This is pretty unbelievable, but multiple sources report that San Francisco’s Millennium Tower, finished in 2009 and at 58 stories the tallest residential tower in the city, is experiencing potentially grave problems with settlement–the building as a whole has settled more than 10 inches farther into its site than originally calculated, and it’s done so unevenly, so much so that the tower now has a supposedly noticeable lean.
The problems with settlement were first reported a few weeks ago, but the national media (and several friends, family, and SCI-TECH alumni) have now caught wind of the problem and it’s making headlines for the potentially explosive political consequences, if in fact the owners received an overly generous assessment from the city. Inspectors were aware of the problems as early as August, 2009, according to some reports, but approved the building anyway.
Politics aside, any building that tall that leans that much is going to attract attention, especially mine. So what’s up? (Disclaimer–all of the following is pure speculation, researched lightly with whatever’s available on line. Take with a grain of salt).
The New York Times yesterday had a particularly misleading take on this, one that SCI-TECH alums should be able to parse out pretty easily:
Mark Garay, one of the lawyers for the apartment owners, says it is too early to pinpoint the precise causes for the building sinking, but that it had already begun significantly before work on the transport terminal started.
“What we do know is that the foundation of this building does not go into bedrock,” he said. “It’s all landfill. It used to be part of the bay.”
That sounds terrifying. After all, shallow foundations on landfill were major factors in the collapse of several much smaller building in the Marina District in the 1989 earthquake. As the earth shook, the soil under the water table liquified (exactly like shaking up a french press full of settled coffee grounds), setting the buildings above afloat.
Garay’s quote makes it sound like Millennium Tower is a big version of the same problem. The excavation for the tower goes 75 feet underground, but bedrock in SOMA is more than 200′ below ground level. The basement is tanked to prevent water infiltration, so there’s a displacement force, but nothing close to the weight of the tower above.
But Garay’s quote doesn’t address the actual system in use here. Millennium Tower’s designers relied on an old-school foundation technique, friction piles, to support the weight of the tower (there’s also a 12-story mid-rise and a three-story connector…more on those shortly). Friction piles work by surface resistance with surrounding fluid soil. Imagine driving a broomstick into beach sand–you can only go so far before there’s enough broomstick in contact with the sand to put up fearsome resistance. This has always been a standard technique for building in liquid soil, and it’s why coastal construction always comes with the dulcet tones of a pile driver. Chicago builders used these to support the dozens of grain elevators that used to line the River, and Dankmar Adler was on record as wondering why building engineers didn’t use friction piles when designing skyscrapers–a foreshadowing of his eventual development of caissons for architectural applications.
According to a case study by the Concrete Reinforcing Steel Institute, Millennium Tower’s core is supported by 950 square piles, each of them 14″ x 14″. The loads from the surrounding building are apparently transferred to these piles, in part, by a 10-foot deep pile cap (with, CRSI, notes a fair dose of “vertical shear reinforcing.” No kidding.) The number of piles, the case study notes, was “governed by gravity design” and not “overturning due to earthquake.”
So that’s the section, at least as I understand it (happy to be corrected by anyone who knows more). This looks like it could be a classic vertical vs. rotational equilibrium problem. If you remember your elementary beam statics, we’re worried about structure moving through space (usually the problem is down) and providing enough resistance to prevent its translational movement. But we’re also worried about things rotating (beams in particular, but bear with me). Rotational equilibrium means providing enough resisting leverage to keep elements from spinning around one point or another. Where that resistance goes is critical–the larger the lever arm, the better able a support is able to push back and resist an unbalanced load. While the piles under Millennium Tower are designed to resist the translational load of the tower against gravity, they’re not in the most efficient place to resist rotation. So the lean could have something to do with this.
The larger-than-expected settlement in general–something like 15″ vs. an expected 5″–may not be that big a problem. Buildings in poor soil settle more or less than expected all the time, and the Chicago experience shows that guessing how much a fluid soil will compress is at best an art more than a science. There’s discussion of adjacent construction at the Transbay Terminal that may have exacerbated the gradual movement of the building down through the soil.
It’s the uneven settlement that’s more worrying. The lean itself isn’t such a big deal–the Fisher Building in Chicago leans a good 6-8″ out over the street, and has done for 120 years now–but the small changes to the building’s structural geometry could have all sorts off serviceability issues, from doors that won’t fully close because their frames are racked to pens that roll across the floor when they’re dropped. If one segment of the building is settling more than another, that can also put unanticipated stress on structural connections.
What happens in an earthquake? That’s a bigger question. Friction pile foundations can have problems with horizontal shear in large seismic events as the momentum of the building above works out of phase with the movement of the liquid soil below, but assuming they’re designed for this force they present a pretty solid keel for a tall building that suddenly finds itself afloat, and the resistance of the 950 piles will theoretically prevent the building from moving or rotating laterally. No building on the planet is “earthquake proof,” but a slight lean doesn’t necessarily make the building any more vulnerable.
What’s to be done? All kinds of possibilities. Adding piles under whatever part is experiencing greater settlement is one. Soil remediation with concrete is another, but unlikely given the surrounding soil. It’s possible that doing nothing is the best option–as long as the settlement is slow it may be that the building is entirely habitable for generations before things get too far out of hand.
But my favorite solution is the one finally employed in 1907 at the Fisher Building. The owners bought the lot to the immediate north of the leaning tower and constructed a slightly taller, heaver structure there, rigidly connected to the existing building’s steel frame. The new structure is tiny, but it balances out the lean of the older structure and literally helps to keep it upright. Sort of like the designated driver (left, below) walking a drunk friend (right, below) home from a late night…