Degree Type

Thesis

Date of Award

2020

Degree Name

Doctor of Philosophy

Department

Civil, Construction, and Environmental Engineering

Major

Civil Engineering (Geotechnical Engineering)

First Advisor

Vernon R Schaefer

Abstract

Determination of the soil parameters for conducting stability analysis of first-time slides in stiff-fissured clays is still a significant challenge in the geotechnical field. Currently, instability analyses for these types of soils are mainly conducted by applying the fully softened shear strength (FSS) concept. However, this concept is only a practical approximation not applicable to all types of failures in first-time slides in stiff-fissured clays. This study presents the compilation of several advanced statistical assessments to evaluate the factor of safety (FS) of slopes to further focus on the specifics of slope instabilities in stiff-fissured clays. For this purpose, the integration of statistical design of experiments with analytical methods (limit equilibrium and finite element analyses) is performed to evaluate the factors and interactions affecting the FS of slopes for different soil configurations. This evaluation allows for understanding of the controlling factors on FS for general soils not prone to softening. In addition, the use of statistical assessments is a novel application for the study of pile-reinforced slopes to determine the effects of reinforcement related factors on FS for optimal design purposes. Bayes' theorem is applied as diagnostic testing to determine the closeness of analytical tools to the field condition. This assessment delineates the accuracy of the current analytical methods and demonstrates the shortcomings of back-calculated parameters. Regarding clays prone to softening, the prediction power of the existing correlations based on FSS concept is evaluated, generating a regression-based neural network (ANN) to simulate the laboratory determined FSS for the study and quantification of the relationships among the variables. Lastly, an original multiple response artificial neural network (MRANN) approach is undertaken to develop a prediction model that estimates three responses simultaneously. These three responses correspond to curve fitting parameters for a non-linear failure envelope to characterize the mobilized shear strength of stiff-fissured clays. Several comparisons between the current methodology, the field condition, and the proposed modeling approach are provided to demonstrate the validity of the presented technique. The outcomes of this study provide three prediction tools that facilitate the application of research findings in the geotechnical practice without the need for special software for stages of preliminary design. Other benefits from this study include contributions to research needs regarding the use and application of the FSS concept, contributions to optimize the selection of slope remediation techniques, contributions to the geotechnical profession by quantifying the difference between analytical methods, and contributions to design optimization for pile-reinforced slopes. Finally, the major contribution to advance the understanding of the mobilized shear strength of stiff- fissured clays resides in providing a general approach that could help overcome the applicability constraints of the FSS concept.

DOI

https://doi.org/10.31274/etd-20200624-192

Copyright Owner

Yuderka Trinidad Gonzalez

Language

en

File Format

application/pdf

File Size

166 pages

Available for download on Wednesday, June 16, 2021

Share

COinS