EFFECT OF SPATIAL VARIABILITY OF SOIL PROPERTIES ON SAFETY FACTORS OF CANTILEVER RETAINING WALLS
Abstract:
Spatial variability is one of the most significant characteristics of soil
properties, even within homogeneous layers. The physical description of this spatial
variation is unclear, due to the highly expensive sampling, uncontrollable
measurement errors, and modeling uncertainty. While the deterministic approach
failed to present a reasonable quantification for the spatial variability of soil
properties, the probabilistic approach has been utilized to assess and quantify the
effects of soil spatial variability on the behavior of some typical soil-structure
systems.
This study investigates the effect of spatial variability of soil properties (mass
density γs, coefficient of friction tanδ, angle of internal friction (φ), and wall mass
density γc, on the factors of safety against sliding (Fs), and overturning (Fo), as well
as the maximum bending moment (M) of cantilever retaining walls.
For risk assessment, soil properties are described using appropriate
probabilistic models. Deterministic analysis combined with Monte Carlo Simulation
(MCS) is employed to analyze both spatially fully correlated and spatially
uncorrelated soil profiles.
A FORTRAN program is developed to: (1) generate random variables with
both Gauassian and non-Gauassian probability density functions (PDF); (2) simulate
possible probability distributions for distinct soil properties; (3) perform repetitive
analysis for each set of variable PDFs to generate the population of Fs, Fo, and M;
and (4) calculate the statistical parameters of Fs, Fo, and M. The program is then
applied to conduct an extensive parametric study.
This parametric study is performed to quantify and assess the effects of model
uncertainty and the effect of spatial heterogeneity of soil properties, utilizing two
different soil profiles: (1) spatially fully correlated soil profile; and (2) spatially
uncorrelated soil profile.
The numerical results show that the effects of spatial variability and
uncertainty of input variables on wall safety vary from one variable to the other.
ABSTRACT
xx i
Consequently, considerable care should be directed to those variables that have the
greatest effects. For typical degrees of uncertainty, it is found that the structural risk is
highest when the PDF of tanδ and φ is normal (Gauassian), or has an upper triangular
shape (non-Gauassian). In addition, it is found that accounting for soil spatial
variability in design of retaining walls leads to more economic designs, as it produces
smaller probabilities of failure. In other word, negligence of the spatial variability in
the design of retaining wall is conservative.
Download
*
Abstract:
Spatial variability is one of the most significant characteristics of soil
properties, even within homogeneous layers. The physical description of this spatial
variation is unclear, due to the highly expensive sampling, uncontrollable
measurement errors, and modeling uncertainty. While the deterministic approach
failed to present a reasonable quantification for the spatial variability of soil
properties, the probabilistic approach has been utilized to assess and quantify the
effects of soil spatial variability on the behavior of some typical soil-structure
systems.
This study investigates the effect of spatial variability of soil properties (mass
density γs, coefficient of friction tanδ, angle of internal friction (φ), and wall mass
density γc, on the factors of safety against sliding (Fs), and overturning (Fo), as well
as the maximum bending moment (M) of cantilever retaining walls.
For risk assessment, soil properties are described using appropriate
probabilistic models. Deterministic analysis combined with Monte Carlo Simulation
(MCS) is employed to analyze both spatially fully correlated and spatially
uncorrelated soil profiles.
A FORTRAN program is developed to: (1) generate random variables with
both Gauassian and non-Gauassian probability density functions (PDF); (2) simulate
possible probability distributions for distinct soil properties; (3) perform repetitive
analysis for each set of variable PDFs to generate the population of Fs, Fo, and M;
and (4) calculate the statistical parameters of Fs, Fo, and M. The program is then
applied to conduct an extensive parametric study.
This parametric study is performed to quantify and assess the effects of model
uncertainty and the effect of spatial heterogeneity of soil properties, utilizing two
different soil profiles: (1) spatially fully correlated soil profile; and (2) spatially
uncorrelated soil profile.
The numerical results show that the effects of spatial variability and
uncertainty of input variables on wall safety vary from one variable to the other.
ABSTRACT
xx i
Consequently, considerable care should be directed to those variables that have the
greatest effects. For typical degrees of uncertainty, it is found that the structural risk is
highest when the PDF of tanδ and φ is normal (Gauassian), or has an upper triangular
shape (non-Gauassian). In addition, it is found that accounting for soil spatial
variability in design of retaining walls leads to more economic designs, as it produces
smaller probabilities of failure. In other word, negligence of the spatial variability in
the design of retaining wall is conservative.
Download
*