Seismic structural response from continuous and discrete models

Abstract

The Northridge Earthquake of 1994 caused unexpected cracking in the connections of Steel Moment-Resisting Frame (SMRF) buildings. This thesis investigates the possibility that higher-than-expected strain rates in the near-field region of the earthquake were partially responsible for the cracking. A shear-beam model is proposed as possibly a better model for evaluating structural strain rate response for near-held ground motion. The shear-beam model is used to predict seismic response of a 20-story structure. The results from the shear beam are compared to seismic analysis using a more traditional vibratory model to see if the shear beam predicts higher relative values in the near-field region. It is found that the shear beam does indeed predict higher strain rates. However, the increased stain rates are not limited to the near-field region. The correlation between relative strain rates and epicentral distance is found to be negative-indicating higher relative slain rates in the near-field region-but the magnitude of the correlation is too low to be considered significant. The analysis is repeated with increased damping and it is found that the magnitude of the correlation increases, but is still not significant. Long-period wave forms are analyzed with the shear beam and vibratory models to determine if the band-pass altering of the seismic records could be altering the results. The analysis indicates that the shear beam does not predict response levels that are significantly higher than the vibratory model for long-period wave forms. A truncated (first-mode) modal analysis is conducted to determine how limiting the structure to a single degree of freedom affects results. It is found that the first-mode analysis predicts much lower stain rates than both the shear-beam and complete vibratory models. However, the results do not show significant correlation with epicentral distance. It is concluded that increased slain rate in the near-held region of the Northridge Earthquake was unlikely to have been a significant factor in the observed connection cracking. It also appears that traditional vibratory analysis may underpredict both the strain and the strain rate in seismic response.