A builder of the new San Francisco-Oakland Bay Bridge failed to disclose that a 19-foot section of concrete in the foundation of the span's signature tower had not hardened before it was tested. By keeping quiet about the problem, the builder prevented further examination or repair.
The Bee found descriptions of the apparent defect in records provided by Caltrans last fall to reassure the public about the overall stability of the suspension segment of the bridge's eastern span. Experts said the problem, combined with other construction and testing lapses by the California Department of Transportation and its contractors, raises new questions about the structural integrity of the bridge.
Kiewit-FCI-Manson, a joint venture, built the foundation as part of a $177 million contract. It did not provide the problematic 2007 test results until after a Bee investigation in November showed that a Caltrans employee skipped required test preparation for separate checks of the same foundation and fabricated results on other structures.
The agency plans to open the $6.5 billion structure, the costliest public-works project in state history, by Labor Day 2013 to serve an estimated 100 million drivers annually. Caltrans said the bridge is sound and can withstand any anticipated earthquake.
Beyond the large area of suspicious concrete in one of the reinforced underground foundation piles, a Bee examination of Caltrans records found numerous other problems with the piles, and gaps in essential data. Experts who reviewed concrete and engineering records for The Bee questioned the ability of the main tower foundation to resist an extreme earthquake – the reason for building the new bridge.
Among The Bee's findings:
Two of 13 piles that rise out of the Bay to hold up the tower contain suspect and inadequately tested concrete. Sonic-wave tests revealed a 19-foot section of poor concrete in Pile 3, in a location subject to profound seismic forces. When tested, the concrete had not hardened to the required strength. It was not retested.
For unclear reasons, Pile 8 either received no sonic test or builders could not locate the test report. Job-site inspection diaries also show construction abnormalities in that pile.
Sonic test reports contained more than 20 errors. Among other slips, they misstated which piles were tested, test dates and pile measurements. Experts said the unusual volume of mistakes casts doubt on the reliability of testing for both the problem pile – Pile 3 – and others deemed free of defects.
Builders treated the piles with an additive meant to increase concrete strength, but known to cause soft or poor-quality concrete when overused – one possible explanation for the 19-foot anomaly. Although batching computers should ensure mixing precision, records for a different pile show unexplained mixing errors by a concrete plant computer.
Caltrans and its experts said the bridge is safe. Many of their supporting assertions were contradicted by agency documents. For example, a Caltrans panel asked to review the work said sonic tests proved that the piles were of sound construction, despite the Pile 3 problem and the lack of testing for Pile 8. Panelists relied heavily on tests of what they called "full scale" mock-ups. Those models actually were a small fraction of the bridge piles' size. Independent experts said the mock-ups offered invalid comparisons.
Larry Olson, president of Olson Engineering, based in Wheat Ridge, Colo., which conducted the sonic tests, declined to comment without permission from Kiewit Corp., based in Omaha, Neb. Kiewit referred all questions to Caltrans.
Olson Engineering detected the problem concrete in Pile 3 in 2007, calling it "a batch of concrete that has not fully set at the time of testing" or "a very poor area of concrete." The company suggested new sonic tests. None was conducted, according to Caltrans.
"The most likely cause for the (19-foot) anomaly is concrete that didn't cure," or harden, said Les Chernauskas, general manager of Geosciences Testing and Research Inc., a Massachusetts firm that specializes in sonic testing. Chernauskas, who examined hundreds of pages of technical documents for The Bee, co-authored a seminal paper used to develop standards adopted by Olson and most similar companies.
Through a spokeswoman, Gov. Jerry Brown declined to answer questions about whether Caltrans has kept his office informed of ongoing concerns about the bridge, or if the agency enjoys his confidence for its construction and testing oversight. The Governor's Office deferred all questions to Caltrans.
Rather than provide engineers or executives for an interview, Caltrans spokeswoman Tamie McGowen responded in writing to Bee questions. "Substantial evidence," including tests of the small mock-ups and the other bridge piles, indicate that the abnormal concrete in Pile 3 eventually hardened properly, she wrote
"We are confident in the structural integrity of the main tower foundation and that the bridge will perform as designed to handle an extreme earthquake," McGowen said. A panel of engineering experts hired by the agency to re-evaluate the safety of the foundation concurred.
Chernauskas expressed skepticism about that conclusion. No one knows if the problem section ever hardened to its required strength, he said, because of the failure to retest.
"A job of this magnitude? Close to 20 feet of material near the top of the shaft showing anomalous readings?" said a university professor and expert in deep foundation testing, who also reviewed the documents. "At least it requires some discussion."
He spoke anonymously for fear of jeopardizing business relationships with contractors for Caltrans, among the nation's largest public-works funders.
Kiewit built the Bay Bridge piles in 2006 and 2007 and had relatively little experience with this kind of deep foundation. It had struggled on prior jobs, according to Caltrans records obtained by The Bee.
Kiewit began building foundations for the Benicia-Martinez Bridge shortly before taking on the Bay Bridge project. Records show that Caltrans engineers recommended rejection of nearly all the Benicia piles built before awarding the Bay Bridge contract to the Kiewit joint venture. Those piles, almost one-third of the Benicia job, were deemed either too flawed for use without repair or required retesting due to construction errors. After Caltrans awarded Kiewit the Bay Bridge contract, engineers similarly recommended rejection for more than 80 percent of the remaining Benicia piles.
Soon after the Bay Bridge sonic testing, Kiewit-FCI-Manson provided Caltrans with results for six of the 13 piles. They showed only minor problems. It failed to deliver the other sonic reports, including one showing the huge anomaly. Caltrans did not request those tests, McGowen said, which technically were not required by the construction contract.
Because the company declined to comment, its motives for withholding test data remain private. Substantial repairs on a giant pile can cost up to $1 million.
Thomas W. Joo, a UC Davis law professor and contract authority, said that even if a builder has no contractual duty to report test data, it could face legal liability for "straight-up bad faith."
"If it's a matter of public safety, the calculus is different," he said. "The whole thing is colored by what's at stake."
Kiewit and its partners provided findings for six piles out of at least 12 tested.
When a contractor provides partial findings it can engender "a duty to disclose" more fully, Joo said. "Partial, misleading disclosure is a species of fraud."
Designers of the new Bay Bridge built in a factor of safety – more capacity than strictly needed. A single problem – even a large one – probably would not cause a catastrophic failure.
After The Bee investigation last November exposed fraud in the Caltrans foundation test unit and described possible design flaws in the piles, officials asked a panel of eminent engineers, described as independent, to review the tower foundation. In March, after deliberations from which the public was barred, they declared the bridge safe.
Soon after, The Bee reported financial and professional conflicts of interest for each panelist, as well as for Earth Mechanics Inc., which wrote a background report under Caltrans direction. Ethics experts said the conflicts damaged the panel's credibility.
Records show that the panel, Earth Mechanics and Caltrans officials misrepresented or dismissed crucial facts that differed with their conclusions. Among the most significant:
Panelists said that sonic tests "showed the expected good construction quality," ignoring the major anomaly found in Pile 3 and the absence of a test report for Pile 8.
Caltrans created what the panel called "full scale" model piles, ostensibly to check pile-design concerns raised by independent experts who reviewed documents for The Bee. Those experts said unduly congested rebar supports could have hampered the flow of concrete, leading to undetected defects. According to the panel report, the models proved that the rebar posed no problems.
Agency documents show that the models were miniature versions, not full scale. They were too dissimilar to provide reliable comparisons, said Cumaraswamy Vipulanandan, an internationally known deep-foundation expert at the University of Houston. For example, the models were built in above-ground steel casings, not drilled into bedrock as was half of each bridge pile. This could have eased the flow of concrete through the rebar, Vipulanandan said.
The other university professor who spoke with The Bee, also respected globally for his deep-foundation work, compared data from strength tests for concrete samples from the Bay Bridge piles and the models. The model concrete was far weaker – suggesting that excess water might have been added to increase flow through the rebar. After evaluating other technical qualities of the concrete mix, he called the models "a poor representation of the actual conditions in the (bridge) piles."
Caltrans said the original sonic report for Pile 3 was not needed, nor was a repeat test warranted, because subsequent examinations that used radiation instead of acoustic waves showed that pile as sound. Yet according to Olson Engineering and the manufacturer of the radiation test meter, radiation can't detect unset concrete.
Olson had been hired to examine every pile. Asked why no results for Pile 8 were released, officials offered incongruous explanations. The builder said that it was "currently unaware" of sonic tests on that pile, according to a memo released by Caltrans. Agency spokeswoman McGowen said the tests were skipped because Caltrans ordered sonic tests only when radiation tests showed possible problems. However, in nearly all other cases, sonic tests preceded, rather than followed, radiation tests.
The backgrounder prepared for the expert panel described all the radiation tests, called gamma-gamma logging, as valid. It cited reviews by the Federal Highway Administration and Caltrans.
Those evaluations were neither complete nor vetted by independent experts. Caltrans formed a "Gamma Data Review Team" after The Bee's November report. That group has launched a more rigorous assessment of all radiation tests conducted from 1998 through 2011, to "resolve any remaining questions regarding the integrity of our test data," according to a Caltrans statement. Outside experts will review the team's methods, the agency said. That work won't be completed for several months.
Experts said the combination of missing and undisclosed data, radiation's inability to detect unset concrete, doubts about the piles' design, uncertain testing reliability, and reliance on model piles creates uncertainties about how the foundation would behave in a worst-case scenario.
"The way that they conducted the testing program, and the results, do raise some serious issues with respect to the quality of the concrete," said the university professor who examined documents at The Bee's request. A chief concern, he said, involves the absence of sonic data for Pile 8 and the location of problem concrete in Pile 3, toward the top of the pile, "subject to the most significant loads during an earthquake."
"If you had (two) out of 13 piles with major zones of defective concrete, at that point I expect you'd start to see some problems," the professor said. "(It) could result in a very large movement of that tower in an earthquake. How that would affect the performance of the bridge structurally is a big question mark. No matter what anyone tells you, no one can answer that question without doing some very rigorous analysis."
If the bridge foundation needs remediation, the state faces a difficult problem. Replacing or fixing the piles "in any significant way is pretty much impossible" in an economically viable way, Bernard Hertlein, a principal scientist at the construction giant Aecom Technology Corp., told The Bee last year.
Multiple scenarios assuming serious problems in the two suspect piles should be explored, experts said, and if needed, those piles should be physically examined.
Using computer modeling, "you simulate the potential defects and see whether that would have any impact on the overall system," said the University of Houston's Vipulanandan, as a way to test "how much the factor of safety in the design would be reduced."
Getting to structures beneath a mammoth 525-foot tower might require drilling through the bridge footing to extract core samples – a significant engineering challenge. But Chernauskas, Vipulanandan and the other professor who reviewed the data said such an effort might be vital to determine whether the bridge could stand up to the most extreme earthquake.
Caltrans spokeswoman McGowen cited "redundant internal and external review and testing," involving numerous engineers, as proof of the structure's safety.
"The test results show the concrete in these piles is acceptable," she said. "Therefore additional testing is unnecessary."
The Pile 3 anomaly presents a perplexing mystery, experts said.
Caltrans and its consultants offered this explanation: The pile received sonic tests too soon, after only four days. It would have hardened in five days.
Concrete specialists pointed to an essential flaw in that reasoning. An entire pile cast from a continuous concrete pour – as with Pile 3 – generally hardens at about the same time, said Gary Knight of HeidelbergCement Group in Doraville, Ga. He serves as a committee chairman with the American Concrete Institute, a nonprofit technical and educational organization.
Most of the pile hardened in four days or less; only the problem section was delayed. No one knows why.
"The first thing you look at is, did they overdose an admixture for those loads?" Knight said, referring to additives within the concrete used for the unset portion of the pile. Additives can enhance flow and reduce the need for water, which can weaken a pile. If overdosed, they also can cause disastrous results.
The additives were mixed with the other concrete ingredients at a central plant. Workers normally monitor the process to ensure consistent dosing across all 26 loads poured over about three hours into a single pile shaft – so that the top, middle and bottom all harden properly.
"If you severely overdose," Knight said, "you may harm the concrete in such a way that it would never gain the adequate design strength."
While Bay Bridge foundation contractors apparently never considered an admixture irregularity, Caltrans previously had encountered serious admixture problems .
The Ruckman Undercrossing on San Francisco's recently renovated approach to the Golden Gate Bridge offers a cautionary example.
Soon after Ruckman was built in 2010 and 2011, it began to fall apart, according to a Caltrans report. Workers removed much of its concrete by hand – it was that soft. The Bee learned about the problem from a Caltrans employee who requested anonymity for fear of retaliation.
The Bay Bridge piles and the Ruckman structure each contained the admixture WRDA 64. The soft concrete on Ruckman was traced to an overdose. W.R. Grace & Co., the admixture's maker, declined to comment.
The Ruckman failure was "a unique event," said McGowen. "WRDA 64 is one of the most common admixtures used nationwide, and it has been successfully used for over 40 years by Caltrans."
Daren Davis, a quality control manager with Utah-based Westroc, one of the nation's largest concrete manufacturers, said the additive's potential to produce soft concrete is well known in the industry. That makes it "critical" to use the specified amount, he said. Other experts said admixture overdosing is common and requires vigilance to detect and correct.
The Ruckman and Bay Bridge projects shared another trait: Computer-controlled machines that allotted concrete ingredients showed correct amounts – even when they sometimes went haywire.
For one Bay Bridge pile, a worker appended a concrete certification ticket with this emphatic handwritten note: "Batch (weights) are wrong. Computer error!!!" The ticket did not indicate why builders poured that load into a shaft anyway. It led to flaws that required repair.
After a pile is poured, laborers chip away loose or soft concrete that often forms on top, to prepare the surface for the next stage of construction. Caltrans' hired researchers said that chipping Pile 3 would have revealed any poor concrete.
But sometimes a top crust forms, hiding soft concrete farther down. Emails from Caltrans' Foundation Testing Branch, obtained by The Bee, described that phenomenon after 2004 sonic tests on a bridge pile near Atascadero. The culprit, according to the messages: an overdose with a similar W.R. Grace admixture.
Caltrans chipping records also pose doubts about Pile 8.
On the other piles, chipping lasted two to four days – considered normal. McGowen said that Pile 8 chipping took three days. Caltrans building summaries and inspector logs contradicted her claim, showing that an extraordinary amount of bad concrete took six days to remove. The records don't explain why.
The Caltrans emails describe a comparable mystery on another job – a pile supporting the San Francisco Fifth Street freeway offramp near the Bay Bridge. In that case, several feet of soft concrete had to be chipped away and replaced. Like Pile 8, radiation tests showed no problem.
Compounding such doubts about Piles 3 and 8, experts said, were uncertainties about the reliability of work by Olson Engineering. A review of the company's sonic test reports contained more than 20 errors.
Some confused which piles were examined. A required test was missing. The reports routinely misstated the interval between a concrete pour and a sonic test, and recorded incorrect dates and pile locations. Measurements conflated meters and feet, creating ambiguities about where to find pockets of irregular concrete.
Except in one case of mislabeling, documents show no sign that anyone at Caltrans noticed problems with Olson's work. When asked about the errors, McGowen called them insignificant.
Occasional small errors in such reports should cause no alarm, said Fari Barzegar, a construction and engineering expert who operates Oakland-based Habitat Engineering & Forensics. But so many such errors reflect poorly on Olson's testing and Caltrans oversight, and on the reliability of both kinds of conclusions: that one pile had a serious anomaly, while others had either minor problems or were flawless.
"It indicates that things aren't being done according to the highest standards," Barzegar said, for "a life-safety issue."
The errors left other experts uneasy.
"This is a little scary," said Vipulanandan. "You have to ask yourself if it impacts the quality of the piles."
Editor's note: This story was changed June 1 to clarify the context of comments from Thomas W. Joo.