Cheaper, faster path led to failure

Construction engineer John Tsikouras suspected that the salesmen were not telling him the whole story. The men from Newman Associates, a bolt distribution company, approached him in August 1999 and said they wanted his professional advice about a tunnel ceiling that was part of Boston's Big Dig. But Tsikouras said their questions seemed intended to get the Rhode Island engineer's approval rather than his expertise.

Would it be safe to hang the concrete ceiling from 7.5-inch-long bolts? No, Tsikouras said. What if the bolts were 5 inches long instead? "I can ask my mother that one," he replied: Even less safe, obviously. Then, the firm that designed the ceiling sent word through Newman that the ceiling would be far lighter than Tsikouras estimated. Would the bolts be strong enough then? Yes, Tsikouras replied, after making more calculations, "the bolts passed with flying colors."

With that, John Tsikouras had been drawn into the most disastrous episode in the 20-year history of the Big Dig, one that would cost a Jamaica Plain woman her life when the ceiling in the Interstate 90 connector tunnel caved in on her last July. Unbeknownst to Tsikouras, construction of the ceiling was already well underway in the summer of 1999, and even as he continued fielding questions into December, some ceiling bolts were popping loose, raising doubts about the durability of the whole ceiling.

Nobody told Tsikouras about the slipping bolts, a tip-off that his initial concerns might be valid. Rather than make the ceiling's supports stronger, Big Dig contractors with powerful incentives to avoid construction delays asked a veteran engineer with almost no knowledge of the ceiling to vouch for its design. The ceiling's builder cited Tsikouras's work as evidence of the ceiling's soundness, and days later Big Dig managers backed off their demand that all the ceiling bolts be retested or replaced.

"They deliberately misrepresented my work," said a dumbfounded Tsikouras, who said he spent only four hours in all answering Newman Associates' questions and never visited the tunnel. He said he made the calculations requested by Newman and the ceiling design firm, Gannett Fleming, but he told them he objected to the low margin of safety he was instructed to incorporate into some of the calculations.

"You must be conservative to cover all the screw-ups that people may do down the line," said Tsikouras, a professional engineer at Sigma Engineering International who consults on industrial projects from Maine to Virginia. "I'd rather be too conservative than on the other side" -- especially considering that the bolts were expected to hold tons of concrete over drivers' heads for decades.

A five-month Boston Globe investigation shows that conservative engineering was the first casualty in the Interstate 90 connector tunnel as Big Dig costs rapidly escalated in the late 1990s. Under unrelenting pressure to save time and money, engineers from Gannett Fleming and Bechtel/Parsons Brinckerhoff, a joint venture hired by the state to manage the Big Dig, repeatedly weakened the ceiling's design, increasing the risk that the bolts would give out:

Bechtel/Parsons Brinckerhoff decided to use concrete ceiling panels that were quicker to install and cheaper than the lighter ones used in the adjacent Ted Williams Tunnel, but more than twice as heavy.

Next, Big Dig managers successfully pressured Gannett Fleming to reduce by 40 percent the number of bolts used to hold up the ceiling.

Then, Gannett Fleming estimated each of the remaining bolts would have to hold far less weight than Tsikouras first estimated, leading them to recommend a bolt-safety test using less than half as much weight as that performed on the Ted Williams ceiling bolts.

Gannett Fleming officials insist that the ceiling they designed was safe and that they followed all the guidelines set out by Bechtel/Parsons Brinckerhoff.

"We are confident that our calculations are correct," said a statement from Gannett Fleming's Braintree office. "This tragic failure is not the result of a design error."

Company officials said they did not mislead Tsikouras or misrepresent what he did. Working through the bolt salesmen, Gannett Fleming officials said they gave him enough information to answer specific questions -- and he did so, placing his professional "stamp" on his two pages of calculations that showed the ceiling bolts were strong enough.

Bechtel/Parsons Brinckerhoff had no comment, but has cautioned against rushing to judgment about who's responsible. Officials at Modern Continental Construction Co. , the ceiling builder, said they followed the design they were given and expressed confidence in the quality of their work. Newman Associates of Canton did not return calls seeking comment.

Ultimately, the connector ceiling design was less sturdy than for any other tunnel in Boston, leaving little room for mistakes by the inexperienced construction crew that would erect the ceiling. The workers themselves sometimes questioned what they were doing. "I remember looking up there, and thinking about how much each slab weighed, and feeling a little uneasy. A lot of guys did," recalled John Lorden, foremen of the ironworkers in 2000.

Like Tsikouras, Modern and Gannett Fleming engineers also questioned the design, but their concerns were brushed aside by Bechtel/Parsons Brinckerhoff. Its engineers insisted on doing the project in the cheapest, quickest way, in part because their client, the Massachusetts Turnpike Authority, was utterly consumed with controlling the soaring Big Dig price tag.

By late 1999 when the ceiling was being installed, the state official in charge of the $14.6 billion project had became a desperate Johnny-one-note harping about the need for cuts. James Kerasiotes , a seasoned bureaucrat who revel ed in his image as a "hatchet man," had his staff at the turnpike combing over every budget item looking for trims.

"This is just a little liposuction," Kerasiotes once instructed his chief of staff to tell an official who would have to carry out an immediate $60 million spending cut. "Most important, tell him to be a pro and not a baby."

Because the Turnpike Authority had put Bechtel/Parsons Brinckerhoff in charge of the Big Dig, state officials were almost invisible in day-to-day decision-making. The joint venture's engineers were considered "authorized representatives" who could sign off on decisions without consulting state officials. Kerasiotes, who was forced to resign in April 2000 amid a Securities and Exchange Commission investigation into cost overruns, said he was never even briefed on connector tunnel ceiling problems.

From the day in June 1999 that laborers drilled the first ceiling bolt hole -- less than 200 feet from the site of the future accident -- they had difficulty installing the unusual bolts, which were fastened to the tunnel roof by a chemical adhesive called epoxy. By the fall, engineers for Bechtel/Parsons Brinckerhoff and the company that manufactured the bolts suspected that workers had made at least eight types of mistakes, from not adding enough epoxy, to over tightening the bolts, to drilling bolt holes through steel reinforcement bars inside the concrete. Any one of these errors, the engineers knew, could severely weaken the bond holding a bolt in place.

When newly installed bolts began coming loose, discussions among engineers and managers focused on who would pay to fix the problem, according to project memos. Officials at Bechtel/Parsons Brinckerhoff initially ordered Modern to replace all 1,000 epoxy bolts it had installed or retest their strength using almost twice the weight that Gannett Fleming had required. One Bechtel/Parsons Brinckerhoff engineer worried in field notes at the time that some bolts may "develop just enough strength to pass the pull-out test" Gannett Fleming devised, then fail under the weight of the ceiling.

But Modern's young project manager, Robert D. Coutts , made it clear that he would fight any attempt to make his company foot the bill for more testing, submitting a 50-page report with Tsikouras's calculations as part of the defense of the epoxy bolts' basic soundness. Bent on avoiding further cost increases, Bechtel/Parsons Brinckerhoff's top engineer for the ceiling, Robert E. Steffy , swiftly retreated from his earlier demand, which could have cost hundreds of thousands of dollars.

Instead, just 187 of the 1,000 suspect bolts were subjected to more rigorous safety testing. Although 10 percent of those flunked, Steffy decided against testing any more bolts -- including the ones where the ceiling would later collapse -- after a memo from his second-in-command attributed the failures to early mistakes by workers still on the "learning curve" of the job. Steffy, a journeyman engineer who would retire to Florida before the tunnel was complete, quietly paid Modern for the small amount of retesting it had done. The bill was only $16,396.50, well below the $50,000 threshold that would have required Steffy to justify the cost overrun to his superiors.

Steffy declined to answer the door when a reporter visited his Florida home. But Phil C. Aikele, Steffy's second-in-command, said, "There was nothing malicious. . . . I spent a lot of years on that project trying to make sure something like that [ceiling collapse] wouldn't happen."

Aikele said his Aug. 31, 2000, memo did not recommend against further testing, but he admitted he did not object when Steffy and others decided not to press for more rigorous safety testing of the remaining bolts. "That was for higher-ups to decide," said Aikele during a contentious interview in his suburban driveway, "They may have thought it cost too much money to justify."

Aikele suspects the real cause of the accident might be a defect in the epoxy Modern obtained from Sika Corporation: "Something happened to the epoxy. It decayed," he said, causing bolts to lose strength.

Officials at New Jersey-based Sika have said they know of no defects in their adhesives. However, there is little doubt that the seeds of the disaster were sewn in the early 1990s when Big Dig officials decided to use epoxy bolts.

Dieter Schrebler was so impressed with epoxy bolts that he drew them onto the ceiling blueprints for the Ted Williams Tunnel before he got permission from Big Dig managers. The engineer, who worked for a minor Big Dig subcontractor called Weidlinger Associates, had learned in the spring of 1992 that epoxy bolts could be embedded into concrete more securely than conventional screw-in bolts. Statistical information about the strength of the bolt he wanted to use was limited, Schrebler admitted in a memo to another designer at the time. However, the crew renovating the ceiling in the Callahan Tunnel already was using epoxy bolts, and reporting that they were cheap and quick to install. "Our work crews required only one day to acclimate themselves," boasted the superintendent of the Callahan project in a 1992 newsletter.

Usually, officials at Bechtel/Parsons Brinckerhoff -- whose engineers liked to boast that they "built the world" -- reacted badly to subordinates acting without their approval. Engineers at the notoriously top-down organization even criticized the rough design of the acclaimed Zakim Bridge over the Charles River because the architects took their idea to state officials first instead of to them. But the Bechtel/Parsons Brinckerhoff engineers endorsed the new-style hardware for the Ted Williams Tunnel, which was then expected to be a model for the nearby I-90 connector tunnel.

Over the next decade, structural engineers everywhere would become skeptical of using epoxy bolts alone to suspend objects overhead because they can "creep" out over time if installers fail to apply enough epoxy or make other subtle mistakes that can weaken the adhesive bond. Today, few heavy ceilings are held up by epoxy bolts -- "There's a reason for this. It's called gravity," quipped one epoxy bolt expert -- but Boston got four of them during the 1990s.

In practice, workers on the Ted Williams Tunnel ceiling discovered that the epoxy bolts weren't nearly as cheap or easy to use as the Callahan superintendent had said. To begin with, the tunnel roof that crews had to drill into was shot full of steel reinforcement bars that they were supposed to avoid, but could not, forcing them to redrill thousands of bolt holes. Then, if they didn't get all the dust and debris out of the holes, the epoxy didn't bond correctly, weakening the bolt's strength. Third, the designers of the Ted Williams set a high standard for safety-testing, requiring each of the 26,000 bolts to be able to hold 7,000 pounds suspended from them. Twelve percent of the bolts installed early in the project flunked, forcing crews to put in hundreds of replacements.

These ceiling bolt problems contributed to a 58 percent cost overrun on construction of the interior of the Ted Williams Tunnel, and no one wanted to repeat the experience. Big Dig managers decided in 1995 that future tunnels would have a special steel track installed in the roof at the time the concrete was poured. Ceiling hangers could be hung securely from this "unistrut" without using the bothersome epoxy bolts at all.

But crews building the connector tunnel ceiling still would have to rely on epoxy bolts because of a quirk in the construction schedule: The eastern 250 feet of the tunnel, where the accident would take place, had already been built without steel channels by 1995, because it was needed to give drivers access to the Ted Williams Tunnel.

Meanwhile, a series of design changes driven by cost-cutting dramatically increased the weight that each bolt would have to support.

In June 1995, Bechtel/Parsons Brinckerhoff decided to use dull gray concrete in the connector ceiling rather than the shiny enameled steel ceiling panels they had installed in the Ted Williams. The Williams panels were elegant, but they had been slow to install and expensive. Bechtel/Parsons Brinckerhoff engineers estimated that the Big Dig could save $65 million by using 4-inch-thick slabs of concrete instead, a move that would more than double the weight of the ceiling.

Three years after they increased the ceiling's weight, Bechtel/Parsons Brinckerhoff engineers pressured Gannett Fleming to reduce the number of epoxy bolts holding it up by about 40 percent, a change with the potential to save significant labor and time. Bechtel/Parsons Brinckerhoff engineer Robert J. Richard told Gannett Fleming in a June 23, 1998, e-mail that, "unless I'm missing something," each ceiling hanger could be safely suspended from two epoxy bolts instead of four. As it happened, Richard was missing something important: His analysis was based on much bigger bolts than Gannett Fleming planned to use, causing him to overestimate their strength by 12 percent.

Gannett Fleming officials reluctantly agreed to reduce the number of bolts, concluding that, even with the correct, smaller bolts, the ceiling was safe. But Gannett Fleming and Bechtel/Parsons Brinckerhoff had made several assumptions that minimized their estimate of the weight borne by each bolt, called the "design load." This load, an estimate of all forces affecting the bolt from the weight of ceiling panels to vibrations from traffic, can be calculated in several ways, and the ceiling designers repeatedly looked to lower the estimate.

First, Bechtel/Parsons Brinckerhoff officials told Gannett Fleming to slash the weight allowance for work crews walking on the ceiling while they made repairs, which by itself reduced the estimated weight each bolt would need to carry by 450 pounds. In March 1995, a consultant hired by the Big Dig had suggested the reduction as a way to save money by allowing designers to scale back the durability of ceiling materials.

Then, in their calculation of the weight burden on each bolt, Gannett Fleming designers left out precautionary factors such as the impact of vibrations when vehicles rumble through the tunnel, and the sudden jolt from an earthquake. Gannett Fleming's weight calculation also did not make allowances for the possibility that, over time, some ceiling hangers could be damaged or disconnected, shifting weight onto the remaining bolts.

As a result, Gannett Fleming estimated that each epoxy bolt would have to carry just 2,600 pounds, compared with the 5,715 pounds that Tsikouras, the Rhode Island engineer, initially calculated. Gannett Fleming said Tsikouras was mistaken and urged him to change his calculations . However, two structural engineering professors who were shown the ceiling's weights and dimensions by the Globe also estimated that the weight on each bolt was at least 5,000 pounds; a third placed the load between 2,750 and 3,150 pounds.

The consequence of Gannett Fleming's lower estimate was profound because the bolts' strength would be tested with just 25 percent more weight than the expected load on each bolt. That meant that this "pull-test" would be done with a simulated weight of 3,250 pounds in the connector, less than half the test load used in the Ted Williams.

But Gannett Fleming officials stand by their 1999 estimate of the weight on each bolt, suggesting the real problem may have been poor installation. "The failure of the ceiling panels occurred when the load on them was considerably less than the design load and far less than at the level at which they were tested," said the company in a statement.

Nonetheless, when Modern won the contract to build the road, walls, and ceiling inside the connector tunnel in November 1998, the company inherited a design that was measurably less sturdy than other tunnel ceilings held up by similar bolts in Boston: The ceiling panels were more than twice as heavy as in the Ted Williams, Sumner, and Callahan tunnels, while the bolts were smaller.

Within days of winning the contract, Modern managers set about trying to get rid of the concrete ceiling. Coutts and senior Modern executives believed they could devise a lighter and cheaper alternative that would still achieve the ceiling's purpose: to create an emergency ventilation chamber that would suck out toxic fumes during a fire .

Modern asked Environmental Interiors, a New Hampshire-based company, to develop plans for a metal ceiling that Modern estimated would be 80 percent lighter than concrete and could save $4 million. However, Bechtel/Parsons Brinckerhoff engineers worried that the metal ceiling would vibrate too much, and they rejected the plan, largely because there was not enough time to modify the design before construction was set to begin in June. A Vermont company was already manufacturing the original concrete ceiling panels.

Anticipation filled the dusty air inside the connector tunnel on June 10, 1999, when two laborers rode 30 feet up on a powerlift to drill the first bolt hole for the ceiling while managers and consultants watched or took pictures like proud parents.

Though Manuel Dias and Fernando Carvalho were longtime laborers, they -- like most of the union workers -- had little experience building a ceiling with epoxy bolts. They quickly learned that the work was a pain, requiring them to hoist a 50-pound drill on their shoulders for 10 minutes at a time to make bolt holes. Often, they had to drill holes again because they had drilled through the steel reinforcement bars inside the concrete roof, making the hole useless because steel forms a weak bond with epoxy. Bechtel/Parsons Brinckerhoff supervisors suspected that crews sometimes concealed the "rebar hits" to avoid having to make yet more bolt holes.

Two-man crews of laborers drilled more than 2,500 bolt holes over the length of the tunnel by 2000, and while the work was slow, they were under immense time pressure. Modern's contract was so tightly scheduled that a delay of even a few days would cause back-ups for other contractors working in the tunnel. Modern also had won the contract with an extremely low bid of $64.1 million -- one-third below what Big Dig managers had expected -- and that placed the company in danger of losing money if the work fell behind schedule.

Because of the time pressure, even when bolts began coming loose in September 1999 after portions of the ceiling were hung, workers kept installing other sections of the ceiling while Modern and Bechtel/Parsons Brinckerhoff debated for months what to do.

The ironworkers who hung the ceiling panels from the bolts were more inexperienced than the laborers. With construction in Boston near an all-time high, ironworkers were so scarce that their union put out a national appeal for workers. The inexperience showed, as 12 ironworkers were injured on the ceiling project from the end of 1999 into August 2000, far more than normal for the Big Dig . At the time, Coutts, Modern's project manager, blamed the injuries on "the hiring of workers which may not have the experience ordinarily expected in heavy construction."

For engineers, construction of a suspended ceiling is pretty routine compared with the challenge and glamour of working on other phases of the Big Dig. That meant the ceiling project's design and construction were left mainly in the hands of junior people, who may have been more susceptible to cost-cutting pressures.

Likewise, Bechtel/Parsons Brinckerhoff assigned journeymen engineers -- neither of them licensed -- to oversee the tunnel's completion. Resident engineer Steffy, then 52, had come to Boston from out of state in early 1998 to oversee the tunnel project, and he would return in 2002, buying a house on a quiet cul de sac in Apopka, Fl a. His lead field engineer, Aikele, then 49, had relocated from out of state to join the Big Dig a year earlier. Neither had the cachet -- or likely the clout -- of the career Bechtel and Parsons Brinckerhoff engineers who managed more high-profile work.

The project manager for Modern, by contrast, was a 38-year-old who grew up in Dedham and was making a name for himself in the largest contractor on the Big Dig. Coutts, who declined to be interviewed, zealously protected his company's interests.

Thrown together with an inexperienced workforce in a highly pressurized situation, Coutts and Steffy clashed over time and money when bolts began coming loose after the ceiling was hung , despite having first passed the pull-test at 3,250 pounds. Looking for an inexpensive way out, Steffy agreed in January 2000 to pay for a small amount of safety testing at a higher weight and, if the results "prove good," he would drop the issue, he wrote in an e-mail to Aikele.

Over the next several months, inspectors tested 187 epoxy bolts in the HOV lane, including those in the section ironworkers had installed first. In that section, called the "mock-up area," 17 of the 75 bolts pulled loose when they were tested at 6,350 pounds, a staggering 21 percent failure rate. Outside of the mock-up, only two of 112 bolts failed the tougher test, less than 2 percent. Bechtel/Parsons Brinckerhoff's Aikele considered the improved performance outside of the mock-up area a positive sign:

"This could be the result of better installation techniques . . . gleaned from going through the 'learning curve' in the mock-up area," he explained to his boss, Steffy, in a memo on Aug. 31, 2000. Since the weak bolts had been replaced, he suggested the issue be "closed" in the HOV lane.

But laborers Dias and Carvalho had actually drilled the first bolt holes in the eastbound lane of the tunnel near where the accident would later occur, meaning those bolts, if tested, might have had as high a failure rate as seen in the mock-up area.

Moreover, Aikele and other engineers had substantial evidence that workers had made mistakes or used the wrong equipment throughout the tunnel, not just in the HOV lane.

Steffy dropped the testing issue entirely in early 2001. Aikele said Steffy consulted about whether to continue safety testing with higher-ranking Bechtel/Parsons Brinckerhoff people, but Steffy refused to comment, and a Globe review of thousands of pages of documents has revealed no documentation on who made the decision.

Peter Zuk , who left as state Big Dig project director in December 1998, said he would have recommended more safety-testing, though he said in a recent interview he understood how the issue could have slipped through the cracks. "There was enormous commercial pressure to all the parties in this," he said.

The first firefighters who arrived at the scene of the ceiling collapse last July 10 at 11:19 p.m. could not believe that there was a car underneath the layers of concrete, let alone that Milena Del Valle's husband, Angel, had somehow managed to climb out the narrow space where the driver's side window had been. But there was no helping 38-year-old Milena, mother of three, who lay slumped across the driver's seat.

Investigators for Attorney General Thomas F. Reilly quickly zeroed in on the key evidence: ceiling bolts lying in the road that had almost no epoxy coating at all -- "like they just came from Home Depot," one turnpike official said -- while the adhesive on other bolts was brittle and discolored as though it had not been mixed correctly.

Five months later, white X's spray-painted onto the pavement are the only visible reminder of the tragedy. Overhead, the suspended ceiling has been ripped out and not replaced in the easternmost 250 feet of the tunnel because the area is already well ventilated.

State engineers concluded the ceiling was never really necessary after all.

Brian Ballou of the Globe staff contributed to this report. Scott Allen can be reached at allen@globe.com. Sean P. Murphy can be reached at smurphy@globe.com.