A Bridge That Didn’t Collapse
ON NOVEMBER 7,1940, LEONARD COATSWORTH, A REPORTER FOR the Tacoma News Tribune , earned a small place in history as one of the last people on the Tacoma Narrows Bridge. In the four months since its opening, the bouncy bridge had become a bit of a tourist attraction. People came from miles around for the thrill of driving across the galloping span. But Coatsworth was no tourist, just a local toting a earful of beach gear and his daughter’s cocker spaniel. And that morning the bridge was no fun at all. Usually, on windy days, the bridge would heave up and down like a dying fish gasping for air. On that morning, though, it was thrashing like a swordfish at the end of a line—angry, strong, and very much alive.
Coatsworth got about halfway across before he lost control of the car and had to stop. He leaped from the swaying auto, slamming his face on the pavement in the process, and began a panicked half-walk, half-crawl to the end of the bridge, leaving the cocker spaniel in the car. Hands swollen, knees bleeding, gasping for breath, Coatsworth finally made it off the bridge. He then stood by the toll plaza with a group of observers and watched the bridge tear itself apart and collapse into the Narrows. One of the observers, Professor F. B. Farquharson of the University of Washington, captured the spectacle on film, documenting for posterity the death of one great suspension bridge and one small dog named Tubby.
Then the story got interesting.
Soon after the collapse a New York bridge engineer named David Steinman sent out a series of letters to politicians and engineers involved with Tacoma Narrows. In the letters he explained how he could have saved the bridge if only the world had listened. Steinman was no crackpot. He was a prominent engineer and author, and his firm, Robinson and Steinman, had been building bridges since the 1920s. He had been interested in the Tacoma project from the beginning but was, to his chagrin, not involved in Leon Moissieff’s final design. Trying to be a gracious loser, Steinman had kept his distance from the project—until the bridge opened in July 1940 and started swinging in the breeze. Then he knew he had to get involved.
Steinman had encountered similar problems on two of his own suspension bridges and had tamed the sway (to some extent) with simple fixes: a couple of clamps to hold the main cable in place and a few wire ropes strategically rigged to steady the roadway. One of these bridges, the Deer Isle-Sedgwick, in Maine, had structural similarities to Tacoma Narrows and thus was especially applicable. Solution in hand, Steinman had contacted Lacey V. Murrow, chief engineer of the Washington Toll Bridge Authority, and offered his services. Steinman’s letters after the bridge collapse noted, a bit smugly, that Murrow had turned him down.
Whether or not Steinman’s solution could have saved the Tacoma Narrows Bridge is debatable. To this day some mathematicians and physicists disagree sharply on what caused the collapse. For all the advances that have been made in mathematical modeling and computer analysis, there is still a significant element of trial and error in bridge design. But two things are certain: First, the Deer Isle Bridge, though smaller, has some striking similarities to the ill-fated Tacoma Narrows. And second, the Maine bridge still stands today, sixty years after it opened, while its cousin in Washington tumbled into the drink before its first birthday.
Why the separate fates for two similar bridges? The answers vary, depending on whom you ask, but it is likely a combination of factors. Deer Isle’s smaller size certainly helped, and most believe that Steinman’s solutions did too. Still, everyone agrees that the bridge hasn’t had an easy time of it. Deer Isle, while still standing, remains quite sensitive to wind. Over the years it has been patched and repatched with all manner of cables, braces, and other gadgets to protect it from gusts and gales. It’s safe to say that the original Tacoma span, if it were still around, would be in similar shape. The story of these two bridges, theirups and downs and intertwined histories, is a lesson not only in engineering but in two of our favorite Yankee attributes: ingenuity and stubbornness.
The Deer Isle Bridge, 2,308 feet long and painted a pale, industrial green, sits on the Maine coast about 35 miles south of Bangor. It begins on the mainland, in the town of Sedgwick, stretches over an inlet of Penobscot Bay called Eggemoggin Reach, and comes down on Little Deer Isle. The bridge was opened to traffic on June 19, 1939. Its design, like that of the Tacoma Narrows and several other spans of the time, was profoundly influenced by the era of its birth, the Great Depression.
In those days bridge engineers were under pressure to lower costs by using as little steel as possible. Style dictated that bridges look economical as well, with thin towers and long, slender roadways. Both Deer Isle and Tacoma Narrows carried only two lanes. Deer Isle’s main span is 1,080 feet long, with a roadway only 23.5 feet wide, while the roadway at Tacoma Narrows, more than double the length of Deer Isle at 2,800 feet, was only 39 feet wide. Narrow roads make for light bridges, with little weight to steady them. A case in point (or more precisely, a study in opposites) is the George Washington Bridge, which connects New York City and Fort Lee, New Jersey. When it opened, in 1931, its 3,500-foot main span was the longest in the world. The bridge was designed to carry a two-deck roadway, but the plans were scaled back because of the poor economy. The bridge opened with only one deck, but that deck was a monster, carrying eight lanes of traffic and weighing about 31,590 pounds per linear foot. The roadway’s bulk was heavy enough to dampen any oscillations, and the bridge has never had any problems with the wind. When the lower deck was added, in the early 1960s, it steadied the bridge further. Tacoma Narrows and Deer Isle weighed in at only 5,700 and 2,400 pounds per linear foot, respectively. Compared with heavyweights like the George Washington, the two bridges were light as a feather and thus vulnerable to wind. This vulnerability increased when designers added devices called plate girders to stiffen the roadway. The plate girders, while cheap and easy to install, led to unforeseen problems.
At a basic level all suspension bridges have the same design. There are usually two or three towers and two main cables, attached (sturdily, one hopes) to the earth on one side of the water, then slung up and over the towers and attached to the ground on the other side. From these main cables dangle many thinner cables, and from the thinner cables dangles the roadway.
It may be discomfiting to think of a bridge as a roadway dangling in the air, but that’s exactly what it is. However, anyone who has driven over a suspension bridge knows that it feels like a solid road, not like one of those treacherous rope-and-plank bridges so popular in playgrounds and adventure movies. That’s because suspension bridges are stiffened. Without stiffening, “if you put a truck in one spot, all the load would go to the ropes that were right next to that truck,” says Frank Mondello, a bridge engineer with Steinman Boynton Gronquist & Birdsall (the successor to Robinson & Steinman). “It would tend to kink the cable, like running your finger on a clothes-line.” When a truck drives over a stiffened bridge, the roadway can’t bend, for the stiffening system distributes its weight to many cables instead of just a few.
The most common stiffening apparatus is a truss, metal beams arranged in a series of triangles pointing up and down along the sides of the roadway. (The Golden Gate Bridge provides the most familiar example.) But during the 1930s, guided by considerations of economy and fashion, the engineers of Deer Isle, Tacoma Narrows, and four other bridges opted for a sleeker stiffener than the truss: a series of large, solid steel plates called, as a whole, a plate girder. The plates line the roadway like a wall, giving it the appearance of a long, slender watering trough.
Plate girders are not only graceful, they’re easier and cheaper to attach than a truss. “You’ve got a lot less parts with a girder,” explained Everett Barnard, a bridge maintenance engineer with the Maine Department of Transportation. “You just throw up some plates, throw in some rivets, and you’re done.” But then the wind problems begin. A breeze that can easily blow through the open triangles of a truss is blocked by the solid wall of a plate girder. It’s what engineers call a bluff body; the wind can’t go through, so it swirls over and under and around, setting the bridge rolling.
Deer Isle and Tacoma Narrows were bluff, they were lightweight, and they had a third problem caused by their length-to-depth ratios. In the 1930s engineers were still trying to figure out how deep a suspension span had to be, relative to its length, in order to be safe. The profession eventually settled on a maximum recommended length-to-depth ratio of 150. The Williamsburg Bridge in New York, famously ugly but steady as a rock because of its very heavy stiffening truss, has a length-to-depth ratio of just 40. Deer Isle, with its 6½-foot-deep plate girders, is in the danger zone at 166. Tacoma Narrows, only 8 feet deep with a 2,800-foot main span, had a ratio of 350. There was bound to be trouble.
Bridge engineers at the time knew much less about aerodynamics than they do now. Steinman, like all his colleagues, was struggling to understand the interplay of steel and wind. Things started to click for him as work was finishing on his Thousand Islands Bridge, in upstate New York. The bridge (actually two separate bridges) was oscillating, undulating, and swaying in the wind. So before it opened in 1938, Steinman added diagonal stays—wire ropes attached to the road near the towers and fanning out to various points on the main cable. More important, he attached cable ties—rigid steel braces that clamped the main cables to the stiffening girders. The devices seemed to quiet the problem on Thousand Islands, so he turned to them again when the same problem emerged on the Deer Isle.
Steinman’s partner, Holton D. Robinson, was confident that the problem had been licked. In January 1939 he wrote: “Recent experience has shown that light bridges of this type with shallow stiffening members … have a tendency in the main span to undulations similar to wave movements [that] can be unpleasant for timid passengers even though it does not indicate any lack of strength or safety in the structure.” The unpleasant movements could “be practically eliminated,” he added, “by bracing the stiffening girder [and] installing diagonal stays.” Robinson estimated the added cost at “slightly less than $5,000.”
With the brace and stays installed, Robinson and Steinman considered the problem solved. A year later, when the Tacoma Narrows Bridge began to oscillate, Steinman suggested the same solution to the Washington Toll Bridge Authority. The authority turned him down and subsequently paid the price for its arrogance. Or did it?
That’s a difficult question to answer. The Tacoma engineers, having witnessed wind problems on other plate-girder suspension bridges, had expected some motion on their span. They believed, though, that because their bridge was so long, its weight would dampen most oscillations. When this proved untrue, they installed devices to control the span, including cable ties similar to those on Steinman’s bridges. But the cable ties on the Tacoma bridge, while they did clamp the main cable to the stiffening girder, differed from Steinman’s. His were rigid steel rods that resisted both pulling and pushing motions. The Tacoma ties were cables made of steel rope. They resisted pulling motions but went slack when pushed. On the morning of the disaster, the ties went alternately slack and taut in time with the motion of the bridge. Eventually the strain was too much to bear, and the ties broke before the bridge collapsed. Some engineers say that stronger, rigid center ties might have helped the bridge that day, but it’s impossible to know for sure.
As for Steinman’s other fix, the crisscrossed wire stays, the Tacoma engineers never seriously considered using them. “Such stays,” said the final report on the disaster, “would have been ineffective if not dangerous on the Tacoma span.” Some engineers today, like Ken Serzan of Steinman Boynton Gronquist & Birdsall, agree that the cable stays probably wouldn’t have helped much. Harold Bosch, a bridge engineer with the Federal Highway Administration, is more generous. “The cables did improve the Deer Isle Bridge, and did help it to survive,” he said. “But they didn’t totally solve the problem. With cables the Tacoma Narrows might have done better. Whether it would have eventually collapsed anyway, who knows?”
In the aftermath of the Tacoma Narrows collapse, Steinman remained confident of the soundness of his bridge. Deer Isle did fare better when its own trial by wind came two years later, on December 2, 1942. That day a severe storm ripped down the Maine coast, smashing into the bridge with winds of up to 80 miles per hour. The bridge oscillated in 12-foot waves that snapped a quarter of the cable stays, cracked the expansion joints, caused the suspension cables to slip through the cable ties, and generally wrought havoc on every part of the bridge. The damage was extensive, but when the gale blew itself out, the bridge was still standing.
The storm left Steinman more convinced than ever that he was on the right track. What the bridge needed, he decided, was more of the same. Steinman added another, far more extensive set of cable stays, and he didn’t stop there. By 1944 there were vertical stays between the tops of the towers and the roadway, transverse stays that crossed over the roadway, and zigzag stays from the roadway to the main cable and back, crisscrossing the secondary cables the whole length of the span. The bridge became an enormous cat’s cradle of metal rope, as if it were woven by the hands of a giant.
With Tacoma Narrows in ruins and Deer Isle being strung together ever more tightly, what became of the other plate-girder suspension bridges that emerged from the Great Depression? How did their designers react to Tacoma’s collapse? Most of them didn’t have to. The hefty George Washington, as we’ve already seen, resisted winds with its massive bulk. The two Thousand Islands spans built by Steinman settled down after they opened. They still get the jitters from time to time, but they aren’t long enough to be in any danger. The Anthony Wayne Bridge, in Toledo, Ohio, 785 feet long, 59 feet wide, and 10 feet deep (for a length-to-depth ratio of 78), is also too sturdy and inflexible to be bothered by the wind. Only the Bronx-Whitestone Bridge in New York City had to be altered significantly.
The Bronx-Whitestone opened in 1939 with a risky length-to-depth ratio of 209 in its 2,300foot main span, worse than Deer Isle but better than Tacoma Narrows. But with a roadway 74 feet wide (later expanded to 100 feet) that carried four lanes of traffic, it was heavier than our two subjects. Nonetheless, the bridge had its share of oscillations, though none so dangerous as those on Deer Isle and Tacoma Narrows. In 1943 its designers added a series of wire stays from the tops of the towers to the roadway.
Steinman mocked this decision for years afterward. “They used single stays and ran them in the wrong direction, from the tops of the steel towers down to the roadway,” he told a Conference of scientists a decade later. “Stays running from the tower tops were effective with masonry towers, as in the monumental Brooklyn Bridge; but such stays are largely ineffective when the towers are flexible.” In 1946 the Bronx-Whitestone engineers, still unsatisfied, decided to go even further, attaching a truss along the top of the plate girders. This improved the length-to-depth ratio and increased the span’s weight. The fix tamed the bridge, but it still experienced small oscillations from time to time. In the 1980s engineers added a tuned mass damper (a large, heavy mass that is calibrated to oppose oscillations) to quiet it down, and it has been relatively calm ever since.
Meanwhile, Tacoma Narrows was rebuilt. This time the engineers greatly overdesigned the span. And who can blame them? They abandoned the elegant plate-girder design in favor of a sturdy truss. At 33 feet deep, it changed the bridge’s length-to-depth ratio from a dangerous 350 to a very safe 85. Engineers also widened the span by 20 feet, put slotted wind grates between the lanes, and installed oil-filled hydraulic damping mechanisms at the towers. The new span weighed almost 3,000 more pounds per foot than the old one. It has never had any problems with wind.
Deer Isle, in typical Yankee fashion, stuck stubbornly to its original design. As a result, it has continued, on occasion, to suffer Tacomalike oscillations. “I went over it one time while it was really bouncing,” a Deer Isle resident named Neville Hardy told a local magazine. “My front tires were coming off the surface.” Back in 1941, in the final report on the Tacoma disaster, engineers compared the vertical and twisting motions of seven modern suspension bridges. Only Deer Isle’s waves were anywhere near so severe as those of Tacoma. Yet, steadied by Steinman’s cables, Deer Isle held on, until the Maine weather struck again. On February 3, 1972, the “Groundhog Gale” hit the bridge at 90 mph, causing wild oscillations but amazingly little serious damage. Even so, officials decided the bridge needed more shoring up and added another set of fixes. Workers installed six stronger cable ties, further clamping the main cable to the roadway. They also added metal plates to the top of the stiffening girder and metal triangles called web stiffeners between the roadway and the wall. These fixes improved matters somewhat but did not eliminate the oscillations. In 1978 Joan Mondale, wife of the Vice President, was stranded on the island for several hours when the bridge began undulating dangerously in what news accounts described as a “light breeze.”
By the time I drove up to Maine last year to see Deer Isle in person, I half expected to find it held together with :hewing gum and duct tape. I couldn’t have been more wrong. Heading down Route 15 toward Deer Isle, I got my first view of the bridge from a hill overlooking Eggemoggin Reach. Out of more than half a million bridges in this country, there are only 144 suspension bridges. As I looked at the graceful, arching Deer Isle Bridge from the top of that hill, the scarcity seemed a shame. Driving over the bridge, I saw Steinman’s cable stays weaving in and out of the main cables, crossing the roadway with giant X’s in some places, creating complex patterns in others.
The day after my arrival I met Everett Barnard on the mainland side of the bridge. The tall, soft-spoken Barnard, who has been with the Maine Department of Transportation since 1959, oversees the maintenance of about 2,800 bridges in the state. He and David Cole, a bridge maintenance manager for the DOT, gave me a tour of Deer Isle Bridge on a bright, breezy day in June.
I soon learned that walking across the bridge is not quite as lovely as viewing it from the hill. It’s not because of oscillations; there were none. But the “sidewalk” is an open grating 15 inches wide, partially blocked about every seven feet by a web stiffener. As we walked, cars and trucks rumbled by. An ambulance with a large, protruding rear-view mirror passed very, very close. To avoid thinking about the traffic heading toward me, I sometimes looked down through the grate to the water below. Near shore I could see straight through the crystal water, clear to the pebbles and mussel shells scattered across the sandy bottom. At the center of the bridge, the Reach is much deeper, the crystal shoreline replaced with a churning green abyss. As we passed the halfway point, Barnard stopped to point out some storm damage from the 1972 Groundhog Gale. It was a bad one for Deer Isle; Barnard says they had to close the bridge for a few days. There was never any danger of its collapsing, he adds.
When we reached the other side, we walked down to the beach to take a good look at the span. Admiring the view, I turned to David Cole and asked him if the bridge was one of his favorites. Cole, who had been silent for most of the tour, replied: “No comment.”
Barnard pitched in: “I think they all have their own unique thing.”
“Their own personalities,” added Cole.
“And what’s the personality of this one?” I asked.
“Arrogant,” he replied.
Later I told Harold Bosch, the Federal Highway Administration engineer, what Cole had said. Bosch laughed.
“I think it does have a personality,” he said.
“It does some interesting things, and it surprises us—it continues to surprise us, even though we’ve learned so much about it.”
Bosch is the research director of the George S. Vincent Aerodynamics Laboratory, part of the Turner-Fairbank Highway Research Center in McLean, Virginia. It was Bosch, armed with modern equipment and 50 years of accumulated knowledge not available to the original Tacoma engineers, who finally calmed the Deer Isle Bridge in 1993. He had been studying the structure since the late 1970s, when the Maine DOT asked the Federal Highway Administration to take a look at it. Bosch was sure that the shape of the bridge—a narrow roadway with a plate girder—was contributing to its problems with wind. The problems were probably increased by the bridge’s lightness and flexibility, as with the original Tacoma Narrows Bridge. To test his theories, Bosch’s lab built a scale model of a section of the Deer Isle—extremely detailed, right down to the grated sidewalk—and put it in a wind tunnel, where he hit it with winds so hard it almost broke. (“Generally we don’t test models to destruction,” he said. “Models cost money.”) As he suspected, the Deer Isle section responded to wind in much the same way as a Tacoma section he had tested: Both twisted and both moved a lot. After about a year and a half of testing, Bosch knew his initial assumptions were right. He also knew the best way to solve Deer Isle’s wind problem: Change the shape of the bridge.
This wasn’t as hard as it sounds. The Deer Isle was having problems because its solid plate girders were about as aerodynamic as a Mack truck; they needed to be streamlined. Bosch had just the plan to do it. He designed a set of attachments, triangular in cross-section, that would be affixed to the outside of the bridge, facing their noses into the wind. The sharp edges would slice into the oncoming wind, turning the truck into a race car.
Bosch had proposed such attachments, called fairings, for other bridges, and they had worked well. But he was unsure whether the Deer Isle fairings should be symmetrical—dividing the wind flow evenly over the top and bottom of the bridge—or asymmetrical. He didn’t have time to run extensive wind-tunnel tests before making his recommendation, so after a preliminary analysis he bet on asymmetric fairings. Further testing later proved him right. His hunch, though counterintuitive to nonengineers, makes sense. A perfect triangle would split the wind evenly, so whatever swirls or turbulence appeared on top of the roadway would be mirrored underneath. “In theory,” says Bosch, “this would be great. The net effect would be zero. Unfortunately it doesn’t work that way.” What happens instead, he says, is that identical and alternating vortices can sometimes induce resonance vibrations. “And that,” says Bosch, “is something we don’t want to see in any structure, unless it’s a guitar string.”
The asymmetrical fairings create a slightly lopsided flow, stopping any resonance before it starts. At least that’s how it has seemed to work on the Deer Isle. The DOT installed Bosch’s fair- ings in 1993 along the entire length of the bridge. Most observers agree that the bridge has quieted down considerably.
Could fairings have helped Tacoma Narrows? Perhaps. Before the collapse engineers had begun to discuss ways to change the bridge’s shape, by either adding fairings or drilling holes in the girder. “The problem with Tacoma is that there wasn’t much time,” said Bosch. “Even if they had designed something, to implement it in three or four months would have been quite a task.” Most engineers agree that the bridge was in trouble from the moment it was designed. Some would argue that the same is true of Deer Isle. To Bosch, though, this air of unruliness is what makes the bridge so attractive. “It’s such an interesting bridge to study,” he said. “It’s so much like Tacoma Narrows; there are so many aerodynamic issues. So many modern bridges we see are, well, kind of dull.”
Even though Deer Isle has calmed down, Bosch keeps an eye on it. He has 18 wind sensors on the bridge and several accelerometers on the roadway measuring vibrations. He also has a video camera on one of the towers. The data feeds into a computer housed in a small white clapboard building on the mainland side of the bridge. Bosch can download the data directly to his office in McLean, but he usually just waits for his man in Maine to send him the data tapes when they’re full. Unless, of course, it’s a special occasion. “If a storm’s coming, I may be on the phone not once a day, but thirty times a day, checking in on the bridge,” he said. “I can check in from home, and a lot of times if there’s a big gale on the coast, or a little bit of leftover hurricane heading toward Maine, I’ll be up at three in the morning, checking in on the bridge.”
Although his fairings have gone a long way toward stabilizing the Deer Isle span, Bosch knows as well as anyone the dangers of being too confident. The science of building and maintaining a suspension bridge remains a surprisingly provisional one, so it’s nice to know that at least one person is watching over it. And nice, too, to know he’ll have something to watch for a long time to come.