Bay Bridge

Do hidden cracks imperil Bay Bridge?

The welding code and construction contract for the suspension-span roadway of the new San Francisco-Oakland Bay Bridge each contain a universal rule: “No cracks.”

That rule applies to any new steel bridge and is particularly important for a “fracture-critical” bridge such as the new span, which opened last fall. Fracture-critical bridges can break, because some of their parts lack redundancy or backup. If a weld crack grows larger, causing such a part to fail, all or part of the roadway could collapse.

In 2008, the no-cracks rule put the California Department of Transportation in a worrisome bind. The Chinese firm hired to build the roadway routinely produced cracked welds that proved difficult to fix. Facing rising costs and increasing delays on a $6.5 billion bridge that was already years behind schedule and billions over budget, Caltrans sought advice about its options from a highly regarded expert in how metal fractures. He said some cracks can remain without compromising safety.

Caltrans then changed its contract and decided to put aside the welding code. Its fracture-critical bridge could now have cracks.

Weld cracks are the latest in the litany of errors and construction problems regarding the new span that have steadily emerged in recent years. Suspect foundation concrete, rusted tendons in the skyway that connects the suspension span to Oakland, broken anchor rods, and corrosion on the main cable are among the issues uncovered by The Sacramento Bee and others.

The California Highway Patrol, an independent expert panel formed by the Legislative Analyst’s Office, and the California Senate Transportation and Housing Committee are investigating how Caltrans handled the weld cracks and other quality-assurance matters. Committee chair​man Sen. Mark Desaulnier, D-Concord, said he awaits his committee’s and the LAO’s conclusions concerning the welds, other construction problems and the bridge’s safety. Those are, he said, “to be determined.”

Caltrans allowed cracks just below the pavement surface, where welds attach steel panels and stiffeners that comprise the span’s box girders – segments that form its roadway. According to Caltrans documents and testimony before DeSaulnier’s Senate committee, the top Caltrans official in China also allegedly instructed inspectors to disregard many obvious cracks. No one knows how many cracks were left in the bridge. The department later studied the welds further, (Caltrans Weld Report) certified that they were handled properly and declared the bridge safe.

The Bay Bridge is a “lifeline” structure. It’s meant to return to service within 24 hours after the largest anticipated earthquake in one of the nation’s most seismically unstable regions. Every day since the span opened last September, a multitude of cars and trucks have crossed.

The Bee examined more than 100,000 pages of construction records, research studies, meeting minutes and memos, and interviewed officials and outside experts, to understand the decision to allow cracks and learn whether it jeopardizes the safety of those travelers. The Bee found that Caltrans skipped important steps in an effort to move the project along quickly. Most notably, an expert review that concluded the bridge was safe and reliable did not perform a rigorous analysis of how cracked welds would perform during the powerful earthquake the bridge was conceived and built to withstand.

UC Berkeley engineering professor Abolhassan Astaneh-Asl is a well-known expert in the seismic performance of steel bridges. He advised Marwan Nader of TY Lin International, engineer of record for the design team of the new bridge, when Nader was a UC Berkeley student. Astaneh and his research associates have spent thousands of hours analyzing and modeling the new suspension span.

After that study, he concluded that the bridge would not survive the ultimate seismic test.

Caltrans doesn’t dispute that it altered its requirements to accept cracked welds on significant portions of the bridge. But officials reject Astaneh’s critical assessment and have drawn confident conclusions ahead of the state Senate and LAO panel deliberations.

“We determined that the new structure was safe and provided a level of protection to travelers on the bridge exponentially higher than that offered by the (old) bridge. …That is a decision I stand by,” said Caltrans Director Malcolm Dougherty, in a February letter to DeSaulnier.

The department points to the new span’s unique design as ample reason for optimism. A single cable – secured to the bridge itself – holds up the roadway, while special engineering features are designed to absorb seismic motions.

To Astaneh, that “self-anchored” design lies at the heart of the problem. “I don’t know any other major bridge located in a seismic area that is fracture-critical,” he said. “We cannot have cracks in the welds – period.”

The cracks, Astaneh said, present a looming threat to public safety.


The decision to leave cracks in place was not an easy one, according to Steve Heminger, executive director of the Metropolitan Transportation Commission and head of an oversight panel established by the state Legislature for the project. But it was the right choice, he said, given the urgency to replace the old, unsafe span before a major earthquake.

High-level Caltrans engineers or consultants who worked for the project in China disagreed. They have warned that the cracks could haunt the new span and the traveling public for years to come – requiring costly and disruptive maintenance.

The larger question is whether the cracks represent a hazard to the traveling public, experts said.

An important way to answer that question, Brian Maroney, Caltrans’ chief engineer for the bridge, said in a recent interview, is to examine history. He said Caltrans reviewed major quakes around the globe and never found a case in which weld cracks caused bridge-roadway fractures.

But such a case did occur. It happened close to home and was discovered by a Caltrans-funded study.

After the 1994 Northridge earthquake in Southern California, Caltrans commissioned Astaneh to lead a group of eminent experts for an in-depth investigation of damage to steel bridges. The group’s report described the Santa Clara River Bridge over Interstate 5 in Santa Clarita. The bridge suffered several fractures in steel girders. The breaks were traced to tiny cracks in welds, likely present before the quake, and worsened by vibrations of heavy trucks passing overhead. When the quake struck, the girders fractured.

The I-beam-shaped girders, like those in most steel bridges, fortunately were not fracture-critical. The Santa Clara River Bridge remained standing.

The Bay Bridge is different. Its roadway consists of box-girder segments welded together. In effect, they create one contiguous, fracture-critical girder, Astaneh said. If welds crack and grow rapidly during a large quake, the entire roadway could fail, he said.

The box girders were manufactured by a firm now called Shanghai Zhenhua Heavy Industry Co. Ltd., or ZPMC, from 2006 to 2011. ZPMC was new to bridge fabrication. Caltrans reports showed its work resulted in hundreds of cracked welds, including those allowed to remain unrepaired.

Those reports said that the firm’s welders commonly made errors in vulnerable areas, and sometimes defied corrective direction from Caltrans. On more than 16 occasions, ZPMC ignored key contract requirements for box-girder welds specifically cited as fracture-critical. Caltrans said the problems were resolved. It provided The Bee with documents for 11 of those cases. Six showed that the resolutions were to accept results of tests conducted by ZPMC. In one case, a document states a “ZPMC unqualified welder” was found working, and “when questioned, ran from the shop.”

Hundreds of similar problems were noted on “seismic performance critical welds” – a term Caltrans used for welds subject to the same requirements as fracture-critical welds.

Maroney approved the order allowing cracks but now says that in some cases, what they found were really benign imperfections, rather than cracks. “Some people like to go, ‘it’s a crack, it’s a crack, it’s a crack.’ And I get that. And that was one perspective that we could take.”

Maroney said he concluded that the “super conservative” contract, which forbid cracks, could be changed to use a “finer, project-specific set of rules” that allow cracks in certain circumstances. The result, he said, was greater efficiency and a safe bridge. Maroney later asked a team of experts, including two co-authors of the welding code, to study how Caltrans had managed welds on the roadway box girders. The resulting 304-page 2011 report, meant as a definitive examination of the issue, concluded that the welds would hold for 150 years – the span’s design life.

Seismic insignificance

The Caltrans report, coordinated by Maroney, contained incorrect assumptions about many of the welds, raising doubts about that prediction, Astaneh said.

Maroney had told his experts that the possible problems were restricted to a few “welds of interest” in critical locations and created with suspect methods or materials, according to the report.

By design, the bridge places most welds oriented east and west along the length of the roadway safely in “compression,” Maroney said. According to the 2011 report, compression presses the welds tightly in place, preventing them from enlarging and fracturing box girders under any circumstances. This automatically excluded them from the “welds of interest” category because they were seen as guaranteed never to cause problems.

Astaneh said his analysis of the suspension span, which included years of computer modeling, showed the opposite – that many of the welds Maroney said are securely compressed by forces of the bridge are actually in tension. That makes them much more susceptible to fatigue and stress that could, over time, cause cracks to enlarge into the roadway girder panels. Astaneh said that charts and images in Caltrans’ 2011 report about the welds confirm his analysis, showing tension in many welds that Maroney said were compressed.

Even for the few welds of interest, Caltrans used a narrow scope for the 2011 report. It calculated how the weight of the bridge and traffic, and the repeated movements caused by truck traffic or wind might fatigue and break welds over many years.

The report did illustrate how the concentrated pressure of truck tires – called “wheel load” – increases stress on the road deck. But it did not apply that analysis to how the pressure would affect cracked welds. The Federal Highway Administration cautions that wheel-load concentrations dramatically stress locations just below the roadway surface, where Caltrans allowed cracks at weld roots. The federal agency’s manual for building this type of box girder warns that “root cracking has the potential to result in the poorest fatigue resistance,” and that such cracks often expand “into the deck plate.”

More important, Astaneh said, the 2011 Caltrans report did not refer to how cracked welds would respond to earthquakes. Unlike the small, cyclic effects caused by traffic or wind, seismic motions can be sharp and dramatic.

In an interview, Bill Casey, a Caltrans supervising engineer and one of the project’s top leaders, said that if cracks exist, they could never fracture the roadway girders in a quake because the roadway’s weight prevents vertical movements. Maroney said he made no calculations or references to earthquakes in the 2011 report, because he saw them as insignificant for the welds, particularly considering the strength of the box girders.

“You only sharpen your pencil when you need to,” Maroney said.

Astaneh said that acceptance of cracked welds without calculating seismic factors defied common sense.

In a large quake, Astaneh said, “the entire length of the suspension span deck will bend between its east and west piers in horizontal and vertical directions, and will twist.” Such movements could be severe because steel box-girder bridges are relatively light compared to concrete bridges, Astaneh said. Some welds in compression under normal use would be placed in tension.

“Those cracks can jump into what we call the base metal,” according to Astaneh. That is, they could expand beyond the welds into the roadway panels, most of them about half an inch thick. “The loss of a single panel could cause the roadway to collapse,” he said.

The Golden Gate Bridge and most other suspension bridges use two cables, anchored into bedrock at either end. The new Bay Bridge uses a single cable that wraps around one end of the roadway and anchors to the other end of the bridge itself.

Maroney agreed that a large quake can deform the bridge in all directions, but said the span is designed to handle such forces with minimal damage.

A quake could not harm the welds because the brunt of ground motions would be absorbed by what he called structural “fuses,” he said. These work like a blown fuse in a home electrical system that prevents a fire by blocking excess current.

For example, the four legs that make up the tower connect via “shear links” – replaceable steel parts designed to contort, sparing the tower. The bridge also features relatively flexible piers that help support the bridge and are meant to absorb ground motions, shielding the roadway from damage.

The fuses don’t compensate for the lack of secure anchors in bedrock, Astaneh said.

Following Maroney’s analogy, Astaneh said if a home’s wiring can take 20 amps, a fuse that blows at 15 amps prevents a fire. Cracked welds, he said, are like corroded wiring – putting up less resistance to a quake, and risking fractures in the roadway girders before the fuses are activated. If the wiring is corroded and can operate safely at no more than 10 amps, he said, “your 15-amp fuse is not going to work, and your house will catch fire.”

Keith Devonport, a box-girder consultant who helped manage the project in China, said in a recent interview that he would be skeptical of the optimistic conclusions reached by Caltrans’ 2011 weld report unless the expert consultants had been told the number and characteristics of cracks allowed by changes to the contract. The report does not mention specific cracks or even the contract change that allowed them.

Asked if he would feel safe driving over the bridge, Devonport said, “It’s a question I don’t want to answer at the moment.”

Astaneh did answer the question. He said he won’t use the new bridge.

Related stories from Sacramento Bee