UC Berkeley

The strong shaking of the 2010-2011 Canterbury earthquake sequence caused widespread liquefaction in much of the city of Christchurch, NZ, including large parts of the central business district (CBD). The most intense event of the earthquake sequence with regards to shaking in the CBD was the 22 FEB 2011 Mw6.2 Christchurch event. A reported 185 casualties were caused by the destruction from this event. The financial damage was immense as many of the structures, especially in the CBD, had to be demolished due to damage sustained from the Christchurch earthquake as well as some of the other less intense events. The performance of structures during these events was often related to liquefaction of foundation soils. Understanding the effects of liquefaction on soil and building response is an area of earthquake engineering that continues to challenge practitioners and researchers. This challenging topic makes the study of case histories, such as those provided by the Canterbury earthquakes, an essential component to characterizing and understanding the effects of soil liquefaction on building performance. This thesis focuses on providing insights regarding the seismic response of liquefiable soils through the use of information and data collected in Christchurch, NZ following the Canterbury earthquake sequence.

Nonlinear effective stress site response analyses are often used by engineers to model the dynamic response of potentially liquefiable soils during strong shaking. For the presented research, a widely used one-dimensional nonlinear effective stress site response analysis program is used to perform this modelling. Ground motions recorded during six events of the 2010-11 Canterbury earthquake sequence are used in conjunction with the extensive site investigation data that has been obtained in Christchurch to complete site response analyses at several strong motion station sites in the greater Christchurch area. Deconvolved Riccarton Gravel input motions were generated, because representative, recorded rock or firm layer base-motions were not available in the Christchurch area. Nonlinear effective stress seismic site response analyses are shown to capture key aspects of the observed soil response through the comparison of acceleration response spectra of calculated surface motions to those of recorded surface motions; however, equivalent-linear and total stress nonlinear analyses are shown to capture these aspects as well. Biases in the computed motions compared to recorded motions were realized for some cases, but they can be attributed primarily to the uncertainty in the development of the input motions used in the analyses.

The study of the consequences of liquefaction on building performance is a complex soil-structure interaction problem that requires the use of well-documented case histories for validation purposes. An extensive site investigation and advanced laboratory testing program was carried out in Christchurch, NZ from April to October of 2014. The aim of this work was to provide characterization of the liquefaction resistance of foundation soils from building sites affected by liquefaction during the Canterbury earthquakes. In-situ penetration tests, such as CPT, are valuable methods for gaining an initial understanding of a site’s characteristics and the expected performance of critical soil layers. However, to understand fully the complex response of soil at the element level, laboratory testing of relatively undisturbed soil specimens provide unique insights. To accomplish these goals, “undisturbed” sampling and triaxial testing (monotonic and cyclic) were performed on soils from key building sites in Christchurch’s CBD. High quality sampling and testing could be achieved for most of the predominantly silty and sandy soils in the CBD. Test results indicate, though, that loose clean sand specimens were densified significantly during the sampling with the Dames & Moore hydraulic fixed-piston sampler (an Osterberg-type thin-walled sampler). The cyclic resistances measured in the tests on “undisturbed” specimens were generally consistent with those estimated using empirical simplified liquefaction triggering procedures.

Important insights regarding the cyclic response of the shallow CBD soils were obtained through the laboratory testing carried out as a part of the research presented in this thesis. Triaxial testing of “undisturbed” soil specimens proved important in understanding not only the stress-strain response of the studied soils, but also allowed for further knowledge regarding the pore water pressure response of the tested soils during both cyclic and monotonic loading. Importantly, insights into how various soil types of the CBD responded to cyclic loading were gained through comparisons of cyclic triaxial (CTX) tests performed on a variety of sand and silty sand soils. It was seen through CTX results comparisons that silty sands (soils classified as SM) and clean sands (SP and SP-SM) responded similarly in cyclic loading, even when the fines content of the tested specimens differed. Monotonic triaxial testing was also performed on reconstituted specimens to characterize the steady state response of several soil units in the CBD. The extensive insight garnered from laboratory testing is critical for informing researchers and engineers studying the case histories provided by the Canterbury earthquakes of buildings founded on liquefiable soils, especially those using numerical-based soil-structure interaction analyses. The results of the monotonic and cyclic tests performed as part of this study provide useful data for calibrating advanced numerical models.

Appendices are included as a part of this dissertation to provide supporting information and data not included in the main body of the thesis. A majority of these appendices are included at the end of this thesis; however, there are several additional appendices included as electronic attachments to this dissertation, which provide supplementary information regarding the work presented in Chapters 2 through 4 of this dissertation. Electronic appendices that support the work presented in Chapter 2 include: rotated (fault normal and fault parallel) seismic records for events studied at the strong motion station sites of interest (zip file that contains text files), deconvolved Riccarton Gravel motions for all events studied (zip file that contains text files), and selected results for completed site response analyses (PDF file). Electronic appendices that support the work presented in Chapters 3 and 4 include triaxial test data and results for “undisturbed” soil specimens (zip file that contains text, data, and PDF files).