...Historical Examples Cover Page Conclusions...
Scenario Earthquakes for Urban Areas Along the Atlantic Seaboard of the United States
Scenario Events, Probability-Based Code Guidelines and Site-Specific Design Ground Motions
Scenario events are useful for providing information on how various sample earthquakes may produce losses and loss patterns depending on the individual earthquake’s location and magnitude. Scenario events, however, do not tell how likely they are. In contrast, seismic hazard maps (e.g. Frankel, 1995) to which seismic code design guidelines are currently still tied, have a different purpose. They attempt to define the seismic hazard, i.e. expected ground shaking level at every location, in terms of a preselected "probability of non-exceedance."
Codes. Currently most U.S. seismic code provisions use a probability of 90% that a certain ground motion parameter (e.g. "peak acceleration") shall not be exceeded in a given time period called "exposure time," which for most U.S. codes is currently chosen to be 50 years. Conversely, the 90% non-exceedance implies, of course, that there is a 10% probability that the mapped ground motion parameter will be exceeded during this exposure time. (Note: using equation (2), the 10% probability of exceedance, P = 10, when combined with an exposure time of Dt = 50 yr, yields an average recurrence period of T = 475 yr for the mapped ground motion). Often the probabilistic ground motion maps (in the strict sense) are modified for code purposes into seismic zoning maps or seismic design maps (for instance in the UBC, the Uniform Building Code, 1990; or the recent draft for a future New York State Seismic Building Code). The purpose is to outline zones with uniform design ground motion parameters, which are easier to administer than the contoured values which require finding the applicable design parameter by interpolation. An example for such a zoned seismic code map (for New York State) is shown in Figure 5.
The mapped ground motion or design parameters are then used to produce a "design spectrum," which yields the horizontal design forces as a function of natural period of the structure in question. Each structure tends to vibrate (like an oversized tuning fork) in a resonance of its own, and does so with a given fundamental natural period (plus overtones at shorter periods). Codes prescribe in detail how to compute the fundamental period of a given type and height of structure. A simple approximate rule may suffice here which tells us that for every story of structure one needs to add about 0.1 seconds of natural period. Hence a ten-story building would have a natural period of about 1 sec and a 50-story high-rise a 5 sec period. Figure 6 shows an example of such code design spectra, taken from the NY City seismic building code. The vertical axis depicts the spectral response acceleration (or applied horizontal force) plotted against the structural natural period (on the horizontal axis).
Figure 6 shows multiple design spectra which apply to different soil or rock conditions on which the structure is founded. Seismic ground motions are strongly modified by the overlying soils, depending on the softness of soil and its thickness. Code spectra try to reflect that soil behavior in a schematic way. That is why Figure 6 shows separate curves for hard rock (S0), soft rock (S1) and ever softer and deeper soils, up to S4 which represents a thick layer of very soft clays or similar
soft materials of poor bearing capacity and low stiffness. In the New York City code, the maximum ratio of design accelerations between S4 and S0 site conditions is 3.75 for building periods longer than about 1 sec. This has been the first U.S. code to specify such high soil amplification. Recent national codes (NEHRP, 1994; and future editions) now provide similar values, but in addition they make the soil amplification ‘nonlinear.’ That is, for a given soil profile the amplification factor becomes incrementally lower the higher the rock input acceleration is.
Many but not all eastern U.S. states have recently adopted seismic building codes. New York State is one that has not (as of 1995), but New York City has (February 21, 1995, to become effective February 21, 1996). Since building codes apply only to newly built structures, it will take on the order of a century before sufficient numbers of old structures are replaced by newly built ones before the earthquake-loss mitigating effects of the seismic code provisions become significant. Building codes apply to a wide range of structures, most commonly, however, to the numerous ordinary buildings. Similarly, bridge codes (e.g. AASHTO, 1992) apply to the ordinary highway bridges. A serious earthquake may well occur before most existing buildings and ordinary bridges are replaced by new ones that are seismically designed. Therefore, it seems only prudent to focus on checking out the seismic resilience of the most critical lifelines and key buildings on whose functioning we must be able to rely, even after a major earthquake strikes. For such priority structures, local communities may press forward with a seismic rehabilitation program although current codes generally do not require such actions.
Site-Specific Design Motions for Important or Critical Structures. For special structures such as nuclear power plants, essential lifelines, monumental bridges with high traffic volume, or even important buildings, it is important and/or cost-effective to go beyond the ordinary code requirements and design procedures. In many such cases, site specific ground motions are evaluated, which typically are tied to annual exceedance probabilities (in case of probabilistic procedures), or are tied to specific design earthquakes (M-d combinations for deterministic procedures, possibly connected to recurrence periods and performance criteria appropriate for the importance of the structure). Often only site-specific design spectra (rather than time history ground motions) are obtained, but more often the ground motion time series themselves are needed, especially if the design procedure allows or requires the nonlinear response of the structure to be considered.
We show examples for design motions that were recently derived for two bridges in or near New York City (Jacob et al. 1994, 1995): the Queensboro Bridge across the East River connecting Queens with Manhattan, and the Tappan Zee Bridge carrying the NY State Thruway across the Hudson River some 20 miles north of the city. Both bridges carry, on average, more than 100,000 vehicles per day and are essential lifelines in the region’s economy.
Queensboro Bridge Hard-Rock Motions. In the case of the Queensboro Bridge, the owner opted for a two-level design choosing a long recurrence period (2,500 yr) to satisfy performance criteria concerned with structural collapse, and a shorter recurrence period (500 yr) concerned with continued serviceability of the bridge. Three-component acceleration time-series and damped response spectra for three magnitude distance pairs (M-d) per recurrence period (taken from Table 2) were computed by a method described in Horton (1994). The response spectra for the three events per constant recurrence period are then enveloped which yields a single envelope spectrum for a suite of events with a given constant recurrence period. Therefore, the constant recurrence period envelope spectrum approximates a nearly uniform hazard level. Samples of the computed ground motions in the time domain are shown in Figure 7. Their envelope spectra are shown in Figure 8. An envelope spectrum represents the largest spectral response at each period chosen from the suite of M-d events representing a given constant recurrence period. Typically, the highest spectral levels at short periods are contributed by small magnitudes at short distances, and the highest spectral levels at long periods are contributed by large events at large distances. Envelope spectra represent deterministically derived ground motions for a constant recurrence period over the range of structural periods, and hence are similar to but not identical with probabilistically derived uniform hazard spectra.
We now compare the computed site-specific envelope spectra with various code-based spectra (Figure 9). The essence of this comparison is: at short periods, T ?/font> 0.1 sec, the site specific Queensboro Bridge hard-rock, envelope spectrum for the 500 yr event suite exceeds all the different code design spectra by a considerable margin. Conversely, at nearly all periods T .15 sec, most code spectra exceed the Queensboro Bridge 500 year spectrum by a considerable margin. A similar comparison of code spectra with the 2,500 year envelope spectrum for the Queensboro Bridge hard-rock site shows that the 2,500 year envelope spectrum exceeds all code spectra for periods T .5 sec, and tracks for longer periods the various code spectra with variable degrees of fidelity for periods up to 4 seconds. The comparisons emphasize that in the eastern U.S. the AASHTO code spectra for hard-rock sites tend to underpredict short-period spectral levels, but overpredict long-period spectral levels. Note that this characterization applies to hard-rock spectra. We discuss next the effects of soils on site specific motions and their spectra.
Tappan Zee Bridge Ground Motions Including Nonlinear Soil Response. Location-specific, soil-response-modified ground motions and their response spectra were computed from the same hard-rock motions as used for the Queensboro Bridge, except that the bridge owner also requested motions for a 1,000 year recurrence period. To account for the nonlinear soil response, we compute layer strains at different depth levels to determine iteratively new shear moduli (or shear wave velocities) and damping values for each layer using empirically derived degradation relations by Vucetic and Dobry (1991). The new shear velocity and damping values are then used to calculate a new soil transfer function. The process is repeated several times until the changes in shear moduli and damping coefficients between iterations become insignificantly small. The final three-component acceleration records are computed at the specified depths using moduli and damping coefficients from the last iteration. These final accelerograms are processed to obtain 5% damped acceleration response spectra.
Figure 10 shows an example where the transverse horizontal motion from a 1,000 yr recurrence-period earthquake, M = 7 at a distance of 92 km (see Table 2), observed at a rock surface outcrop is compared to the transverse motion at a depth of 1.5 meter below the mudline near the main-span caisson of the Tappan Zee Bridge, assuming linear and non-linear soil response. The comparisons both in the time and frequency domains clearly show how high frequencies are suppressed (above 4 Hz), and how longer periods (at about 1 Hz) are strongly amplified by factors of up to about 10; they also show how the nonlinear soil response shifts the soil resonance peaks to lower frequencies (or longer periods), in this case from about 1 Hz to about 0.6 Hz, thus exciting many of the long-period (i.e. fundamental) modes of the main-span structure much more than if they would be founded on hard rock.
Nonlinear soil amplification is very prominent for large earthquakes, e.g. the M = 7 at d = 92 km for the constant recurrence period of about 1,000 yr. This event provides large input motions at long periods due to a nonlinear ‘runaway’ degradation in the soil parameters from large transient strains. The spectral behavior and non-linear response at this site and for this event show how complicated the interactions of the different factors influencing the ground motions sometimes can be. At shorter spectral periods, and for sites on rock or with thin soil cover, the motions associated with smaller earthquakes at short distance (e.g. M = 5 at 15 km) dominate the constant recurrence period spectral response. These examples illustrate that different scenario events, even if they belong to the same constant recurrence period, and hence are equally likely, produce quite different response-spectral behavior depending on the M-d, and soil combinations at play.
In Figure 11 we compare the soil response spectrum for 1,000 yr events at the Tappan Zee Bridge site with code design spectra. It is quite apparent that the code spectra are incapable of properly predicting the strong site-specific soil response. This example indicates the need and value of site-specific studies of design ground motions for essential structures, and points to potential risks that could be incurred if only code spectra would be used instead.
Final results from the engineering analysis of the Queensboro Bridge and Tappan Zee Bridge using the site specific ground motions discussed here are not yet available. But preliminary results indicate that both bridges may be vulnerable to a variety of damage patterns given the site-specific motions provided. These are only the first steps towards gaining insight on what engineering efforts and financial resources may be needed in the future for making bridges that are quite critical for the economy of the New York City region more resistant against earthquakes.
Home | Background | HAZUS | Research Activities | Education/Outreach
Who We Are | Technical Documents | Contacts | Links
Last modified: 4/18/2002 10:08:12 AM
Suggestions and comments to the webmaster.