” Goulas said.

A bridge in New Hampshire, incidentally, is defined by law as spanning at least 10 feet, which means that a not-insignificant number are held up by culverts. The federal definition, by contrast, says a bridge has to span at least 20 feet. There are three main components to a bridge, explained Goulas: The decking that we drive on, the superstructure that crosses the water or roadway being spanned, and the substructure that holds up the bridge.

Goulas gave some examples of how the inspectors proceed.

For the decking, a coat of paving covers many sins, so they go underneath to see if reinforcing steel is visible.

On the superstructure, he said, “You look for corrosion-related deterioration that’s starting to cause measurable change – your ¾-inch bolt on your flange is measuring 0.4 inches – maybe there are some holes in the steel beam.”

For the supporting substructure, suspicion centers on the expansion joint, left to provide a place for structures to swell and shrink as temperature changes. “Salt-laden bridge runoff goes into the joint, no matter how careful you are,” Goulas said. Also they check for timber or steel elements that have been exposed by flooding or changes in water flow.

All of this data is analyzed with templates and spreadsheets and computer-aided design software, with engineering calculations for such things as changes in a bridge’s load-bearing ability and descriptions a federal document with the staggeringly uninteresting name of “The Coding Guide.” This is used to assign each bridge a score of 0 to 9, from worst to best, for each of the three categories.

Any score of 4 or less in any category puts the bridge on the red list, where more than 300 bridges of various sizes, about half owned by the state and half owned by the local municipality, already sit.

How to choose which ones to fix? Well, consider three city-owned bridges in Concord:

A half-mile stretch of Washington Street crossing a canal on a bridge built in 1975 gets a 5 or “fair” rating for deck and superstructure, but only a 3 (“serious”) for substructure.

Several miles of Route 9, or Loudon Road, over the Merrimack River, which dates from 1966 through 1996 depending on section, gets a 6 or “satisfactory” for superstructure and substructure, but a 4 or “poor” for the decking.

Birchdale Road over Bela Brook, which was built in 1928, gets a “poor” 4 on all three categories – but it’s only 22 feet long.

All need work, but if the city doesn’t have the money to fix them all right now, how does it choose among them? By length? Amount of work needed? Traffic flow? Number of letters in their name?

Each municipality can decide on the priority of spending local money (usually matched at least 80 percent by the state), but if these were state-owned bridges, there’s a set procedure.

They would be ranked according to condition (all the physical stuff that inspectors found); importance (a function of number of vehicles and detour distance if the bridge was closed); risk of failure (both scour-critical and fracture-critical, which you’ll have to look up yourself because I’ve only got so much room to write here); whether weight restrictions have been placed; and, finally, type (girder, movable, culvert, timber, truss) and size.

Toss those factors into the hopper, meaning the algorithm, and the resulting score presents a totally objective measure of which bridge work should be done first. Right?

Well, not entirely.

There is a weighting given to the five components. In this case, 37 percent of the final score takes into account the condition, 21 percent the importance, 6 percent each the risk of failure and weight restrictions, and 30 percent the type and size.

Those percentages are chosen because of experience and judgment about what matters. But unlike calculations about, say, the load-bearing ability of a span, they’re not mathematically objective measurements of physical reality.

In other words, the priority list that determines which projects will actually get done reflects, at least partly, our wishes and hopes as society. This is inevitable because, after all, the state Legislature and Executive Council, not some algorithm, approve the budgets that govern the actual repairs. We can’t keep human whim out entirely.

All well and good, although I have to point out that this process is much clunkier than my alternative: 1. Does David drive on it? 2. If “yes,” make it better.

Now that is an efficient algorithm

(David Brooks can be reached at 369-3313 or dbrooks@cmonitor.com or on Twitter @GraniteGeek.)