EFFECTS OF LONG-DURATION GROUND MOTIONS ON LIQUEFACTION HAZARDS

Abstract

Soil liquefaction during past earthquakes has caused extensive damage to buildings, bridges, dam, pipelines and other elements of infrastructure. Geotechnical engineers use empirical observations from earthquake case histories in conjunction with soil mechanics to predict the behavior of liquefiable soils. However, current empirical databases are insufficient to evaluate the behavior of soils subject to long-duration earthquakes, such as a possible Mw = 9.0 Cascadia Subduction Zone earthquake. The objective of this research is to develop insight into the triggering and effects of liquefaction due to long-duration ground motions and to provide recommendations for analysis and design. Recorded ground motions from 21 case histories with surficial evidence of liquefaction showed marked differences in soil behavior before and after liquefaction was triggered. In some cases, strong shaking continued for several minutes after the soil liquefied, and a variety of behaviors were observed including dilation pulses, continued softening due to soil fabric degradation, and soil stiffening due to pore pressure dissipation and drainage. Supplemental field and laboratory investigations were performed at three sites that liquefied during the 2011 Mw = 9.0 Tohoku earthquake. The recorded ground motions and field investigation data were used in conjunction with laboratory observations, analytical models, and numerical models to evaluate the behavior of liquefiable soils subjected to long-duration ground motions. Observations from the case histories inspired a framework to predict ground deformations based on the differences in soil behavior before and after liquefaction has triggered. This framework decouples the intensity of shaking necessary to trigger liquefaction from the intensity of shaking that drives deformation by identifying the time when liquefaction triggers. The timing-based framework promises to dramatically reduce the uncertainty in deformation estimates compared to conventional, empirically-based procedures.