How do gridshells and longspan roofs perform in earthquakes?

The 500km rupture of the 2011 M9 Great East Japan Earthquake resulted in extensive damage in over a half dozen prefectures from Tokyo to Iwate.  Several lessons can be drawn from the response of spatial structures, particularly long span roofs. While the global behavior was generally excellent, nonstructural element damage and local failure modes were widely observed. This is unfortunate, as such structures play a vital role in post-disaster recover as shelters (e.g. Shigeru Ban) and minor design changes could have prevented much of the damage. In the aftermath, the Architectural Institute of Japan [1] conducted a detailed reconnaissance of dozens of gymnasiums, sports stadia and halls and found several reoccurring damage patterns:


Miki Disaster Management Park Beans Dome, Sport Stadium and Emergency Staging Area in Hyogo Prefecture (photo credit penccil::Slowtechture)

Shear failure of baseplate anchors

Typical gymnasiums in Japan have a cylindrical double layer steel lattice roof supported on RC columns or walls. A popular connection is a steel baseplate with either post-fixed chemical anchors or studs cast in-situ. These were often found to be embedded in thick (100~150mm) unreinforced leveling mortar or with insufficient confining links, resulting in extensive spalling and shear failures. Such damage was observed both in truss structures due to the horizontal seismic demand and in arching structures that impose an additional thrust. Such damage can render the structure unsafe during the critical post-disaster months as further aftershocks might precipitate unseating and collapse.

© Takeuchi Lab, Tokyo Institute of Technology

Buckling and fracture of tension-only bracing

Slender angles or rods arranged in an X pattern are popular as wind bracing, but also find use as roof diaphragms or vertical braces for spatial structures in seismic zones.  Such “tension-only” systems are well known to perform poorly when subjected to inelastic demands as the negligible compressive stiffness and cumulative tensile yielding creates slack, leading to progressively increasing drift and cyclic impact loads [2]. Furthermore, poor detailing resulted in a number of bolted connections developing concentrated buckling mechanisms and fractures. Often such systems are justified on the basis of elastic response, but it should be kept in mind that seismic demands have greater uncertainty than ultimate live or wind loads and code-defined demands can and occasionally are exceeded, both in low and high seismic zones.  If not for the specific frequency content of this earthquake (yielded braces elongate the structural period to a lower energy portion of the response spectrum), a number of these structures may have collapsed.

Slack due to yielding of tension-only bracing © Takeuchi Lab, Tokyo Institute of Technology

Ceiling collapse

Japanese ceilings [3] differ slightly from the grid system used in the US, but both can suffer from various unexpected local weaknesses, particularly in the detailing at brace and edge connections. Ceilings and mechanical equipment are often delivered as proprietary products late in the design stage. As the design actions are often deceptively low (N, not kN!), structural engineers do not always subject the detailing and fitness-for-purpose of these components to the same level of scrutiny as would be afforded to the primary structural members. While this is not a unique problem to spatial structures, the issue is aggravated by the large spans, coupled vertical-horizontal response and high occupancy rates. It should be kept in mind that ceilings are generally quite heavy, thus posing a risk to human life, and are time consuming to clean up, potentially rendering the facility unusable in the critical post-disaster months.

© Takeuchi Lab, Tokyo Institute of Technology

In conclusion, the successful design of spatial structures in seismic zones requires more than a global buckling analysis. The performance will be judged not on whether the member sizes were optimized and fine-tuned, but on the quality of the connection detailing and whether the load path is properly capacity-designed. A holistic design considering the total structural, mechanical and architectural solution is essential, as a robust gridshell is useless if collapse of nonstructural components impairs the building function.

Large-scale ceiling collapses in public venues © Takeuchi Lab, Tokyo Institute of Technology

Author: Ben Sitler

Ben Sitler is currently a PhD Student focusing on earthquake engineering at Tokyo Tech. Ben  graduated from Princeton in 2010, having worked in the Form-Finding Lab on his Senior thesis. After graduating Ben worked for ARUP, contributing to projects in Singapore, New Zealand and Australia.

[1] AIJ (2011), Joint survey report of the Great East Japan Earthquake – Part 3 Shell – Space Structures. Architectural Institute of Japan. (in Japanese)

[2] Sato Y, Motoyui S, MacRae G, Dhakal R (2011), Ceiling Fragility of Japanese Ceiling Systems. Proceedings of 9th Pacific Conference on Earthquake Engineering, Auckland, New Zealand.

[3] Sabelli R, Roeder CW, Hajjar JF (2013), Seismic Design of Steel Special Concentrically Braced Frame Systems: A Guide for Practicing Engineers. NEHRP Seismic Design Technical Brief No. 8.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s