Earthquake Characteristics and Damage Statistics

Abstract

A simple Markov distribution relates the probability of occurrence of five discrete damage states for a specific building type in an earthquake. Within the model the damage distribution depends on parameters that represent the building's structural system, the matching of the site and building periods, and the site materials, and the site's acceleration. The constants in the model were determined using a maximum likelihood formulation and damage observations for a series of California and Chinese earthquakes. Model predictions of damage distributions agree well with reserved damage data not used in determining the constants. A maximum likelihood method allows determination of ground motion, attenuation, and/or earthquake moment magnitude from observations of damage for different types of structures located at diverse sites. A simple relationship exists for average damage estimate that closely matches historical observations.

Chapter C. The Loma Prieta, California, Earthquake of October 17, 1989 - Building Structures

“…Many methods can be used: fragility curves, Markov models, or DPM's. An example of a Markov model approach is Thiel and Zsutty (1987). We use DPM's here.…”

Probable maximum loss estimation in earthquakes: an application to welded steel moment frames

“…The reasonableness of this representation for the damage function has been established by the extent of its agreement with a wide range of observed data (Thiel and Zsutty, 1987b). 4 Use of the Markov model has successfully reproduced the damage statistics for a wide range of building types and site conditions for earthquakes throughout this century, from the 1906 San Francisco earthquake through the 1985 Tangshan earthquake (Thiel and Zsutty, 1987b) and subsequent unpublished analysis of data through to the 1994 Northridge earthquake. The model requires the acceleration of the site, a, and values for three parameters to describe the structure and sites: s, b, and m. The ®rst parameter, s, characterizes the relative damageability due to site characteristics and depends on the geotechnical nature of the site and the depth to the water table.…”

“…Similar differences in performance in all types of structural systems are observed in every earthquake (Thiel and Zsutty, 1987a,b). 3,4 The result of this series of uncertainties is that the current state of the art requires that probable maximum losses be EARTHQUAKE MAXIMUM LOSS ESTIMATION determined in a statistically consistent manner that recognizes the uncertainties of: where, when and with what magnitude earthquakes will occur; the ground motion of the site; and the damage to the building from the ground motion. Building earthquake damageability projections are therefore naturally expressed as probability distributions, not as single numbers.…”

“…4 This model was developed to replicate damage statistics observed in historic earthquakes and extended to current construction types using the expert opinion results of the Applied Technology Council (1985) 5 using only the mean estimates of damageability for MMI VIII. The Model gives the distribution of possible damage in a given peak horizontal ground acceleration Figure 1.…”

Assessment of regional earthquake hazards and risk along the Wasatch Front, Utah; Volume III

“…High shaking levels in these events contributed significantly to expected culinary water damage. At first, we speculated that significant damage can occur to seismically vulnerable pipelines even at Modified Mercalli intensities as low as VIII and when no permanent ground displacement occurred (see [43] on the Coalinga earthquake, for which later estimates [48] suggest intensity IX or X may have occurred in the city of Coalinga). Eventually we determined that pipe vulnerability models need to be modified (downwards) at these higher intensities.…”