1,070-Foot Salesforce Tower Elevates Seismic Design
There are no outriggers or belt trusses. There are no buckling restrained braces or seismic braces of any kind. There are no dampers—tuned, sloshing or otherwise. And there are no perimeter megacolumns, diagrids or moment frames.
The structure of the tallest building in earthquake-prone San Francisco—the 1,070-ft Salesforce Tower—is plain and simple. Still, the 1.4-million-sq-ft supertower is the city’s only high-rise, without an obvious seismic system, designed to perform 25% better in a quake.
Absent the quake-resisting devices that take up rentable space, block views and disrupt construction tempos, the nearly complete project, formerly known as Transbay Tower, is its developer’s dream and its architect’s delight. “The marvel of the engineering is what’s not there, not what’s there,” says Edward Dionne, senior associate principal in the office of the designer, Pelli Clarke Pelli Architects, which is working with architect-of-record Kendall/Heaton Associates Inc. (KH).
Fred Clarke, PCPA’s senior design principal, adds, “It’s a quiet structure. That’s the magic.”
|The city considers Salesforce Tower a marker for, and the front door to, its multimodal transit center: a 1,425-ft-long groundscraper with a curvy facade that is expected to open for bus operations in March.|
To get to a quiet structure, the structural engineer, Magnusson Klemencic Associates (MKA), used a sophisticated engineering method called performance-based seismic design (PBSD), instead of following the building code’s prescriptive provisions for tall buildings. The 61-story tower’s superstructure has a reinforced-concrete shear-wall core, which takes all lateral loads, and perimeter structural-steel columns, for gravity loads only. The floor systems connecting the perimeter to the core consist of structural steel beams and a composite concrete-on-metal deck.
“We were told we could have nothing that would encumber the leasable space and, ‘By the way, don’t make the [lateral system] bigger than the core,’ ” says Ron Klemencic, MKA’s chairman and CEO.
“It wasn’t all that hard,” he adds. “When you put a box around all the passenger and service elevators, the stairs, the toilet rooms and the mechanical rooms, it happens to hold up a building that is 1,070 ft.”
Klemencic says the “chances are good” that a quiet structure could be even 100 ft taller than Salesforce. “We aren’t at the extreme limits of what was allowed” in terms of sway, interstory drift and stresses, he adds.
The quiet structure and a single-run elevator system—with no transfer floor—were key to the speculative development making marketing and economic sense, says Paul Paradis, senior managing director for Hines. In 2007, Hines, with PCPA, won a design-development competition for the tower.
“Ron came up with a fabulous design,” says Paradis. There are only three columns on each 167-ft-long face of the obelisk-shaped building and none at the rounded-glass corners. That configuration allowed an “incredible” glass curtain wall with 10-ft-tall windows and 45 ft between columns, Paradis adds.
Ronald O. Hamburger, a senior principal with engineer Simpson Gumpertz & Heger (SGH) and a member of the structural peer-review panel for Salesforce Tower, calls the tower “an impressive structure and an impressive design.”
The superstructure design is just about the only “quiet” part of the speculative development, which was delayed—and nearly derailed—by the Great Recession that hit in late 2008.
Soon after, the project mostly went into hiding through 2011. Construction, which started in 2013, has not been easy. The hemmed-in site, in a district full of high-rise projects, is too close for comfort to the 1,425-ft x 171-ft “groundscraper,” recently renamed Salesforce Transit Center but formerly known as Transbay Transit Center (ENR 6/29/15 p. 40). Work on the two megaprojects, both developed by Hines and designed by PCPA, is overlapping.
“There was so much construction in the area, it was tough to stay out of each other’s hair,” says David M. Wilson, vice president of Clark Construction Group-California, the lead general contractor in a joint venture with Hathway Dinwiddie Construction Co. “We had traffic planning meetings every Monday” with all the nearby-project players and the city, says Wilson, who declines to provide the value of the joint venture’s contract.
The building team also had to deal with poor soil conditions. From the top down to weak bedrock, more than 250 ft below grade, there are layers of dirt fill with debris, sand and then clay.
“There is potential for liquefaction of the soils” in a quake, says Jeffrey Dunn, a principal of the project’s geotechnical engineer, Arup, which also is the geotechnical engineer for the multimodal transit center and 181 Fremont, a high-rise nearby. If soil liquefies, it loses capacity to support a structure, says Dunn.
A spotlight has been on soil and foundations since it became known that Salesforce Tower’s neighbor across the street—the 645-ft Millennium Tower—had sunk 16 in. and listed 2 in. since its completion in 2009.
Millennium Tower piles, about 68 ft long, go into the Colma clay layer. They are not socketed into bedrock. The Millennium Tower experience, coupled with Salesforce Tower’s greater size and its close proximity to the transit center, led MKA, with Arup’s input, to design deep caissons socketed into bedrock.
“We were aware of the problems with Millennium Tower and wanted to be sure to avoid anything like that performance,” says Dunn.
After value-engineering, the caissons became 42 load-bearing elements (LBEs) socketed 70 ft into rock. The 10.5-ft x 5-ft LBEs—a first for San Francisco—reach down as deep as 310 ft from grade. They are the city’s deepest foundations, says MKA’s Klemencic.
The LBE work did not go smoothly. Problems with voids and soft tops led to a fix that was a big contributor to a six-month construction delay, says Wilson.
The development is costing “about $1 billion,” says Bob Pester, executive vice president and regional manager for Boston Properties, Hines’ 95% equity partner and the lead developer.
Wilson says, with few exceptions, crews have been working 24/7. A temporary certificate of occupancy is expected this month. The first tenant move-in is scheduled for December.
Without the estimated $6-billion transit center and Caltrains Downtown Extension Program, there would be no tower. The obelisk serves as a marker on the skyline to locate the multimodal facility.
In 2006, a year after the San Francisco Board of Supervisors adopted the Transbay Redevelopment Plan, the Transbay Joint Powers Authority, created in 2001 to run the transit center project, launched a combined design-development competition for the transit center and the tower. In 2007, TJPA selected the PCPA-Hines scheme. Hines agreed to pay TJPA $350 million for the land, with $50 million designated for a 5.4-acre public park on the transit center’s roof.
PCPA’s scheme called for a 1,200-ft-tall tower, encompassing 1.8 million sq ft. In late 2008, when Hines and TJPA were close to a final agreement but before any documents had been signed, the Great Recession hit, the bottom fell out of the real estate market, and the project was virtually shelved.
“We knew immediately that the price was too high, the timing was too short, and either the design was too expensive or the building was too big,” says Hines’ Paradis.
A renegotiation followed, starting in 2010, which was “very painful,” says Paradis, because the TJPA wanted the original land deal to help pay for the transit center.
The development nearly collapsed. But Paradis hung in. “We could have easily walked away, but I fiercely believed in the project,” he says.
Sporadically during the recession, PCPA, Hines and the city continued negotiations related to approvals. “We thought the tower ought to be the tallest in downtown and that a roof height of 1,000 feet was appropriate for shadow impacts and urban-design considerations,” says Joshua Switzky, a program manager for the San Francisco Planning Dept.
In 2011, work started on the transit center, after TJPA secured $400 million from The American Recovery and Reinvestment Act of 2009. Bus operations are expected to begin in March, says Dennis Turchon, TJPA’s senior construction manager.
In November 2012, a month after Boston Properties came aboard, the city approved the 1.4-million-sq-ft building with 61 occupied floors. The tower would be 217 ft taller than the city’s then-tallest building: the 853-ft Transamerica Pyramid Center.
At 61 floors, however, the tower was still 150 ft short of the agreed-upon 1,070-ft height. To compensate, a latticework crown of structural steel was added. By allowing the sun’s rays to pass through it, the crown breaks up shadows. By comparison, in the 1,200-ft scheme, some of the crown’s lower floors would have been occupied, says Paradis.
TJPA on March 26, 2013, transferred the land to the developer at a reduced cost of $190 million, with no funds earmarked for the rooftop park. “The land was a big bargain,” says Switzky.
Schematic design began in 2013. But with no tenant prospects, the developer had a contingency plan to stop construction at grade if the market changed dramatically for the worse, says Hines’ Paradis.
To reach the 1,020-ft height of the occupied portion, PCPA designed 14-ft, 9-in. floor-to-floor heights, including a raised floor one foot above the structural slab. A raised floor is “somewhat unconventional” for a spec office tower because it adds cost, but it also allows underfloor air distribution, says Clark Bisel, currently senior mechanical consultant for Meyers + Engineers. Bisel designed Salesforce Tower before he left WSP USA, the project’s consulting engineer.
Underfloor air distribution minimizes fan energy and eliminates overhead ductwork. To take advantage of San Francisco’s mild climate, outside air—brought in at each floor through louvers in the curtain wall—is used for ventilation and free cooling much of the time.
Instead of mechanical floors, each level has two mechanical rooms in the core. That avoided big duct risers to distribute air from the mechanical floors but required openings—one for supply air and the other for return air—in the 3-ft-thick core walls.
To minimize the number of openings, which interrupt reinforcing steel, the team decided on a single opening. The upper part, for supply air, is just above the structural slab, and the lower part, for return air, is just below the slab. This detail required “very rigorous” design coordination meetings with WSP, Pelli, KH and MKA, says MKA’s Klemencic.
The supertower will rank as the second-tallest building west of Chicago, according to data from the Council on Tall Buildings and Urban Habitat. It is 30 ft shorter than the 1,100-ft-tall Wilshire Grand Center, a hotel-office building in Los Angeles that opened in June.
The Wilshire Grand, designed by engineer Thornton Tomasetti with engineer-of-record Brandow & Johnston, has a composite structure consisting of a reinforced-concrete shear-wall tube, supplemented at three levels by sets of transverse outriggers with buckling-restrained braces. It sits on a 17.6-ft-thick mat that bears on a siltstone formation.
Salesforce Tower and Wilshire Grand were designed at about the same time. Like Salesforce, Wilshire Grand is engineered using PBSD to a higher seismic safety standard.
PBSD uses sophisticated nonlinear seismic time-history computer modeling—practical only in recent years, thanks to advancements in computing capacity and user-friendly analysis programs—to examine building performance during multiple predicted seismic events. With PBSD, engineers are able to analyze more-complex building geometries and precisely allocate strength and stiffness to achieve an efficient design that meets performance objectives, says MKA.
MKA opened the floodgates for tall-building PBSD, initially resisted by building officials in California and especially in San Francisco (ENR 1/12/09 p. 48). Klemencic was the “first to try to push through designs of tall buildings using the performance-based technique,” says SGH’s Hamburger. Previously, PBSD was used for only seismic retrofits, he adds.
PBSD, which requires peer review, is a way to avoid the code’s prescriptive provisions. For buildings taller than 240 ft, the provisions mandate a dual frame to resist lateral loads. Dual systems typically have a shear-wall core and a heavy perimeter moment frame.
Architects don’t like dual systems because they limit design expression. Developers don’t like them because buildings with heavy perimeter frames and, consequently, smaller windows are less attractive for leasing.
In San Francisco alone, MKA has designed 19 PBSD high-rises. However, Salesforce was the first designed to a higher seismic-risk category—RC 3—a requirement because it will have more than 5,000 occupants. Before that, the designs were for RC 2, which is for buildings with populations fewer than 5,000.
Under RC 2, there is a 10% probability of collapse in the maximum considered earthquake, says Klemencic. Under RC 3, there is a 6% probability.
To meet RC 3, MKA made the building stronger and controlled deflection, decreasing the allowable drift. “The core walls, due to their size, were inherently robust,” says SGH’s Hamburger. That meant designing for the higher-occupancy category didn’t require too much extra cost or effort, he adds.
The deep foundations did require extra effort. The poor bearing soil and proximity to the transit center—the two have a common basement wall—led Arup, collaborating with MKA, to create a 3D nonlinear structure-soil-structure interaction analysis to confirm the satisfactory performance of the tower, in relation to the transit center, in a strong quake. Thornton Tomasetti, the transit center’s engineer-of-record, reviewed and approved the SSSI analysis, which Hamburger thinks is the most extensive ever done.
Foundation work began in late 2013. In its bid, foundation contractor Bencor Global Inc. had proposed the LBEs in part because a rectilinear bearing element offers more bending capacity, says Giancarlo Santarelli, Bencor’s president. Also, filling the hole with slurry eliminated the casing needed for a circular shaft. Bencor’s LBE price was 30% less than its bid for the caisson scheme and was to take half the time, says Santarelli.
After a thorough review by Clark, Arup, MKA and foundation consultant Clyde Baker, the LBE design was accepted by the peer-review panel and the city issued a permit, says Klemencic.
LBE work started from the existing grade, before crews dug the tower’s 60-ft-deep excavation. The plan was to build one permanent test pile, with two Osterberg load cells in place. But the work took longer than allowed, which reduced the required friction on the pile, says MKA. Bencor then built a second test pile. Friction values were equal to or greater than required.
The rest of the 42 LBEs followed, one after the other. First, crews built a 4-ft-deep guide wall to define the limits of the bearing element and guide the excavating tool. Then, they excavated the earth, replacing it with slurry. Crews used a clamshell for the first 100 ft and then switched to a hydromill, in part to keep the shaft vertical.
Once at the specified depth, crews placed the long rebar cage. Each cage held six cross-hole sonic logging (CSL) tubes. After the concrete cured, a transmitter and receiver were inserted into a tube and sound waves transmitted, as a way to measure concrete density.
After a cage was set, crews placed the concrete, from the bottom up, using the tremie-pipe method. Each LBE took three to five days, says Santarelli. “We worked 24/7” to meet a very tight schedule, he adds.
By the time MKA got the CSL test results for the first pile, crews were working on pile six, says Klemencic. The report showed anomalies—changes in density at specific locations in the LBE’s upper 20 to 30 ft.
“That suggests a soft spot—clay or rock in place of concrete,” says Klemencic. The hope was that the data was flawed. “CSL testing is an indication, but it is not necessarily conclusive,” Klemencic adds.
Bencor proceeded with the work, as instructed, says Santarelli. Concurrently, MKA investigated protocols for how to take care of any anomalies.
Soon, crews drilled a 4-in.-dia core in the LBE and sent a camera down. There was a void “the size of a basketball,” says Klemencic.
Additional test data showed consistent anomalies in the LBEs. And further scrutiny of CSL data suggested the LBEs had “soft” tops—1,000-psi concrete, instead of the specified 8,000 psi, adds Klemencic. Ultimately, MKA insisted on remediation for the tops of all 42 LBEs.
The LBEs had been built from grade, with an extra 60 ft at the very top that would be sacrificed during the excavation. Clark decided to excavate the site before the fix, cutting off the sacrificial 60 ft to get closer to the problem tops as much as 90 ft below grade. That meant deepening the excavation and the braced shoring walls to access the 12 LBEs along the perimeter. Clark built a cofferdam around the LBEs in the center of the site so it could isolate and recast the tops.
Klemencic declines to speculate about the cause of the problems. Santarelli blames the voids and the soft tops on the greater rebar congestion near the top, which “prevented the concrete,” installed from the bottom up, from flowing properly. During the fix, crews worked from the top down, which made it easier to deal with the rebar congestion, says Santarelli.
In any case, the fix seems to have worked. To date, the building has performed slightly better than the predicted 11⁄8 in. of settlement, which Klemencic says is from anticipated downward displacement due to axial shortening of the LBEs.
“We have a very extensive monitoring program,” he adds. “The dirt is not moving.
“I’ve said many times that the foundation for Salesforce Tower is probably the best foundation ever built in San Francisco because it’s the only one anyone dug up and looked at,” says Klemencic.
Remaining work went more smoothly. Clark’s Wilson said the biggest surprise was this year’s strong winds, which prompted tower-crane shutdowns. “We probably lost a month to a month and a half on the exterior wall—starting at floor 59 and above,” he says.
There was a fear that the below-grade common wall, between the tower and the transit center, would move, but that went fine, Wilson says.
The strategy for the superstructure was to lead with the core and follow a few floors behind with the structural steel. “I believe we were the first to build a leading concrete core in San Francisco,” says Wilson.
John Reitmeier, vice president of operations for the steel fabricator-erector, the Herrick Corp., says the biggest challenge of the leading-core strategy was the coordination of the core’s 1,538 steel embeds, installed by core concrete contractor Conco Cos., and the connections to the structural-steel floor beams.
The root of the issue is the different allowed tolerances for core work and steel work, says Reitmeier.
According to American Concrete Institute standards, concrete contractors have a 2-in. allowable vertical tolerance for the core. According to the American Institute of Steel Construction, steel erectors have only 1 in. of vertical tolerance for the steel columns.
At each floor, if the concrete wall leaned in or out too much or if embeds were recessed too far into the wall, Herrick had to make field adjustments in the beam-to-embed connection plates to adjust the tolerance. If not, the curtain wall may not have fit, says Reitmeier.
Herrick had to adjust many of the connections to the embeds. The fieldwork required adding longer, deeper plates and more bolt holes, he says.
Dan Fink, Conco’s project manager, declined to comment on the project, including the embeds.
Beyond the embed problem, the 10,956 tons of structural steel went in without incident. “It was big, dumb iron,” says Reitmeier. “Nothing exotic.”
Looking back nearly a decade, Hines’ Paradis says the tower’s development phase was tougher than the construction phase, with the exception of the LBE issues.
And the leasing of the tower has been far more successful than Paradis ever anticipated. “We began the project with no leasing and no leasing prospects,” he says. “Leasing has gone very well.”
In 2014, Salesforce, a cloud computing company, signed a lease for 700,000 sq ft and bought the naming rights. “It was the largest lease in San Francisco history,” says Paradis. Since then, Salesforce has taken an additional 100,000 sq ft, he says.
The building is roughly 90% leased. Negotiations are active with prospective tenants. Paradis says he is “very happy” he didn’t pull the plug during the “depressing” times of the recession.