Millennium Tower Is Sinking; And Waiting Is The Hardest Part

San Francisco’s Millennium Tower is sinking. Since its completion in 2009, the 58-story, 645-foot tall residential building has settled 16 inches and tilted perhaps 2 inches to the northwest. Since the foundation issues came to light in August 2016, the vertiginous ultra-luxury highrise has become the subject of outrage, ridicule, and at least two pieces of pending litigation.

Nothing that we build is static. Our office towers, apartment complexes, and single family homes move in response to loads applied by the environment. Buildings sway in the wind, expand and contract in response to temperature changes, and shift with the land upon which they rest. In most scenarios, these deflections are so minuscule that the occupants never even notice. Millennium Tower happens to be a large enough project with a severe enough problem that the whole world can’t help but gawk.

Millenium Tower located in San Francisco's SOMA, near the Financial District
Millenium Tower located in San Francisco’s SOMA, near the Financial District.

In foundation design, not all terra is firma. While a one or two story wood-framed building can be built safely with a shallow foundation on crummy soil, a major skyscraper requires a foundation that can transfer extremely high loads into the earth. But the strata below our city streets can consist of anything from sand to clay to solid rock, and many cities, including San Francisco, have infilled former marshes and bays with soil in order to expand their coastlines and generate valuable real estate. Millennium Tower was built in South of Market, a neighborhood that mostly used to belong to San Francisco Bay.

Stick in the Mud

Many tall buildings in this area rest on foundation systems that extend all the way down to bedrock, on what are called end-bearing piles. But due to the geology of the Millennium Tower’s site, end-bearing piles would have reached down over 200 feet before hitting competent rock. Instead, the geotechnical engineers chose to support the tower on a series of 950 precast concrete friction piles driven between 60 and 90 feet into the soft clays below. A 10 foot thick concrete mat was then built across the tops of these piles to serve as a solid base for the tower’s superstructure.

Millenium Tower pile system. 14" square piles can be seen surrounded by the concrete pile cap [Image Source: CRSI]
Millennium Tower pile system. 14″ square piles can be seen surrounded by the concrete pile cap [Image Source: CRSI]
While an end-bearing pile transmits loads directly into bedrock, a friction pile transfers loads through friction into layers of softer soil. When a pile is driven or drilled into clay or sand, the soil presses against the newly introduced member. The friction at the pile-soil interface causes the pile to resist being pushed down further or being pulled out. If you’ve ever hammered a tent stake in place, then you’ve essentially driven a tiny friction pile: the stake isn’t sitting on bedrock, but it can resist vertical loads (in compression or tension) and lateral loads all the same.

However, this kind of foundation system relies on the stiffness of relatively weak clay, meaning that the building will most likely undergo some settlement. This process typically occurs in three broad stages: In the first stage of settlement, the soil deflects elastically, squishing as a result of the new compressive forces that have been applied to it. The second stage is called primary consolidation, wherein all the excess water in the clay is forced out, resulting in a decrease in the volume of the clay and a further settlement of the building. In the final stage, called secondary consolidation, soil particles within the clay layer reorganize themselves to adjust to the newly dry, highly compressed state. This leads to a further reduction in volume and even more settlement. Any current movement at Millennium Tower is probably due to primary or secondary consolidation.

Settlement Happens

Some degree of settlement is acceptable for a new building, and the geotechnical engineer typically provides an accurate calculation for anticipated movement well in advance of construction. However, Millennium Tower has already sunk over three times more than its total anticipated settlement. Also, since primary and secondary consolidation and can take years, it’s hard to know exactly when the tower will stop sinking.

You expected the Tower of Pisa but we give you the Tower of Suurhusen [image by Axel Heymann CC-BY-SA 3.0]
You expected the Tower of Pisa but we give you the Tower of Suurhusen [image by Axel Heymann CC-BY-SA 3.0]
The other issue is that of differential settlement, when one side of the building sinks more than the other, leading to the kind of tilt we’re now seeing in Millennium Tower. Differential settlement can occur when a building imposes uneven stresses onto the soil, or when the soil itself is weaker on one side. (The most famous leaning tower is the campanile of Pisa, but unsung examples like the Tower of Suurhusen and Oude Kerk in Delft are worthy of mention as well.) Differential settlement can lead to unanticipated structural stresses, damage to architectural finishes, leaks at connections to water and sewer lines, and problems with high speed elevators.

Who’s to Blame?

Millennium Tower is an unusually heavy building for this area of the city and for this type of foundation, and its weight problem is mostly due to the developer’s choice to build with concrete rather than with steel. (Concrete is often preferred for residential construction because the thin, flat floor slabs allow a developer to squeeze more rentable floors into a limited height.) Although steel has a higher unit weight than concrete (490lb/ft³ for steel, compared to 145 lb/ft³ for concrete), steel buildings tend to be lighter because much less of the material is necessary. As a point of comparison, the recently built 33-story steel office tower at 555 Mission Street applies a soil pressure of 2.4 ksf (2,400 pounds per square foot) whereas the 58-story Millennium Tower weighs in at a corpulent 11.4 ksf. The additional weight has probably caused the soft clays to settle more than for neighboring towers.

For its part, Millennium Partners blames the Transbay Joint Powers Authority (TJPA), the agency that has been excavating the Transbay Transit Center just feet from the tower’s footprint. Millennium claims that the soil dewatering operations for the new train station have accelerated the settlement of its skyscraper. The TJPA’s damning response states that the Millennium Tower was already sinking badly before the Transit Center began excavation, that no other nearby towers experienced settlement due to its construction, and that the tower’s extreme settlement is due only to its excessive weight and inadequate foundation.

How to Fix It?

Differential settlement can be a source of unanticipated structural stresses, but Millennium Tower does not seem to be in trouble at the moment. An engineering review has shown that the effect of the settlement is “negligible” and that the building would almost certainly be able to resist extreme earthquake loads. Journalists have reported a range of conflicting values for the true measure of the building’s tilt, but its out-of-plumbness doesn’t appear to be visible to the naked eye.

quote-skyscraper-as-counterweightThe clearest solution for the Millennium Tower’s foundation issues (without demolishing the building and starting over) would be to inject a cementitious grout into the weak clay beneath the foundation. This would strengthen the soil and halt the settlement, but the costs would mostly likely be prohibitive. Alternatively, the owners could stop the building from listing further by placing a massive counterweight on the opposite side of the tilt. When the Leaning Tower of Pisa’s tilt became worrisome, conservators placed a heap of lead weights on the high side to keep the tower in place during stabilization. Unfortunately, Millennium Tower is so large and heavy that the only effective counterweight would be the construction of a new skyscraper (with friction piles, of course) to the southeast.

The reasonable solution to Millennium Tower’s sinking and tilting is the least intrusive and most cost effective: Do nothing and keep an eye on it. Cracked stone should be replaced and elevators should be monitored for excessive wear, but a major structural rehabilitation does not yet seem necessary or worth the cost. As long as the tower is habitable and doesn’t show major signs of distress, the best thing to do is to watch and wait.

Alex Weinberg, P.E. is a structural engineer living and working in New York City. You can e-mail him at

87 thoughts on “Millennium Tower Is Sinking; And Waiting Is The Hardest Part

      1. Hey that’s not fair to those civil engineers who do their best to be civil. ;) Perhaps such civility prevented engineers from saying; Listen you knuckleheads, drilling the additional 120 feet could be the best wager, for the long term.

  1. It’s always interesting to see things like “cost/money” and how it influences (sometimes badly) decisions. And yes the world might be a better place if that influence wasn’t there, but here we are talking about a leaning building because someone didn’t want to go down to the bedrock.

    1. Excuse me? Even if capitalism were to disappear tomorrow projects and problems would be evaluated based on comparing the amount of effort needed to the predicted results. In fact we all do that constantly in our daily lives subconsciously. There really isn’t any other way to properly manage the limited time and resources we have at our disposal.

    2. >>because someone didn’t want to go down to the bedrock.

      It’s more than that.
      Not only did they not want to go down to bedrock, they were assured (and with good reason) that this method provided the same protection and stability that going down to bedrock would. Friction/floating piles are commonly used so it’s not unreasonable to think they’d work here.

      It seems like someone made a mistake here, either in the modeling, or skimping on construction materials, or maybe they didn’t get a representative sample of the material they were putting the piles into.

        1. it doesn’t “liquefy”, it’s more like the liquid trapped in the clay gets shaken and forced up more towards the surface making it wetter and more loose on the surface and denser further down ….

          but you missed the point where the weight of the building resting on the piles causes the clay to reform into a denser drier material over time around the piles. the clay outside the affected zone will likely be affected and you’ll have “liquification” on the surface around the building’s foundation but the building won’t be affected much by that process in a year or so and far less over a decade

  2. Another solution could be to build an identical tower north-west. Hence the 2 towers will tilt toward each other. When their top will touch they will stop tilting. Then the assembly could be renamed the triangle. :-)

  3. To the author: A well written and informative article, well done!

    To the editor: I have the feeling that this article is going to generate several postings about why it should or should not be considered appropriate for hackaday. I think it is appropriate and welcome for a few reasons:
    – It is about real world engineering (granted, not at the scale of the normal content of the site).
    – The discussion of the possible solutions is hack-related (when “building another skyscraper to hopefully balance this one” is on your list of solutions, you are dealing with hacks at an epic scale).
    – It is good to have offer diverse content.

    As long as the articles are of similar quality to this one, I would love to see more. While I do not want to see the site become a “technology new site”, discussing technology in terms of problems and solutions is a welcome addition.

    For those who don’t think the article is appropriate, just skip it.

    Perhaps such articles could be tagged somehow to make it easier for the readers to know that the article is technology-related so they could skip them without starting to read the content?

    Keep up the good work!

    1. Of course this article belongs here. Hearing about problems ignite thought and then lead to invention. If every article was about arduino HAD would be pretty dull and predicable.

  4. This article (mostly) omits the treacherous, dynamic nature of the Bay Area itself. If the 1988 earthquake demonstrated anything it’s that the ground itself works as a wonderfully destructive amplifier for seismic movement. It’s not clear whether the Millenium Tower has a damping mass or other dynamic compensation built in, but given the miscalculations on other parts of the building “keeping an eye on it” would seem the minimum effort.

    (Good read on tuned-mass damper systems in structures here [PDF]: )

    Video of dynamic action of tall buildings during an earthquake in Japan.

  5. Bad materials decisions based on maximizing profits has bitten the owners of the towers on their rich, corpulent behinds.

    Look they wanted to max out profits by cutting costs which meant in this case going with cheaper but far heavier cement rather than steel.,

    It’s the not the first time developers have pulled such a stunt. there are newer housing tracts around Southern California where the developers cut corners on firming up the ground where homes were to be built in order to save time and money. Lo and behold, within months of being completed, the new homes had cracks not only in their foundations, but running right up the walls to the ceiling. End result? Homeowners suing the pants off the developers.

    1. They probably didn’t even save that much. Sure it costs a lot more to drill 200′ than 60-90′, but when you consider how MANY friction pilings were used, it couldn’t have been that much cheaper.

    2. Is it my imagination or is this a stereotypical American response: “Homeowners suing the pants off …”

      Surely it is merely a claim against the developers’ public liability insurance policy? Or is there some emotional turmoil involed in cracks in the walls? Can some enterprising scoundrel fire a bullet through the crack exacerbating the occupants’ paranoia?

      1. I imagine the main complaint people have is that they won’t be able to sell it for half of what they paid, due to the reputation of the building. I would sue too, if I was set to lose half a million $.

  6. Ultimately, it is the city’s fault for approving the tower design, construction, and location. It’s the city’s responsibility to condemn the tower should it be necessary, or there is no way for the city to indemnify itself.

    It worries me that the tower might be or become dangerous and no one would admit it because of the financial ramifications or legal fallout. We might all be gawking at a deathtrap of not only the building itself but its neighbors, and the people near the building.

    Has anyone actually done a real study on the building to determine if any parts are sinking faster than others, that could cause stresses that the building was never design to withstand? Has anyone done a finite element model of the building to understand what is happening to it.

    Or is everyone just talking out of their ass?

    1. Having twenty+ years in a “County”, I can assure you, the engineers that approved the plans used the “Rubber Stamp” approach.
      Real engineers seem to be hired by competent engineering firms, and the dregs seem to wind up at city or county and use the “Design by Catalog” approach to further their career. Sad to see this repeated, as in “Deja Vu all over again.”

      1. Originally it was all based on water. 1 cubic centimeter of water = 1 milliliter = 1 gram, but only at 3.98 degrees C where water is at maximum density. As it’s cooled further, it starts to shift from amorphous liquid to crystalline solid, aligning the molecules and spreading them apart a bit.

          1. Uhm, no… 1 cubic metre is 1000 liters since 1 liter is 10 x 10 x 10 cm. 1 liter (water) weighs 1 kg and contains 1000ml with 1ml weighing 1 gram. 1 calorie is what you need to heat 1 gram of water by 1 degree centigrade. It is a very well thought out system.

    1. I would dig under the foundations and start extending pilons down to bedrock enough to halt the movement. You could actually start on the low side, then wait for the other side of the building to settle more and correct the tilt before adding the pilon extensions on those sides.

  7. A possible fix for the lean might be to freeze the ground under part of it which could slow the sinking on that side.
    This would be expensive and difficult but there are probably no cheap and easy fixes for a problem of this magnitude.

  8. you know i they got one crew started drilling and adding the extra material now to shore up one side then they would probably be done in 20 years it would take for them to make up there minds and end up saving money in the long run.

  9. To give some perspective, 1500 pounds per square foot is the maximum allowed (in my county) without a geotechnical report. Our county is comprised primarily of clay topsoil, not much really solid rock to be found. Their soil loads are 7.5 times the acceptable allowance for a generic clay soil. I can’t see how friction piles and a big cement block are going to address the loading issues associated with huge pressures. I suspect the reason they are encountering these problems is that nobody has done the research to determined what 11.4ksf does to this type of soil. It may be extruding the soil as it settles, a mode they are not familiar with? Or perhaps it’s the 200ft of wet clay below, the stresses are traveling further down into the soil than they anticipated?

  10. Would it be possible, instead of raising the lowest corner, to lower the rest of the tower to level it off? Tricky, sure, but removing soil rather than adding, and wouldn’t address any issues of pipes/electrical (although, as the article mentions, the building’s built to withstand an earthquake, so the utilities should be flexible enough to still be fine.)

  11. “If you’ve ever hammered a tent stake in place, then you’ve essentially driven a tiny friction pile”

    I don’t think I’ve ever driven in a stake that hasn’t hit a rock. You would think those would be end bearing piles, but they just pull right out EVERY FRIGGIN TIME.

  12. My degree is in civil (structures) and I used to say that structural eng is based on science and soils and foundations is based on voodoo. They used to teach us about the screw-ups and most of them were related to foundations. One amusing story from the foundations prof was about an idiot who completely screwed up the foundation on a house. The story ended with “You know who that idiot was?… me!”

    I left engineering after a few years to be a programmer. Instead of the occasional screw-up, I got to see perpetual screw-ups and wished programmers were actually engineers as so many claim to be.

  13. They placed a SKYSCRAPER on 60 to 90 foot friction piles? WTF?

    Here in southern Louisiana there is no bedrock and there is no such thing as an end bearing pile. The friction piles under the Lake Pontchartrain Causeway, which is a freaking automobile bridge not a skyscraper, go 70 ft into the silt. The piles under our major bridges across the Mississippi River go over 300 feet down. The piles beneath One Shell Square, at 50 stories about the same height as Millennium Tower, are over 200 feet deep. In SF this would apparently be deep enough to make them end bearing. It’s not like we don’t know how to build stable structures on such questionable fill, it’s just that somebody got a bad case of the cheaps. And as for earthquakes, which we don’t have in Louisiana, did these morons never hear of liquefaction?

  14. Is there room around the outside to dig then punch piles down to bedrock? Do one at a time, then as each pile hits rock, attach a ‘shelf bracket’ extending beneath the base slab. Start on the low side, putting the brackets up against the slab.
    Then shift to the high side, spacing the brackets down the right amount so the building will settle level onto them.

    When that’s complete, excavate in sections to cast a concrete skirt wall around it to brace between the piles.

  15. As per other architectural clusterfucks, this tower won several industry awards for its design & construction. Wonder if they ever have to give them back?

    London has the Walkie Talkie, which was award-winning and (as everyone except the architect expected) focused the sun & melted things in the street.

    Local award winner (now demolished) was the Tricorn Centre in Portsmouth. Very striking from a distance, utterly useless and horrible to actually be in. Was a place businesses went to die. Only high point was being used as a set for Dr Who.

  16. There might be lessons to be learned from Seattle.

    Back around 2010, the 25 story McGuire Building was deemed not structurally sound and needed to be demolished. It was completed in 2001. It is a facinating story and I’d like to learn more about how this went so badly, and what came out in the various civil actions.

    “Defects include corroding and rusting cables, defective reinforcements in the building’s exterior concrete and structural problems, the company said in a news release.”

    “The city of Seattle first learned of the building’s flaws and demolition plan a week and a half ago, when the owners showed city officials a March 2010 engineering report detailing the building’s structural problems, said Alan Justad, deputy director for the city’s Department of Planning and Development. Until then, as far as the city knew, the owners were repairing the building, he said.”

    The demolition was complicated, and performed one floor at a time. Lots of dust and noise for neighbors. It is very fortunate that inspectors caught this, and weren’t bought off. Imagine the magnitude of this screw up, and the amount of money involved.

  17. How about the Sandra Day O’connor courthouse in Phoenix, AZ? It gets HOT in Phoenix, so you’d think they’d construct the building with air conditioning. Nope! It also has a lobby with glass walls. It’s supposed to be passively cooled.

    Science to idiot architects – when the air temperature outside is over 100F, you can’t use it to *cool* anything.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.