Engineering CCTV by Arup* Proposed headquarters for CCTV in Beijing, designed by Rem Koolhaas, engineered byArup. Public space and circulation.
1 To support the rapid expansion of Because the seismic design of theCCTV building lay outside the scope of the prescriptive
Chinese codes of practice, Arup proposed a performance-based design approach from
the outset, adopting first principles and state-of-the-art methods and
guidelines to achieve set performance targets at different levels of seismic
event. Explicit and quantitative design checks using appropriate linear and
nonlinear seismic analysis were made to verify the performance for all three
levels of design earthquake. The basic qualitative performance objectives were: · no structural damage
when subjected to a level 1 earthquake with an average return period of 50
years. · repairable structural
damage when subjected to a level 2 earthquake with an average return period of
475 years. · severe structural damage permitted but
collapse prevented when subjected to a level 3 earthquake with an average
return period of 2500 years. Studio and broadcast spaces within the CCTV headquarters, Beijing, designed by Rem Koolhaas, engineered byArup. Staff and VIP facilities. Part of the tube structure: regular grid of columns and edge beams. Part of the tube structure: patterned diagonal bracing. The CCTV's tube structure is a composite of a regular grid of columns and edge beams plus patterned diagonal bracing. Internal columns starting from pilecap level. Internal columns supported on transfer structures. The CCTV foundation system.
2 For
the CCTV development site, the peak horizontal ground acceleration values
associated with the three levels of design earthquake are 7 , 20, and 40
percent of gravity respectively. Elastic
Superstructure Design With
the structural bracing pattern determined from the initial concept work, a full
set of linear elastic verification analyses were performed, covering all
loading combinations including level 1 seismic loading, for which modal
response spectrum analyses were used. All
individual elements were extensively checked and the building's global
performance verified. Selected elements were also initially assessed under a
level 2 earthquake by elastic analysis, thus ensuring that key elements such as
columns remained elastic. The
elastic analysis and design was principally performed using SAP2000(limited nonlinear
structural analysis and design with static, dynamic, and push-over capability)
and a custom-written postprocessor for the Chinese steelwork code, which
automatically combined the individual load cases applied to the building for
the limit-state design. Capacity
ratios were then visually displayed, allowing detailed inspection of the
critical cases for each member. Due to the vast number of elements in the model
? 10,060 elements representing nearly 300,000 feet (90,000 meters)
of steel and steel-reinforced concrete (SRC) sections ? and the
multitude of load cases, four postprocessors were run in parallel, one for
steel columns, one for SRC columns, one for braces, and another for the edge beams
that together form the continuous tube. The
SRC columns used a modified postprocessor to account for the differences
between the steel and SRC codes; section properties of these columns were
determined using Xtract(nonlinear
large strain composite cross-section analysis), which also computed the
properties for the subsequent nonlinear analyses. The
postprocessor provided a revised element list which was imported back into SAP2000, and the analysis and
postprocessing repeated until all the design criteria were met. As the
structure is highly indeterminate and the load paths are heavily influenced by
stiffness, each small change in element property moves load around locally. Optimizing
the elements only for capacity would result in the entire load gradually being
attracted to the inside corner columns, making them prohibitively large, so
careful control had to be made of when an element's section size could be
reduced and when there was a minimum size required to maintain the stiffness of
the tube at the back face. To
further validate the multidirectional modal response spectrum analyses, level 1
time-history checks were also made using real and artificially generated
seismic records. Design
and Performance Verification For
the performance-based design, a set of project-specific "design
rules" were proposed by the design team and reviewed and approved by the
Chinese Ministry of Construction's expert review panel, creating a "road
map" to achieve the stated seismic performance objectives. Appropriate
linear and nonlinear seismic response simulation methods were selected to
verify the performance of the building under all three levels of design
earthquake. Seismic force and deformation demands were compared with the
acceptance limits established earlier to rigorously demonstrate that all three
qualitative performance objectives were achieved. Inelastic
deformation acceptance limits for the key structural brace members in the
continuous tube were determined by nonlinear numerical simulation of the
postbuckling behavior. LS-DYNA (software for nonlinear explicit time
history analysis), commonly used to simulate car crash behavior, was used for
this work. The
braces are critical to both the lateral and the gravity systems of the building
and are also the primary sources of ductility and seismic energy dissipation.
Nonlinear numerical simulation of the braces was needed to establish the
postbuckling axial force/ axial deformation degradation relationship to be used
in the global 3D nonlinear simulation model. This
simulation was also used to determine the inelastic deformation (axial
shortening) acceptance limit in relation to the stated performance criteria.
Postbuckling inelastic degradation relationship curves illustrate the strength
degradation as the axial shortening increases under cyclic axial displacement
time history loading. Models illustrating the development of the CCTV headquarters' facade pattern, designed by Rem Koolhaas, engineered byArup. Brace stresses for a uniform grid. Unfolded view of the structure showing areas to densify or rarefy the mesh. Nonlinear finite element simulation model showing local buckling of a typical steel brace. GSRaft model of the piled foundation. The von Mises stress distribution of a large connection plate under the most unfavorable loading combination.
3 The acceptable inelastic deformation was then determined
from the strength degradation "backbone" curve to ensure that there
was sufficient residual strength to support the gravity loads after a severe
earthquake event. Having established the inelastic global structure and
local member deformation acceptance limits, the next step was to carry out
nonlinear numerical seismic response simulation of the entire 3D building
subjected to level 2 and level 3 design earthquakes. Both the nonlinear static
pushover analysis method and the nonlinear dynamic time history analysis method
were used to determine the seismic deformation demands in terms of the maximum
inelastic inter-story drifts and the maximum inelastic member deformation. These deformation demands were compared against the
structure's deformation capacities story-by-story and member-by-member to
verify the seismic performance of the entire building. All global and local
seismic deformation demands were shown to be within their respective acceptance
limits, demonstrating that the building achieves the quantitative and hence
qualitative performance objectives when subjected to level 2 or level 3
earthquakes. Foundation Design The design of the foundations required that the applied
superstructure loads be redistributed across the pilecap (raft) so as to engage
enough piles to provide adequate strength and stiffness. To validate the load
spread to the pile group, a complex iterative analysis process was used
adopting a nonlinear soil model. The superstructure loads were applied to a discrete model
of the piled raft system. Several hundred directional load case combinations
were automated in a spreadsheet controlling GSRaft, iterative nonlinear soil-structure interaction analysis
software. This procedure iteratively changed the input data in
response to the analysis results to model the redistribution of load between
piles when their safe working load was reached. The analysis was then repeated
until the results converged and all piles were within the allowable capacities.
The envelope of these several-hundred analyses was then used to design the
reinforcement in the raft itself. Connections The force from the braces and edge beams must be
transferred through and into the column sections with minimal disruption to the
stresses already present in the column. The connection is formed by replacing
the flanges of the steel column with large "butterfly" plates, which
pass through the face of the column and then connect with the braces and the
edge beams. To simplify the detailing and construction of the concrete around
the steel section, no connection is made to the web of the column. The joints are required to behave with the braces, beams,
and columns as "strong joint/ weak component." The connections must
resist the maximum probable load delivered to them from the braces with minimal
yielding and a relatively low degree of stress concentration. High stress
concentrations could lead to brittle fracture at the welds under cyclic seismic
loading, a common cause of failure in connections observed after the 1994
Northridge earthquake in Two connections, representing the typical and the largest
cases, were modeled from the original AutoCAD drawings using MSC/NASTRAN, a heavy-duty finite element analysis package. The
models were analyzed, subjected to the full range of forces that can be
developed before the braces buckle or yield ? assuming the maximum probable material
properties ?
to evaluate the stress magnitude and degree of stress concentration in the
joints. The shape of the butterfly plate was then adapted by
smoothing out corners and notches until potential regions of yielding were
minimized and the degree of stress concentration reduced to levels typically
permitted in civil and mechanical engineering practice. CAD files of the
resulting geometry of the joints were exported from the finite element models
and used for further drawing production. The structural design of CCTV posed many other technical
challenges to the large international team which delivered the design through
global collaboration, transcending time zones, physical distance, cultures,
cost centers, and even the SARS outbreak. In the end, the team delivered a
complex design on time and won approval from the Chinese construction
ministry's expert panel. A longer version of this article first appeared in The Arup Journal, 2/2005, and is
excerpted here with permission. * The Arup author team: Chris Carroll,
Paul Cross, Xiaonian Duan, Craig Gibbons, Goman Ho, Michael Kwok, Richard
Lawson, Alexis Lee, Andrew Luong, Rory McGowan, and Chas Pope. source:http://www./2008/0827/tools_1-1.html Addition: Metal guru: Arup’s Chris Caroll on CCTV Headquarters, 25 July 2008 |
|