Files in this item



application/pdfMEI-DISSERTATION-2018.pdf (36MB)
(no description provided)PDF


Title:Pore pressure generation and liquefaction analysis using nonlinear, effective stress-based site response analysis
Author(s):Mei, Xuan
Director of Research:Olson, Scott M.; Hashash, Youssef
Doctoral Committee Chair(s):Olson, Scott M.
Doctoral Committee Member(s):Rutherford, Cassandra J.; Elbanna, Ahmed
Department / Program:Civil & Environmental Eng
Discipline:Civil Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Liquefaction evaluation
Site response analysis
Soil constitutive model
Effective stress analysis
Pore water pressure generation
Abstract:Porewater pressure (PWP) generation leading to soil softening and potential liquefaction in sandy soils are pervasive problems during earthquakes. However, our ability to predict the impacts of PWP generation on ground motions and the resulting response spectra are not particularly good. While software is available to perform nonlinear effective stress site response, there are no guidelines for assessing the accuracy of such models. Also, most design codes classify sites susceptible to significant softening as Site Class F and require site-specific ground response analyses. Even if site-specific analyses are performed, many of these codes require that the resulting site-specific acceleration response spectrum not be less than 80% of the code-prescribed spectrum computed using Site Class E based on experience from “total stress” analyses and does not explicitly account for sites where significant PWP increase is likely to occur. Unfortunately, little guidance is available to develop more defensible code commentary. In this research, cyclic and monotonic shear tests have been collected to evaluate and validate the modified Vucetic and Dobry (1986) PWP generation model coupled with a generalized/Hyperbolic constitutive model (GQ/H+u) implemented in 1D site response analysis code DEEPSOIL. Results indicate that the GQ/H+u model provides reasonable estimates of PWP increase and stress-strain behavior during cyclic shear element tests but, as expected, cannot simulate soil dilation when the excess PWP ratio is high (ru> 0.8). Available centrifuge tests and field case studies have been simulated to further validate the performance of GQ/H+u model. Again, the GQ/H+u model reasonably approximates PWP generation and shear stress - shear strain response in these cases over a wide range of relative density but cannot simulate soil dilation when ru > 0.8. Despite this limitation, the GQ/H+u model was able to reasonably capture acceleration response at in the centrifuge tests when either of the following criteria were met: (1) computed ru < 0.8; or (2) computed ru> 0.8 and computed maximum shear strain was less than the limit shear strain (γmax <γlimit). A parametric study of both synthetic and published soil profiles has been performed using effective stress-based, nonlinear site response analysis. The parametric study yielded liquefaction resistances that generally were in excellent agreement with published liquefaction resistance curves (Andrus and Stokoe 2000; Kayen et al. 2013) for Vs1 < 200 m/s. The cases where agreement was not as good involve sandy soils with Vs1 > 200 m/s, input motions that exhibit near-fault effects, and cases where soft clay layers underlie the loose, liquefiable sand layers. Each of these conditions are explained and quantified. Response spectra were developed based on GQ/H+u model, a hybrid method, and code requirements for a number of centrifuge experiments and well-documented field case histories, and these spectra then were compared with measurements to assess the validity of these methods. For cases in which the computed ru < 0.8 or ru > 0.8 and γmax < γlimit by the GQ/H+u model, dilation tends to have a limited effect and the GQ/H+u model is likely to provide reasonable estimates of response spectra at T > 0.4s. In layers/cases where ru > 0.8 and γmax > γlimit, the occurrence of dilation spikes can increase spectral accelerations at both short and long periods, making them larger than the 80% of Site Class E spectral accelerations. Here, analysis results from the GQ/H+u model, as well as results from more sophisticated constitutive models, do not match the measured spectral accelerations, particularly if the maximum spectral velocity (Sv) of the input motion occurs at long periods (T > ~1 second). Until models are better able to capture this response on a consistent basis, an interim approach may be to envelope the GQ/H+u and 80% of Site Class E spectra for use in design. The effect of PWP generation on response spectra has been investigated through a parametric study, the results of which indicate that a factor of safety against liquefaction triggering (FSliq, computed by the cyclic stress method) < 1.4, commonly corresponds to ru < 0.8. As such, a cyclic stress-based FSliq < 1.4 can be used as a threshold for defining when NL-ES analysis should be conducted at a potentially liquefiable site.
Issue Date:2018-12-03
Rights Information:Copyright 2018 Xuan Mei
Date Available in IDEALS:2019-02-08
Date Deposited:2018-12

This item appears in the following Collection(s)

Item Statistics