Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
24 CHAPTER FOUR SEISMIC HAZARD ANALYSIS As indicated in chapter one, PBSD requires an estimate of the seismic hazard for the site in question. This hazard includes effects of both the regional tectonics and the local site characteristics. From either a deterministic or probabilis- tic viewpoint, this design step is perhaps the best developed. Deterministic seismic hazard analysis was the first method- ology developed; it comprises four steps (Reiter 1990): 1. Identification of the seismic sources that affect a site 2. Selection of a source-to-site distance parameter, often taken as the closest approach to the site and 3. either hypocentral or epicentral distance, along with an attenuation relationship that defines hazard as a function of distance to the site from the source 4. Selection of the controlling earthquakeâthe earth- quake that could produce the largest shaking 5. Generation of the hazard at the site (often ground acceleration or spectral acceleration) using the source-to-site attenuation relationship. The deterministic form allows one to assess the shaking at a site as a function of the controlling earthquake that can occur on all the identified faults or sources. The methodology was extended to a probabilistic basis by Cornell (1968) and first mapped for the country by Alger- misson et al. (1982). This approach, with subsequent refine- ments, still forms the basis for the seismic maps used with the AASHTO specifications and the various building codes. With probabilistic seismic hazard analysis, the individ- ual steps listed previously are each put on a probabilistic basis that takes into account uncertainties in the occur- rence of earthquakes on any sources affecting a site, in the attenuation relationships for a local region, and in the local seismic settings. These relationships are then combined to form the probability that a given acceleration may be exceeded during a given window of time (e.g., a 7% chance of exceedance in 75 years). Alternatively, the probability may be related in terms of an average return period for a given value of seismic hazard (e.g., ground acceleration or spectral acceleration). Today seismic hazard is mapped by the U.S. Geological Survey (USGS) and is provided on a gridded basis for the entire country and territories. The hazard data that USGS provides are for firm soil or rock. The site seismic hazard is then built considering site effects. This process may use the so-called National Earthquake Hazards Reduction Pro- gram (NEHRP) site classification factors that form the basis of the AASHTO methods or may use a site-specific study to develop the accelerations at the ground surface. USGS is continually improving the source, uncertainty, and attenua- tion data for seismic hazard and issues updated geographic acceleration data on a several-year cycle. The process of developing site hazard in this manner is clearly defined in the AASHTO specifications and their associated commentary. The USGS data that AASHTO currently uses in its tools to assist engineers in determining site hazard are normally in the form of uniform hazard spectra. These then relate struc- tural response (spectral acceleration) in a form that has a uniform chance of being exceeded in some window of time. For AASHTO, this window is nominally 1,000 years (actu- ally 975 years), or approximately a 7% chance of exceedance in 75 years. Seismic hazard is generally included in design using three parameters: site-adjusted peak ground acceleration, spectral acceleration at 0.2 seconds, and spectral accelera- tion at 1.0 second. These values are normally used to deter- mine the shape of a design response spectrum for the site, or they could be used with site-specific response spectra to define a unique seismic input for each site. The former is more common, especially with conventional bridges. When considering PBSD, the concept of using more than one return period of seismic hazard is often included. For instance, this might result in use of both a 100-year return period and 1,000-year return period for the design checks of a bridge. The performance in each event might be dif- ferent, with operational performance for the more frequent event and avoidance of collapse for the larger event. Such an approach is rational in terms of providing higher protection and less risk in the more frequent event. Then, significant damage and potential loss of service are considered for the larger event. However, it is important to recognize that the relationship between the acceleration levels for two events is not the same across the country. This is illustrated in Figure 7,
25 excerpted from the Multidisciplinary Center for Earthquake Engineering Research MCEER/ATC-49 reports (2003). In the figure, the ratios of the 1-second spectral acceleration for return periods higher than 475 years, up to 2,475 years, are shown for the central and eastern United States (CEUS), the western United States (WUS), and California. Ratios are provided for some 24 locations across the United States, including Alaska and Hawaii. In the CEUS, the ratio is over 5 for Memphis, Tennessee, and Charleston, South Carolina, at 2,475 years. This value is over 2 at the 1,000-year return period level. By contrast, the ratio never exceeds about 1.6 in California, even for long return periods. Other areas in the WUS have intermediate values from 1.5 to 3.5 at 2,475 years and 1.5 to 2.2 for 1,000 years. The trends are similar for the short-period spectral accelerations (0.2 seconds), and for the data more recently developed by USGS. Clearly, the hazard of more frequent earthquakes relative to rare earthquakes is different across the country, and this difference may affect the approach used if multiple-level earthquakes are used for PBSD of bridges. In recognition that the seismic hazard varies signifi- cantly between frequent and rare events across the United FIGURE 7 Ratios of long-period spectral acceleration at various return periods to the long-period spectral acceleration at 475 years (after FHWA 2006 and MCEER/ATC 2003).
26 States, and that the risk of collapse or attainment of other damage levels is not uniform across the county, the ASCE 43-05 Seismic Design Criteria for Structures, Systems, and Components in Nuclear Facilities standard (2005b) presents a method to adjust design spectra to account for the uneven hazard. The commentary of the ASCE 43-05 standard has one of the better descriptions of the approach. This meth- odology has also been picked up in a slightly modified form in the ASCE 7-10 (2010) standard, as described previously. Typically, as PBSD is taken into the full probabilistic for- mat, median predictors are desired for all parameters that are used. In these median predictors, factors that provide the dispersion around the median value are used to bring in the probabilistic nature of the parameters. With uniform hazard spectra, often no single earthquake can produce the shaking indicated by the uniform hazard design spectra. To address the statistical likelihood that accelerations all along the spec- trum cannot be created by a single earthquake, the conditional mean spectrum (CMS) has been developed. In the CMS, a single period of vibration is selected; a statistical spectrum that would be associated with the control period is then devel- oped. These spectra are typically lower than uniform hazard spectra, and the CMS is often used in conjunction with the fully probabilistic PBSD process. This is different than the seismic hazard process that AASHTO is currently using.
