In 1968, a gas explosion on the 18th floor of a high-rise in London’s Ronan Point area blew out a load-bearing wall. All the rooms directly above the damaged apartment collapsed, which then caused the remaining floors below the blast to also collapse. The event shook the neighborhood, but it also shook the construction industry. Since that time, in the UK and the U.S. alike, government safety standards have been instituted to regulate the structure of steel multistory buildings so that they can withstand explosions.
"Establishing guidelines that early was both a blessing and a curse," says Professor Ted Krauthammer, director of the Center for Infrastructure Protection and Physical Security at the University of Florida (Gainesville). "It was a blessing because the new standards undoubtedly saved lives and protected property; and a curse, because we haven’t revised them significantly based on what we have learned about progressive collapse since then."
The safety standards served well for decades, in part because there wasn’t much new knowledge in the field of progressive collapse. Since 2000, however, there have been extraordinary strides in computer hardware and software that have made complex analyses and simulations possible and affordable. "Both experimental and numerical data assembled in the past decade have advanced our understanding of building robustness," Krauthammer says. "It’s time we did a thorough review of our safety standards for building blast."
According to Krauthammer, it is especially important to make finite-element analysis (FEA) a design requirement.
Guidelines vs. Nonlinear FEA
Current U.S. design guidelines for steel connections in structures subjected to blast loads are based on recommendations in Department of Defense Technical Manual (TM) 5-1300. But the guideline was originally written for single-story steel frames, which are not subjected to any significant dead loads. Similarly, the existing design procedures to prevent progressive collapse are based on GSA and/or DOD guidelines (GSA and DOD progressive collapse guidelines have been recently combined into a single document).
Real-world accidents, and the terrorist attacks of the last 15 years, have proven that some of the assumptions behind existing standards for multistory structures should be re-examined. For instance, engineers and analysts should include vertical loads in their assessment of building safety, even where a single load-bearing element has been damaged or removed.
Doctor Krauthammer has been re-assessing building blast guidelines by means of computer models. "We used FEA to simulate the effects of vertical loads on multistory structures," Krauthammer says. "First we modeled steel connection behavior, then we analyzed a 10-story building subjected to normal and abnormal loading conditions."
Analyzing Connection Behavior
Krauthammer and his team used Abaqus/Explicit software from SIMULIA, the Dassault Systèmes brand for realistic simulation, to assess the behavior of steel moment connections (joined beams and columns) under dead loads. They also applied loads equivalent to the shock and gas pressures from an appropriate explosive charge in the middle of a room.
When blast pressures were applied to floor and sidewalls, the predicted global rotations of the beams were close to the TM 5-1300 results for beams near frangible (breakable) walls. However, the beams near reflecting walls rotated much more than TM 5-1300 computation predicted, transferring greater impulse and energy to the beams and column. All local rotations for the different cases clearly exceeded the limit of 2 degrees specified in TM 5-1300.
These findings indicate that extensive damage in the connections comes from the blast radiating in three dimensions as well as the vertically applied pressure. Deformation data for beams and columns in the various analyses indicate that the beam cross sections twisted further due to dead loads.
"Clearly, it is valuable to investigate structural connections using high-resolution finite-element analysis," Krauthammer says. For example, a steel moment connection judged safe based on TM 5-1300 criteria failed in the simulations. This suggests that TM 5-1300 guidelines may need revision to reflect the findings of these analyses.
Simulating Building Collapse
Modeling the collapse of a 10-story building presented different challenges, Krauthammer points out. "In progressive collapse, local damage leads to large-scale structural failure – an intrinsically transient, nonlinear phenomenon that is hard to model, understand or design against without FEA."
The team used Abaqus to model 10-story, 3-D moment steel frames with rigid and semi-rigid connections for their sensitivity to material, buckling and connection failures of specific columns. Six initial failure cases with rigid and semi-rigid connections were used to analyze the frames for progressive collapse. The team used both ideal (rigid plus hinge) and semi-rigid connections for the progressive collapse analyses.
In the simulation with ideal connections, only the case where three columns were removed caused total, instant collapse of the building. In a case with semi-rigid connections, the building also collapsed, but differently. "We initiated failure at a single connection," Krauthammer says, "and it started a cascade of failures." As additional connections failed and columns buckled, the floors above the removed columns fell, causing columns to buckle in the 6th floor. This column buckling initiated a horizontal failure propagation of the columns on the 6th floor, and the whole floor failed. After that, the columns in the first floor buckled because the floors collapsed—leading to the total collapse of the building.
Even though the ideal and semi-rigid connection cases both caused total collapse for the two cases, nonlinear finite-element results showed very different qualitative behavior. The collapse of the semi-rigid connection case was caused by a cascade of local failures, such as connection failures and columns buckling. However, the collapse of the ideal connection case was caused by column buckling in the first floor. The analyses also showed that once failure propagation initiated (i.e. horizontal column buckling across a floor), it would not stop until it caused total, or almost total, collapse. Horizontal column buckling propagation appears to be the most critical factor to control to ensure building safety.
Modeling Safer Structures
The FEA analyses that Krauthammer and his team performed prove that it is possible to represent the true physical behavior of steel building structures under actual loading and failure conditions. The analyses results corresponded closely with the data obtained from recent structural failures and collapses. The simulations point the way towards making FEA a required part of guidelines for building design.
The main resistance to changing the standards comes from inertia and the additional cost of analysis, but Krauthammer thinks the results would justify the costs. "Finite element analysis software can predict the behavior of a structural design," Krauthammer says. "It also can guide civil engineers toward modifications that increase building safety and prevent collapse." Updating the standards for building blast to include nonlinear FEA results would ensure that the engineers would have a large body of reliable models and data libraries to draw on for future projects.