Study of response of buildings to the 1994 Northridge, California earthquake

Aditya Tan, Purdue University

Abstract

Estimating building damage due to strong earthquake ground motions is a difficult endeavor. Various earthquake intensity scales, such as the field survey-based Modified Mercalli Intensity (MMI) scale and the instrument-based Instrumental Modified Mercalli Intensity (Imm) scale, are used to describe the strength of ground shaking during an earthquake. If ground motion recording stations are available in the affected area, the current approach employed by the United States Geological Survey (USGS) to estimate the shaking intensity is the automated Imm approach, in which the peak values of ground acceleration and ground velocity are used. As rapid as it is, this instrument-only based approach does not account for the characteristics of buildings and, therefore, may not provide useful information about the damage state of the built environment following an earthquake. Studies have shown that this method cannot estimate the degree of building damage accurately. The inaccuracy can be attributed mainly to the following: (1) ground shaking intensity is an ambiguous representation of building damage, and (2) structural damage does not depend solely on the ground motion but also on how the structure interacts with ground shaking, which depends on the characteristics of the buildings. This research focuses on finding reliable building damage indicators using the inspection records of 104,025 buildings surveyed in the aftermath of the 17 January 1994 Northridge, California earthquake and publicly available ground motion records from that earthquake. In this study, damage is represented using damage severity levels as prescribed by the inspectors. Each structure is associated with the ground motion parameters obtained from the closest ground motion recording station or the closest geographic grid point provided by the USGS. As the nature of the dependent and some of the independent variables are ordered and integer-valued, besides the regular statistical correlation analysis, a random parameter ordered probit statistical model is considered in the study. A critical evaluation of parameters that have strong influence on building damage is provided. The impact of distance to recording station on observed correlations is also presented.

Degree

M.S.C.E.

Advisors

Irfanoglu, Purdue University.

Subject Area

Civil engineering

COinS