Structural Simulations for Scenario Earthquakes in the Los Angeles Basin (Heaton)
Focus: End-to-end simulations of steel moment-frame buildings for Puente Hills blind thrust earthquake scenarios.
The overall goal is to assess the potential performance, including collapse, of high-rise flexible buildings to the simulated ground motions from hypothesized large earthquakes in the metropolitan Los Angeles region. The responses of 20- and 6-story steel frame buildings have been simulated for each site using the seismic nonlinear structural simulation program, Frame 2-d, developed by Hall (1997). It is based on a planar-frame fiber-element model that includes both material nonlinearities as well as geometric nonlinearities. Most importantly, Frame 2-d includes the nonlinear effects of torques on load carrying columns that result from the finite horizontal displacements of the building
(often referred to as the P-delta effect). The inclusion of P-delta nonlinear effects provides a collapse mechanism. That is, the building collapses when the columns are no longer able to resist the torques that result from the weight of the building that are applied to tilted columns.
Frame 2-d also has the capability to include different criteria to model weld fracture.
Prior to the 1994 Northridge earthquake, it was commonly assumed that welded connections were significantly stronger than the plastic yield strength of structural steel and that steel frames would only yield because of plastic deformation of the steel.
However, numerous welded connections were observed to fracture for SRMF’s in the near-source area of the Northridge earthquake. In the United States, emergency code changes were implemented in 1995 to correct this problem. However, the vast majority of existing buildings must be assumed to have brittle welds which would dramatically decrease the global yield strength of the buildings.
The buildings were designed to the 1994 Uniform Building Code (UBC94) at seismic zone 4 and soil site S2 (corresponding to soil type C in UBC97). Figure 1 gives the frame elevation and floor plan for the 20-story building. These buildings are symmetric which allows us to model their response using two-dimensional models. Figure 2 shows a pushover analysis of this building that demonstrates the overall yielding characteristics of the building.

Fig. 1 The left figure is frame elevation of U20. The right top figure is half of its floor plan. The right bottom figure graphically shows the definitions of response parameters used in this study.

Fig. 2 Pushover analysis of the 20-story steel frame buildings designed to the
UBC94 code. Horizontal loads are applied dynamically and the roof displacement is calculated as a function of the loads which can be carried by structures (given as a percentage of the building weight). This analysis can measure the actual strength of buildings. The blue line correspond to buildings with perfect welds, whereas the other lines correspond to buildings with welds that fracture randomly when local strains exceed a fracture criterion that is compatible with the observations of welds in the 1994 Northridge earthquake.
The presence of brittle welds decreases the global yield strength of the buildings. In the case of the 20-story building, brittle welds significantly reduce the ductility of the building (i.e., the ultimate deformation that is capable of achieving because of the importance of P-delta effects for tall buildings)
Simulated earthquake ground motions are from several sources, with the majority coming from a collaboration of several SCEC research teams to develop reproducible long-period ground-motion simulations for large earthquakes in the Los Angeles metropolitan area
(Day, S.,J. Bielak, D. Dreger, S. Larson., R. Graves, K. Olsen, 2005, 3-d Ground Motion
Simulation in Basins, Final Report to the Pacific Earthquake Engineering Research
Center, Lifelines Task 1A03). These simulations are for finite ruptures in a 3-d velocity structure that is appropriate for the Los Angeles metropolitan area (excluding very low velocities often encountered in the top several hundred meters) and the ground motions have been low-pass filtered at 2 seconds. Ten different rupture scenarios have been developed for known faults in the immediate vicinity of Los Angeles (Figure 3) and six different realizations have been prepared for each scenario (different hypocenters and

randomization of slip). Ground motions have been simulated on a 2 km square grid of
1,600 total sites.
Figure 3. Several SCEC research teams collaborated to generate long period ground motions from simulated earthquakes on ten faults in the Los Angeles area.
In the region defined by the green box, the SCEC researchers archived threecomponent ground motions on a 2km-by-2km grid, for a total of 1600 sites per simulated earthquake. At present, we have simulated building responses for one slip distribution-hypocenter combination on each of the ten faults. We provide the peak ground velocities and peak building responses for three of the ten faults:
Puente Hills, Los Angeles Segment (6); Puente Hills, all segments (7); and
Newport-Inglewood (9).
A total of 64,000 nonlinear building simulations have been performed to analyze the response of both 6- and 20-story buildings with the assumptions of both brittle and perfect welds. Figure 4 shows a contour map of peak ground velocities for a simulated M
7.1 earthquake on all three recognized segments of the Puente Hills fault. This particular

simulation turns out to generate the most severe ground motion of the 10 different scenarios that have been studied to date.
Figure 4. This simulated earthquake on the Puente Hills Fault ruptures across all three segments with Mw 7.1. The maximum dynamic displacement is 2.1m, and the maximum velocity is 3.1m/s. The Puente Hills Fault dips to the north, and the largest ground motions are observed up-dip of the fault, as expected.
The simulated response (maximum inter-story drifts) of 20-story buildings with perfect welds are shown in Figure 5. Strong nonlinear deformation (i.e. structural damage) is implied for any drifts exceeding 0.01 (any colors other than blue). Any regions shown in red (approximately 270 km
2
) have peak inter-story drifts that exceed 0.07 and most of these simulations indicate collapse. While regions shown in yellow do not indicate collapse, the drifts are so large that the buildings would almost certainly be damaged beyond repair. Although it is believed that buildings constructed since 1995 have welds that will not fracture in future earthquakes, the majority of existing buildings in the Los
Angeles region were constructed prior to that time and it must be assumed that their welds are brittle. Figure 5 shows the maximum inter-story drifts of 20-story buildings with brittle welds. In this case the region of collapsed buildings grows to more than

1,000 km
2
, nearly a four fold increase when compared with the buildings with perfect welds.
Figure 5. This earthquake on all segments of the Puente Hills Fault affects model twenty-story buildings with good welds in all of the Los Angeles Basin and San
Gabriel Valley. Sixty-seven buildings collapse in the simulation. The collapses and severe deformations are in buildings located in the north and east of the Los
Angeles Basin.

Figure 6. This rupture on the Puente Hills Fault affects twenty-story buildings with brittle welds located in all of the Los Angeles Basin and San Gabriel Valley.
Buildings in 262 locations experience simulated collapse, which is 3.9 times the number for buildings with good welds in the same earthquake. For the ten simulated earthquakes that cause collapses, buildings with good welds are much less likely to collapse than buildings with brittle welds.
Building response simulations were repeated for each of the 10 scenario fault rupture segments and the results were aggregated into composite maps that are shown in Figures
6 and 7 for 20-story buildings with perfect and brittle welds, respectively. It is clear that known active faults are located such that most sites within the Los Angeles basin could be shaken by ground motions that could cause severe deformation and collapse of 20story steel frame buildings. This is especially true of steel frame buildings with brittle welds, which are about four times as likely to collapse as buildings with perfect welds.
In addition to simulating the response of 20-story steel frames, 6-story steel frames have also been studied using this suite of ground motions. However, this phase of the work is still ongoing. While it is clear that these ground motion simulations are also very damaging to 6-story steel moment frame buildings, the analysis is complicated by the fact that the 6-story building has a linear fundamental period of 1.5 seconds and this particular

set of ground motions is low-pass filtered at 2 seconds. It is expected that broad-band ground simulations of these events are more appropriate for the analysis of 6-story buildings. Rob Graves has produced broad-band simulations of Puente Hills ruptures that will be used to simulate both 6- and 20-story building simulations.
Figure 6. This figure shows the composite results from a simulated earthquake on each of the ten faults. The maximum building response of all earthquakes at each location is shown. There is no information about predicted recurrences on each fault included in this composite. A building collapse at a location is counted only once, even if two or more earthquakes simulate a collapse at that location. In this view, 113 twenty-story buildings with good welds collapse in the simulation.
Earthquakes on the faults selected for this study most significantly affect model buildings in the northern half and eastern side of the Los Angeles Basin.

Figure 7. As described in Figure 6, this figure shows the composite results from twenty-story buildings with brittle welds to ten earthquakes. 402 locations experienced collapse of the model building in one or more of the simulated earthquakes. A twenty-story building with brittle welds located anywhere in the
Los Angels Basin is at risk of permanent deformation. The northern San
Fernando and San Gabriel Valleys also show significant building responses to the ten simulated earthquakes.
In addition to potential earthquake sources within the Los Angeles Basin, there is still tremendous uncertainty about the performance of tall flexible buildings in future large earthquakes on more distant faults, especially the San Andreas fault whose southern segments have not ruptured for more than 300 years. A recent collaboration of SCEC scientists have simulated the long-period ground motions that might occur in the Los
Angeles area if a rupture similar to the Denali earthquake were to occur on the southern
San Andreas fault (Olsen, K. B., S. M. Day, J. B. Minster, Y. Cui, A. Chourasia, M.
Faerman, R. Moore, P. Maechling, and T. Jordan, 2006, Strong shaking in Los Angeles expected from southern San Andreas earthquake, Geophysical Research Letters , 33 ,
L07305, doi:10.1029/2005GL025472). This simulation indicates that there is a potential for surprisingly large long-period motions that might occur in metropolitan Los Angeles if a large San Andreas event were to initiate in the Salton trough and then propagate northward to the Cajon Pass. The peak ground velocities for this simulation are shown in

Figure 8. Figures 9 and 10 show the peak inter-story drifts for 20-story steel frame buildings with perfect welds and with brittle welds, respectively.
Figure 8. Peak ground velocities from a simulation of a large earthquake (M7.8) propagating from the Salton Trough to Cajon Pass on the San Andreas Fault
(Olsen and others, 2006). Surprisingly large ground velocities are indicated for the Los Angeles region.

Figure 9. Peak inter-story drift for 20-story buildings with perfect welds and for simulated ground motions of a M 7.8 earthquake on the southern San Andreas fault that ruptures from south to north.

Figure 10. Peak inter-story drift for 20-story buildings with brittle welds and for simulated ground motions of a M 7.8 earthquake on the southern San Andreas fault that ruptures from south to north.

Principle findings

Near-source ground motions from recognized faults in the Los Angeles basin are large enough to severely damage and potentially collapse 20-story steel frame buildings in many parts of the basin.

Retrofitting brittle welds could decrease the number of collapses to less than half.

The ground motions from the Terashake simulation of a southern San Andreas earthquake produce ground motions that could severely damage and potentially collapse 20-story buildings in much of the Los Angeles metropolitan region.
Publications
Heaton, T., A. Olsen, and J. Hall, 2006, Flexible Steel Building Responses to a 1906 San
Francisco Scenario Earthquake, Seism. Soc. Am., abstract SSA-000458.
A. Olsen, Heaton, T., and J. Hall, 2006, Flexible Steel Building Responses to Scenario
Earthquakes in the Los Angeles Basin, Seism. Soc. Am., abstract SSA-000455.