When Millimeters Matter
Putting a square peg into a square hole is no trick. But when the “peg” is a new, 6,500-ton, 350-foot-long section of the San Francisco-Oakland Bay Bridge, and the “hole” is an opening in the existing bridge with just 3 inches to spare, the feat becomes mighty impressive--especially when it’s done in record time. That’s just what happened on Labor Day weekend last year when an old section of the bridge was removed and the new section was slid into position, on specially constructed rails, to replace it.
Behind the ScenesThe Bay Bridge, a crucial connection between San Francisco and Oakland and one of the busiest bridges in the United States, was shut down Friday, August 31, 2007 at 8:00 p.m. Immediately after the shutdown, multi-ton strips of the bridge were sliced off and pulled out by clawed backhoes. Seventy hours later--11 hours ahead of schedule--the bridge was again open for traffic. Local news organizations declared the operation “as smooth as sliding in a drawer.” California Department of Transportation (Caltrans) news releases shortly after the event were understandably exuberant; Caltrans Director Will Kempton said, “We’re tremendously proud to be able to open the bridge early.”
But a deeper, more interesting story was somewhat obscured in all the excitement about “the slide.” The success of Labor Day weekend resulted from four months of painstakingly accurate survey work, from control setup through as-builts and layout, and an innovative monitoring technique used on the day of the big move. Throughout the whole project, inaccuracies of just a few millimeters could have led to a transportation nightmare instead of a triumph.
Innovative Technique“This was a relatively new technique, only used a few other times,” explains Jim Brainard, a partner at Folsom, Calif.-based R.E.Y. Engineers Inc. and lead surveyor on the project. “The existing Yerba Buena Island section of the Bay Bridge could not be seismically retrofitted properly after the 1989 Loma Prieta earthquake [the “World Series Earthquake” that killed 67 people and caused $6 billion of damage] and this was a way to replace it with minimal traffic disruption.”
In essence, the “retrofit-by-replacement” strategy meant that a wider and seismically safer section of bridge deck was built 30.95 meters (approximately 100 feet) south of the section to be replaced. Since this project was initiated during the phase from 1993 to August 2004 when Caltrans was mandating use of metric, all measurements were in meters, centimeters and millimeters. New columns, to accommodate the new deck’s greater width, were built alongside the existing columns. Meanwhile, temporary columns were built to the south that matched the existing and new columns, and the new deck was built on the temporary columns. Eight rails, or skids that looked like sidewalks with a rail in the middle, were built for use between the two spans.
When the big day came, the existing deck section was torn out and 16 enormous jacks were used to move the new section into place. While it may sound easy, or at least straightforward, tolerances were extraordinarily tight. One rail, for example, was rebuilt during construction because it was a little more than 6 millimeters out of alignment. And each column, or “bent,” had a socket that received a 3-foot-long, 6-inch diameter “bent pin” sticking out from the deck; according to Brainard, these had to be a “perfect match.”
Multiple as-built surveys were required, beginning with the existing columns, which averaged about 30 feet in height. The existing columns varied from design by 1 to 5 centimeters, so the designs of the new deck, new columns and temporary columns were altered to match. The new and temporary columns were also surveyed after construction and they too proved to vary slightly from the revised design. So the new deck design was adapted again and adjusted during construction. Several rounds of as-builts kept construction within tolerances.
Tight Control NeededTolerances were tight for this project: as-builts and points set for bridge layout had to be within 3 millimeters. Consequently, control also had to be tight … which meant some extra work. “When we got to the site, we had control from Caltrans,” Brainard explains. “There were plenty of points to work with but, due to time or seismic activity, they didn’t fit the required accuracy for the bridge layout.”
To verify that the accuracy issues were local, R.E.Y.’s crew performed static GPS work using Trimble (Sunnyvale, Calif.) R6 and R8 receivers, and conventional traversing using a Trimble 1" S6 total station tying into 13 primary control points from the existing network and establishing six new control stations. Each traverse station had four sets of horizontal angles and at least eight distances recorded. The data was adjusted through the least squares adjustment portion of Trimble Geomatics Office software. The final adjustment resulted in only four primary control stations being constrained, yielding a network that was accurate to within 0.0015 meters. A digital level was used to further strengthen the vertical control.
Martin Gonzales and Tim Pringle were lead surveyors for R.E.Y. and assisted by Jay Vanyi; all three provided layout and construction of the retrofit and were the onsite surveyors during the weekend of the replacement. After R.E.Y.’s control update, they did initial control checking with the Trimble S6 robotic total station. “We used the gun throughout the project,” Gonzales says. “It was quick and accurate, and the target recognition worked well.”
Careful Layout“Instruments are so good these days, with all the robotics and other controls, that there’s hardly any human error,” Gonzales says. “It’s made construction layout much simpler-we work with a lot more confidence and less checking than we used to.” Nevertheless, on this job, Gonzales and his crews went to great lengths to verify the position of each point set. “We ran a traverse around the site, closed it and adjusted it, and then we used it for a lot of checks. We’d check at least a few points from each stakeout from multiple setups, and on really critical parts of the project we’d go over everything at least twice. Basically, we double-tied everything.” All of the checking confirmed that the layout was going very well and that the robotic total station--and its operators--were getting it right the first time.
“This was a very smooth job,” Gonzales says. “The only real issue we had was with the layout on several of the rails, and that had nothing to do with our work or the instrument.” On the initial rail layout, Gonzales was asked to set the beginning and end points, and construction crews would snap chalk lines between the two. This proved to be a poor technique, as there was significant deflection along the snapped lines. Fortunately, a foreman asked for a spot check. “It looked like the lines weren’t pulled tight enough,” Gonzales says. “We found some marks that were off by 28 millimeters. If the rails had been built with those lines, they would have had an arc to them, and that could have been bad. Fortunately, we caught them in time.”
The Big DayR.E.Y.’s team was assigned to be the onsite monitor on the day of the move, and team members all fretted a bit about the proper technique. After all, construction surveying is very rarely a high-speed, real-time event, and few surveyors have had to shepherd a 6,500-ton object across a 30.95-meter gap with all of San Francisco waiting on the outcome. R.E.Y. was in largely uncharted waters. “I’ve done high-rise surveying and a lot of bridge work,” Gonzales says, “but I’ve never moved anything this massive.” Initial plans called for plumb bobs to be hung from the new deck and checked against reference points during the slide, but Gonzales realized that wind and the movement of the deck would make this impractical. He had some restless nights thinking it over, which proved to be a good thing: “I woke up one night and thought, ‘Hey, all I have to do is stake out two lines!’”
What Gonzales realized was that four points-one each at the beginning and end of the two outermost rail lines-would be enough to track the movement of the entire deck. Since the move was not continuous, but rather a series of precise, one-meter steps, he would be able to check deflection continually during the move so that adjustments could be made as needed.
Gonzales set up two tripods on the rail lines farthest away from his setup. On the nearer line, for a more secure setup, he set “story poles.” These were two-by-fours laid horizontally and projecting 25 millimeters beyond the deck, about a third of the way into the projected gap between the new and existing decks. He placed reflective targets on the ends of the two-by-fours.
In the days before the move, one hydraulic jack was replaced because it failed to meet specs, and a test lift and 9-inch move were performed successfully. On Labor Day weekend, demolition and removal of the old section of deck was first on the agenda, followed by the fitting of rails into place with hydraulic lifts. And then it was R.E.Y.’s turn in the spotlight.
“I don’t think I could have done this job without the robotic total station,” Gonzales says. “I had two lines to stake out, and after each 1-meter move, which took several minutes, I had about 30 seconds to take four shots and calc the offsets. The tracking feature and instant updates worked perfectly.”
Crews were trying for equal movement at each corner of the replacement deck, and, according to Gonzales, everything went amazingly well: “It was beautiful! After every move I took four readings and gave crews the displacement. The biggest adjustment was less than 40 millimeters.”
The move went slower in the final steps. The last few hundred millimeters took five hours, as very fine adjustments were made prior to lowering the deck into place. It landed within 6 millimeters of its design location. Paradoxically, when it was all done, not much was different-two years of design and four months of construction and a very ambitious weekend of work resulted in a bridge that was much safer and a bit wider, but looked much the same.
There were a few glitches during the weekend, like a shortage of trucks and tougher-than-expected rebar, but overall this was a project that wildly exceeded expectations.
Congratulations All AroundCrews must have been tempted to take advantage of the rare Bay Area bridge closure by getting out the grills and throwing a party on the new deck. They wouldn’t have been the only ones celebrating-Caltrans had arranged special access during the closure for four weddings on nearby Treasure Island, and those “big days” also went as planned. Instead, Caltrans and the Bay Area Toll Authority (BATA) quickly called a triumphant press conference to announce that the Bay Bridge would be reopened at 6 p.m. on Labor Day, well before the projected 5 a.m. Tuesday reopening.
Other work was happening during the deck replacement. Since most of the bridge is scheduled for retrofitting, with work to be completed in 2012, R.E.Y.’s crew traversed and did additional as-built surveys of the East Span deck where a future detour bridge will be built by multifaceted development firm C.C. Myers of Rancho Cordova, Calif. BATA took advantage of the closure to paint, redo striping, install signage and add a new FasTrak toll lane. The new lane may be the only change that many motorists notice. But those in the know, those who sweated the millimeters on this retrofit for more than four months and over one long weekend, will all agree with Caltrans Director Will Kempton’s simple summation: “It was an amazing project!”
A time-lapse video of the entire move may be viewed at www.mtc.ca.gov/news/info/movies/bay_bridge_video_9-07.htm.