Академический Документы
Профессиональный Документы
Культура Документы
A~stract: Engineering judgment and reliance on factors of safety have been the conventional tools for dealing with
sot! heterogeneity in geotechnical practice. This paper presents a review of recent advances in treating soil variability.
It presents the implications of geostatistical techniques and up-scaling methods used for quantifying the heterogeneous
permeability of soil as addressed in the petroleum industry. Moreover, the interest of geotechnical practice to incorpo-
rate the statistical properties of soil in a probabilistic design framework is also discussed. This ranges from conven-
tional Monte Carlo simulation based design and stochastic finite element analysis to the recent techniques that take into
account the effect of spatial correlation of soil properties. Example applications of these techniques to different types
o.f field pro~l~ms: such as foundation settlement, seepage flow, and liquefaction assessment, are discussed with empha-
s~s on the IlIrutatlOns of the current practice and trends for future research. In addition, different decision making algo-
nthms are addressed with examples of their applications to geotechnical field problems.
Key words: heterogeneity, spatial variability, geostatistics, stochastic analysis, decision making.
Resumi:
Mots etes :
rrraduit par la Redaction]
Can. Geotech. J. 40: I-IS (2003) doi: IO.I139rr02-090 e 2002 NRC Canada
Monte Carlo simulation into detenninistic numerical analy- Fig. 1. Sand-shale sequence in a petroleum field, WY. U.S.A.
sis schemes. It is worth noting that almost no attention has (modified from Norris et aI. 1991).
been given to assess the effect of lithological heterogeneity
on the macro (overall) behavior of heterogeneous soil media.
The treatment of such heterogeneity has been exclusively
left to local experience and engineering judgment.
The main objective of this paper is to discuss the different
techniques developed to deal with soil heterogeneity in both
petroleum and geotechnical engineering and their applicabil-
ity to geotechnical field problems. In addition. attempts will
be made to identify the difficulties associated with obtaining
representative parameters that honor detailed ground hetero-
geneity. In the following sections. techniques developed in
the petroleum literature to deal with lithological heterogene-
ity are discussed. Then. different elements of inherent soil
variability will be presented along with their implications on
geotechnical field problems. such as settlement of shallow
foundation. liquefaction susceptibility. and seepage flow.
Limitations of the current practice will be addressed. and po-
tential trends for future studies will be suggested. Finally.
different decision algorithms will be discussed together with
examples of their applications in geotechnical analyses. tropy of the shale. i.e.. the ratio between the vertical to the
lateral extent of shale. The major advantage of this method
Lithological heterogeneity is its simplicity. while the main drawback is that the shale
blocks were assumed to be uncorrelated to each other.
The impact of lithological heterogeneity of the ground on In the renonnalization technique (King 1989). a simula-
the production of oil and gas reservoirs has been a major tion grid is generated across the analysis domain and a con-
area of study in petroleum engineering practice. Several up- stant value of soil penneability is assigned to each element
scaling techniques were developed to deal with complex of the simulation grid. Then. these elements are grouped into
ground profiles. such as the sand shale sequence shown in blocks of four and assigned an effective (equivalent) penne-
Fig. 1. The main aim of these techniques was to scale-up ability value. kef as illustrated in Fig. 2. This effective per-
fine scale penneability to coarser scales amenable to flow meability was obtained based on the analogy between water
simulation and engineering calculations. These averaging flow through soils of different penneabilities and electric
techniques can be classified into the following methods: current flow through a network of resistors. The above pro-
(1) Empirical techniques. such as the power averaging cedure can be applied to the new grid and repeated several
technique (Deutsch 1989). These are the simplest fonns of times depending on the scale of interest. This method was
up-scaling laws. originally developed for uncorrelated penneability fields. but
(2) Semi-empirical methods. such as the renonnalization it is also valid for correlated fields. It is worth noting that
(King 1989) and the representative elementary volume only isotropic media of equal penneabilities in vertical and
(REV) - renonnalization (Norris et al. 1991). They are more horizontal directions were considered during the develop-
sophisticated than the previous type but have limited theoret- ment of this technique. However. the method can be ex-
ical basis. tended to anisotropic media by applying the up-scaling
(3) Analytical techniques. such as that proposed by War- procedure to both the vertical and horizontal directions. In
ren and Price (1961). These methods are rather cumbersome spite of the theoretical basis implemented in this technique.
to implement in practice. it can be regarded as a relatively complicated method com-
The power averaging method was obtained through non- pared with the empirical fonnula presented in eq. [1].
linear regression of the results obtained from a three- The REV-renonnalization approach (Norris et al. 1991) to
dimensional numerical simulation of flow through sand- up-scale sand-shale fonnations for flow simulation com-
stone-shale fonnations. The analysis was carried out under bined the representative elementary volume (REV) theory
different target shale volumes. and the equivalent penneabil- with the renonnalization technique. The REV theory defines
ity was regarded as that of a homogeneous soil mass produc- a specific averaging volume at ~hich all microsc~pic ~aria
ing similar flow under the same head difference and tions are averaged out producmg a representatlve smgle
boundary conditions. This equivalent penneability. kef was macroscopic value. which is usually referred to as the repre-
found to satisfy the relation sentative elementary property (REP) (Norris et al. 1991).
The REV technique was originally developed to assess the
[1] ke =[Vsbk~ +(1- Vsb)~fCO representative property of porous materials. where the repre-
where ksh and kss are the penneabilities of the shale and sentative property at the smallest scale repres~nts th~ prop-
sandstone. respectively; Vsh is the volume fraction of shale; erty of either a void or a solid. By gradually mcreas~g the
and co is an averaging power. averaging volume. more voids and solids are included m the
The value of co was suggested to range from -1 to 1 de- averaging volume. resulting in fluctuation in the representa-
pending on the direction of flow and the geometrical aniso- tive property. as shown in Fig. 3. The REV can be defined as
Fig. 2. Schematic diagram of up-scaling using the renormalization technique (modified from King 1989).
Fig. 3. Representative elementary volume concept (modified Fig. 4. Fine tailings - sand mixture in a laboratory model of co-
from Norris et al. 1991). depositional mixed fine tailings embankment (modified from
Hutcheson 2000).
..
Range of REV
.
1" .'
REP
Volume scale
Fig. S. Detrending of CPT tip resistance data (a) identifying linear vertical trend, and (b) detrended data.
(a) -2 (b)
-- E
-2.5
-3 qo (z)
--
E
......, -2.
~~
~;
1
..
c -3.5 c -3. ~
I
0
-4
;
j .~.-4
~
w -4.5 w .1
-5.5
-5
0 1500 3000 4500 6000 7500 -3000 -2000 -1000 0 1000 2000 3000
qc(kPa) q (kPa)
0.2
aauro 0.4
0.2 8
.
I
I
I
I
I
I
0 O~--~--~--~----~--~--~
0 0.25 0.5 0.75 1 1.25 1.5 o 0.25 0.5 0.75 1.25 1.5
NormaIIzecIeeparatlon dI8w- (hili) Normalized eapar8IIon ~ (hili)
1.2 1.2
IIIl=VIIIIance sill =\8Itance
-=.--;;.:-::a----
1.0 ----------------~......_---- 1.0 - - - - - - - - - - - - - - - - - -'___
.
I
I I
I 0.95 I
0.8
I
I
0.8
I
I
I I
I I
SpI!ertcaI var!ogram mode!
I
... 0.8 ... 0.8 I
I
tiel_an var!ogram model
'[(Il) .. C I 1.5 {l\'8)-0.5 {I\'&11
0.4 0.4
'[(Il) .. C'I 1 -IIXP (~/a2) 1
0.2 8
0.2 8
O~--~--~----~--~--~--~
o 0.25 0.5 0.75 1 1.25 1.5 o 0.25 0.5 0.75 1 1.25 1.5
Normalized eeparatIon ca.r- (hili) NormalIzed aaparatIon dJ-.- (hili)
Spatial correlation structures are usually characterized by Limit of spatial continuity between field data
their model types and the limit of spatial correlation between The limit of spatial continuity is defined as the separation
field data. Spatial correlation models are parametric relation- distance between field data at which there is no, or insignifi-
ships used to curve fit the experimental variograms, or cant, spatial correlation. This limit can be expressed in terms
covariance functions, obtained from analysis of the field of the spatial range (Deutsch 2002), the scale of fluctuation
data. Deutsch (2002) has provided an excellent summary of (Vanmarcke 1977), or the autocorrelation distance (DeGroot
common variogram models used in practice. Examples of and Baecher 1993). The spatial range, a, can be defined as
these models (i.e., spherical, exponential, and Gaussian) are the separation distance at which the semivariogram reaches
shown in Fig. 6. These models help to determine the spatial the sill (variance) and correlation between data no longer ex-
correlation between field data at any separation distance and ists, as shown in Fig. 6. For vario8!'am models where the
in different directions. In addition, they can incorporate variogram is asymptotic to the sill J2), as in the case of ex-
other geological information such as the direction of maxi- ponential and Gaussian models, an effective range can be
mum continuity and maintain numerical stability of stochas- considered as the separation distance at which the variogram
tic simulation (Deutsch 2002). The limit of spatial continuity reaches a value equal to 0.95 times that of the sill. The scale
is discussed in more detail in the following section. of fluctuation, a,
estimates the distance within which soil
properties show relatively strong correlation and data be- Table 1. Comparison between different measures of the limit of
come either above or below the mean value. Vanmarcke spatial continuity between field data.
(1977) developed a Simplified procedure to estimate the
scale of fluctuation for different spatial correlation structure Type of correlation Spatial Scale of Autocorrelation
models. The autocorrelation distance. R. is the separation structure model range fluctuation distance
distance at which the covariance function decays to a value Exponential a 2aJ3 a/3
of (J/e. where e is the base of the natural logarithm. and cor- Gaussian a a O.58a
relation between data can be considered relatively weak. A Spherical a O.55a a
relationship between these different measures was developed
by Elkateb (2002). as shown in Table 1. The above expressions are based on the assumption that
the averaging process occurs in one direction only. These ex-
Volume-variance relationships pressions can be easily extended to the three-dimensional
The volume-variance relationships are analytical expres- case by assuming separable correlation structures
sions used to obtain the variance of spatial averages of field (Vanmarcke 1984). Such an assumption implies that the
data over certain volumes of interest. These spatial averages three-dimensional variance reduction factor could be ex-
usually have a narrower probability distribution function pressed as the product of its one-dimensional components in
than those associated with field data (Vanrnarcke 1977) and the form
consequently a smaller variance. The variance of these spa-
tial averages can be correlated to the point variance using [8] ry = rTJ'T,rTt
the variance reduction factor. r; .
as discussed by Vanmarcke where r Tx ' r Ty and r Tz are the one-dimensional variance re-
(1984) through duction factors in the x. y. and z directions. respectively.
[6] (Jr =r,p The variance reduction factor approaches 1 when the pa-
rameter T is small compared to the limit of spatial continu-
where (J is the standard deviation of field data (point statis- ity. For many geotechnical applications. the size of the
tics); (Jr is the standard deviation of the spatial average of averaging volume in the horizontal direction is usually small
the data over volume v; and r; is the variance reduction fac- compared to the spatial range of the horizontal correlation
tor. structure. As a result. it has been a common practice in
The variance reduction factor depends on the averaging many geotechnical implementations of the variance reduc-
volume. type of correlation structure. and the limit of spatial tion factor to assume that its value will be affected only by
correlation between field data. Several analytical expressions the size of the averaging volume in the vertical direction.
for the variance reduction factor were introduced by i.e.. layer thickness. This is because the variance reduction
Vanmarcke (1984). in the form factor in the horizontal direction can be reasonably assumed
to be equal to one.
[7a] rf = 2(;r(~ -1 +e-T/R ) for exponential In- a similar fashion. the variance reduction factor for
spherical correlation structures was developed in this study.
correlation structures as explained in Appendix A, and can be expressed in the
form
and
T T3
[7b] rJ = 2(; r[/1t ~ ;(~)-1 + e-
T/R
] for
[9] rJ =1- 2a + 20a 3
Spatial averages of random variables are spatially corre-
Gaussian correlation structures lated in a way similar to point (field data) statistics. This
where rf is the one-dimensional variance reduction factor; R correlation can be quantified in a pair-wise manner by as-
is the autocorrelation distance; T represents the size of the sessing the coefficient of correlation between any couple of
average volume; and ;(TIR) is the error function. which var- one-dimensional spatial averages. as shown in Fig. 7.
ies from 0 to 1 as T increases from 0 to 00. through the relationship (Vanrnarcke 1977)
[10]
_ D6r2(Do) - D6Ir2(DOl) - D&r2 (Dm) + D612r2(DoI2)
Pt2 - 2~n~)~n~)
where Pl2 is the correlation coefficient between the spatial Stochastic analysis techniques in
averages over the depths DI and D 2; r 2(D o). r 2(D ol ). geotechnical engineering
r2(Do~. and r2(Dol~ are the variance reduction factors over
averaging thickness equal to Do. DOl' DOl' and D OI2' respec- Stochastic analysis provides an excellent tool to account
tively; and r(D l ) and nD~. are the square roots of the vari- for the variability of soil properties and to develop rational
ance reduction factors over averaging thickness equal to Dl algorithms to estimate soil.desi~ parameters on a pro~abil
and D 2 respectively. istic basis where the asSOCIated nsk level can be quantified.
Several approaches have been adopted by geotechnical prac- Fig. 7. Schematic diagram illustrating different terms used to ob-
titioners to implement stochastic analyses in geotechnical tain correlation between spatial averages of random variables
field problems, such as liquefaction assessment, slope stabil- (modified from Vanmarcke 1977).
ity analysis, and foundation settlement. Examples of these u(z)
approaches are (i) application of reliability principles to
limit equilibrium analyses; (ii) stochastic finite element anal-
ysis; and (iii) application of stochastic input soil parameters
into deterministic numerical analysis. LAYER I {
A detailed discussion of the above approaches is provided
in the following sections together with examples of their ap-
plications to geotechnical field problems.
02
Application of reliability principles to limit equilibrium
analysis -- -
Statistical analysis of limit equilibrium problems was pri-
marily developed to perform probabilistic slope stability Depth z
analysis using different techniques, such as analytical ap-
proaches, approximate solutions, and Monte Carlo simula-
tion. Analytical approaches, such as those proposed by tween probability distributions and distributed vertical loads
Tobutt and Richards (1979), were primarily concerned with on a horizontal rigid beam resting on two pin supports (Harr
obtaining closed form solutions for the statistical properties (1987). For symmetrical probability distributions, p+ and p_
of earth slopes factors of safety. These solutions do not pro- are taken as equal to 0.5; while X+ and x_ are taken as one
vide information about the output probability distribution standard deviation above and below the mean value of the
and become cumbersome when considering different sources random variable, respectively. The mathematical details of
of uncertainty. this method are complicated enough to be beyond the scope
Approximate solutions of probabilistic slope stability of this paper and readers interested in these details can refer
analysis, such as the first order second moment (FOSM) and to Harr (1987). This technique is quite useful when it is dif-
the point estimate method (PEM), have been advocated by ficult or even impossible to apply the FOSM method as a re-
several authors, such as Christian et al. (1994). The basic sult of difficulties in obtaining derivatives of the factor of
idea of the FOSM method (Harr 1987) is to express the fac- safety with respect to different random variables. The main
tor of safety as a function of different random variables con- limitation of this technique is the complexity in calculations
sidered in the statistical analysis. This function is then when considering multiple random variables in the assess-
expanded about the mean values of these random variables ment of safety factors. An excellent summary of the accu-
using the Taylor expansion, retaining only linear (first order) racy and limitations of this method has been provided by
terms, where the mean and variance of the safety factor can Christian and Baecher (1999).
be assessed through The early implementation of Monte Carlo simulation to
limit equilibrium analyses considered soil or rock properties
[lla] E[F.S] =F(E[xtl, E[X2], ........ .E[xn )) as uncorrelated random variables (Kim and Major 1978).
Several realizations of soil design parameters were obtained
[llb]
i-n(d )2
Var[F.S] = L ...!.CJlC/ and used to develop a histogram for the factor of safety of
i=l dXj earth slopes. Recently, the effect of spatial correlation be-
tween soil properties has been accounted for through the ap-
plication of geostatistics principles and volume-variance
relationships (El-Ramly 2001). A complete probability dis-
tribution of output variables, such as the factor of safety, can
be obtained and the failure probability can be reasonably as-
where E[F.S] and Var[F.S] are the mean and variance of the sessed. This is a major advantage over other analysis tech-
factor of safety, respectively; C[Xi' x) is the covariance be- niques where some assumptions have to be made about the
tween the random variables Xi and Xj; and n is the number of probability distribution of output variables. It should be
random variables. noted, however, that Monte Carlo simulation has its own
The major advantage of this technique is its simplicity, es- limitations, which can be summarized as follows:
pecially when considering different sources of uncertainty, (1) The need to define a reliable input reference distribution,
as it provides a direct estimation of the mean and variance which requires a considerable number of field data. In addi-
for the factors of safety. However, the accuracy of this tech- tion, older versions of Monte Carlo simulation algorithms
nique is questionable, especially when dealing with highly used to deal only with parametric probability distribution
nonlinear relations and large soil variability, due to the trun- functions, i.e., probability distributions that can be defined
cation of high order terms in the Taylor expansion. through mathematical relationships such as normal and log-
In the PEM (Rosenblueth 1975 and 1981), the probability normal distribution. Field data, however, do not necessarily
distribution of each of the random variables is represented fit into any of these parametric distributions. This problem
by two points estimates X+ and X_ with probability densities has been overcome by recent versions of Monte Carlo simu-
of p+ and p_, respectively. This is based on the analogy be- lations, such as that of Deutsch and Journel (1998) that are
capable of dealing with nonparametric distribution functions Fig. 8. Clustering of the outcome of Monte Carlo simulations re-
directly inferred from field data; sulting from an insufficient number of realizations (modified
(2) Clustering of the simulation outcome into a limited zone from Palisade Corporation 1996).
of the input probability distribution, as the drawn samples 1~----------------------~~--~
are more likely to be in areas of higher probability, as shown c
0
in Fig. 8. This problem mainly arises in cases where an in- 1:;
sufficient number of realizations (number of iterations in
Monte Carlo algorithm) are used in the simulation process .=cc 0.8
~
(Palisade Corporation 1996). This may result in sampling 0
~
values of the random variable away from the tails of the in- ;, 0.6
.2
put probability distribution, which can be on the unsafe "C
(nonconservative) side. This problem, however. can be over- -;
come by using a number of realizations large enough to re- ;:; 0.4
produce the input distribution; and ~
(3) Depending on the number of variables involved in the 11
simulation process, Monte Carlo simulation may require a :; 0.2
significantly large number of iterations and consequently a E
;,
considerable computational effort. However, the authors be- u 0
lieve that this problem has been overcome by the new gener- -3 -2 -1 0 1 2 3
ation of fast computers.
Random variable
Stochastic finite element method
The stochastic finite element method (SFEM) is a modifi- comes quite tedious for irregular element shape and
cation of the traditional fmite element method to capture the complicated boundary conditions;
effect of soil spatial variability on numerical analysis. This (3) It is limited to small variability due to the error associ-
is carried out by using finite element discretization to obtain ated with the truncation of higher order terms in the Taylor
a direct assessment of the mean and variance of nodal dis- expansion (which is used for the determination of mean val-
placements together with the covariance between displace- ues of the response (output) variables, such as nodal settle-
ments at different nodes of the numerical analysis mesh ments);
(Baecher and Ingra 1981). This assessment is usually ac- (4) Integration of the random variable field over each ele-
complished by calculating a covariance matrix whose value ment may result in a change in the anisotropy ratio of the
depends on the characteristics of spatial correlation between spatial correlation structure of soil properties; and
soil properties, such as variogram model and spatial range. (5) It is usually limited to the linear elastic behavior of soil
These characteristics are captured into the finite element to avoid extreme complexity in the computation process and
scheme by introducing the matrix of differentials thereby as- does not adequately capture the behavior of soil properties
sessing the effect of the variation of mechanical soil proper- with skewed probability distributions.
ties from one element to another on the global stiffness Due to the above limitations, the use of stochastic finite
matrix. For more details about SFEM, the reader can refer to element analysis has received limited attention from
Baecher and Ingra (1981) and Auvinet et al. (1996). Differ- geotechnical practitioners and researchers.
ent modifications of SFEM have been developed by intro-
ducing different numerical techniques to capture soil spatial Application of stochastic input parameters into
variability. Examples of these modifications are the probabil- deterministic numerical analysis
istic finite element method (Righetti and Harrop-Williams Deterministic numerical analysis with stochastic input soil
1988) and stochastic integral formulations (Zeitoun and parameters has been recently adopted by many researchers,
Baker 1992). such as Paice et al. (1994) and Popescu et aI. (1998), as a
The major advantage of the SFEM is the direct assessment technique to incorporate soil spatial variability in
of statistical characteristics of output variables, such as the geotechnical design. Monte Carlo based simulation tech-
mean and variance. This helps avoid long computational time niques have been used to generate several realizations of soil
associated with incorporating several realizations of spatially properties that vary from one point to another across the do-
variable soil parameters into deterministic analysis scheme, as main of interest, as shown in Fig. 9. This spatial variation is
discussed in the following section. On the other hand, differ- usually employed into the numerical analysis scheme by as-
ent limitations of the SFEM have been discussed by several sessing soil properties at the center of each element of the nu-
authors, such as Baecher and Ingra (1981) and Auvinet et al. merical simulation grid and assuming them to be constant
(1996), and they can be summarized as follows: within that element By analyzing several realizations of the
(1) The analysis results are not affected by the probability spatially variable soil medium, histograms of response (out-
distribution of the input random variables. Furthermore, a put) variables can be obtained. Examples of the simulation al-
distribution has to be assumed for output variables as SFEM gorithms used in practice are the sequential Gaussian, the
provides only an assessment of the mean and standard devia- sequential indicator simulations (Deutsch 2002), and the local
tion; average subdivision teclmique (Fenton and Vanmarcke 1990).
(2) Element variance and covariance matrices are functions The sequential Gaussian simulation (SGS) is the most
of element shape and geometry and their determination be- commonly used technique, especially in the field of petro-
Fig. 9. Defonned mesh with spatially variable elastic modulus below flexible strip footing (modified from Paice et aI. 1994).
15.00
y
,
(m)
Darker gray areas signifies
zones of lower elastic modulus
12.00
9.00
6 . 00
3.00
0.00 x(m)
throughout the history of geotechnical engineering practice about the probability distribution of output variables, and the
to assess the impact of ground variability on geotechnical effect of different spatial correlation structures was not ac-
field problems. In the following sections, an attempt is made counted for. Finally, the use of SFEM implied some restric-
to address the current state of practice in some of these ap- tions on the range of soil variability used in the analysis to
plications and its limitations together with potential trends prevent development of plastic zones as discussed in the
for future research. previous paragraph.
The effect of random soil stiffness on foundation settle-
Stochastic analysis of shallow foundation settlement ment was reinvestigated by Paice et al. (1994) through the
Early attempts to perform probabilistic analysis of foun- use of deterministic finite element analysis with stochastic
dation settlement started in the late 1960s. Wu and Kraft input soil parameters. Soil elastic modulus was regarded as a
(1967) estimated the uncertainty in soil bearing capacity and spatially random variable resulting in a ground profile as
foundation settlement through assessing the uncertainty in previously shown in Fig. 9. The elastic modulus was as-
applied load, soil strength, and deformation parameters. The sumed to follow a lognormal probability distribution and ex-
uncertainty in soil strength was estimated by assessing the ponential correlation structure. This study has some limiting
variability of laboratory undrained shear strength for clayey assumptions such as the linear elastic soil behavior, the iso-
soils and that of SPT data for sandy soils. Resendiz and tropic correlation structure, and the symmetry of spatial dis-
Herrera (1969) carried out a probabilistic analysis of settle- tribution of elastic modulus around the footing centerline. In
ment and rotation of flexible and rigid footings over ran- addition, the effect of different types of correlation struc-
domly variable compressible soils. A one-dimensional tures and the sensitivity of the results to the number of real-
settlement model was adopted in which the coefficient of izations were not considered.
volume change was characterized as a normally distributed The effect of random fluctuations of the interface between
random variable. The analysis results were used to obtain soil layers on the uncertainty in foundation settlements has
design parameters that satisfied tolerable settlements and r0- been accounted for by Brzakala and Pula (1996). This uncer-
tations criteria together with the minimum expected mone- tainty in soil geometry was converted into a new random
tary loss. These studies can be considered as a good start field expressed in terms of the interface fluctuation and was
towards addressing such a complex problem. However, they incorporated into the stochastic finite element analysis. The
were fairly primitive as some elements of soil inherent vari- main limitations of the study lie in the linear elastic soil be-
ability, such as spatial correlation between soil properties, havior and the neglecting of the inherent soil variability
were not adequately considered. within layers. In addition, the effect of different types of
The modem approach to deal with ground variability and probability distributions and correlation structures on the
its implications on foundation settlement started in the early stochastic analyses outcomes was not accounted for.
1980s with the pioneer work of Baecher and Ingra (1981). In It is worth noting that the above studies have not consid-
their study, two-dimensional stochastic finite element analy- ered the effect of changing the state of stresses in soil mass
ses were carried out to assess the uncertainty in total and dif- on the outcome of statistical analyses. In other words, the
ferential settlement. Soil elastic modulus was treated as a sensitivity of the statistical characteristics of output variables
random variable, whereas Poisson's ratio was assumed to be to wide ranges of applied vertical and horizontal stresses
constant across the soil mass. 1\vo spatial correlation mod- was not adequately addressed. In addition, no attempt has
els, the exponential and the squared exponential (Gaussian), been made to provide risk-based representative soil parame-
were considered and the response variables (total and differ- ters that can be implemented in deterministic analyses while
ential settlement) were assumed to be normally distributed. continuing to honor detailed ground heterogeneity.
A limiting assumption of the study was the linear elastic soil
behavior, which implies, together with the use of SFEM, a Stochastic analysis of liquefaction problems
small variation in soil properties to avoid the development of The early attempts to quantify the stochastic nature of liq-
plastic zones and the onset of nonlinear constitutive behav- uefaction problems were focused on developing analytical
ior. In addition, the effect of different probability distribu- expressions to estimate the uncertainty in liquefaction poten-
tions of soil properties on the expected uncertainty was not tial assessment. Yegian and Whitman (1978) conducted a pi-
assessed. oneer study to provide a statistical evaluation of the annual
In a similar fashion, Zeitoun and Baker (1992) proposed a probability of failure for potentially liquefiable sites. This
stochastic approach for settlement prediction of shallow was ciuried out by combining the annual probability of
foundations using the stochastic integral formulation (SIF) given earthquakes with the probability of ground failure un-
technique, which is a modification of the SFEM. It was as- der these earthquakes. In addition, an analytical expression
sumed that soil would exhibit linear elastic behavior under was developed to assess the uncertainty in a limit state pa-
both axisymmetrical and plane strain conditions. Soil shear rameter, proposed to estimate the maximum shear resistance
modulus was treated as a random variable and the Gaussian of the ground. A major limitation of that study was that the
model was chosen to represent the spatial variation of shear effect of spatial correlation between soil properties was not
modulus across the problem domain. The technique used in accounted for and that the uncertainty in the results were as-
the study had serious limitations as unrealistic spatial corre- sumed to be insensitive to the probability distribution of the
lations were assumed either through the use of a very high input random variables. Furthermore, the derivation of the
horizontal autocorrelation distance or by considering the soil expression for the limit state parameter was based on the as-
medium to be in the form of concentric rings of constant sumption that the soil shear resistance and vertical effective
elastic modulus. Furthermore, no information was provided stress were two independent random variables.
Recently. the use of deterministic finite element analysis deterministic finite element analysis was carried out. using
with stochastic input soil parameters has gained much different percentiles of the recorded CPT data. to estimate
popularity in the field of probabilistic liquefaction analysis. equivalent (representative) soil parameters that can capture
Several attempts have been made to apply this technique to the implications of soil spatial variability. These parameters
study case histories that involved potentially liquefiable were considered to be associated with th~ upper limit of the
ground conditions. Monte Carlo simulation response range. It was concluded
Fenton and Vanmarcke (1991) performed one-dimensional that the initiation of soil liquefaction could not be accurately
finite element analyses to assess the effect of inherent vari- predicted by deterministic models. which could not account
ability of soil properties on liquefaction potential at the for the presence of loose pockets within soil masses. This
Wildlife site. California. Soil properties. such as porosity. study. as in the case of the previous one. did not take into
Poisson's ratio. elastic modulus. permeability. and the dila- consideration the sensitivity of the analysis to the number of
tion angle. were considered as random variables. The first realizations of soil properties. In addition. the proposed
two properties were considered to be normally distributed. equivalent parameters could be considered overconservative.
whereas the rest were assumed to follow a lognormal distri- as they were associated with the most critical response of
bution. The effect of correlation structure was taken into the stochastic analysis. Moreover. the effect of the type of
consideration through the application of the variance reduc- correlation structure and its spatial limit was not accounted
tion factor proposed by Vanmarcke (1977). Several realiza- for.
tions of soil properties were generated using the local Once again. the effect of changing the state of stresses in
average subdivision technique. and were excited by various soil mass on the outcome of stochastic analyses has not been
earthquake motions applied at the base of the soil columns adequately considered in any of the above studies. For exam-
using DYNAID software. A main limitation of the study is ple. more investigation is needed to ascertain whether or not
that the results sensitivity to the number of realizations was the same values of the representative cone tip resistance per-
not taken into consideration. Moreover. the effect of differ- centiles of Popescu et al. (1998) would be obtained if poten-
ent probability distributions of soil properties was not ac- tially liquefiable layers were at different depths below
counted for together with the use of one-dimensional ground surface.
analysis. in which the analysis domain was divided into soil
columns neglecting the coupling between soil elements. Stochastic analyses of seepage flow
Popescu et al. (1996) carried out one of the pioneer inves- The problem of water flow through heterogeneous porous
tigations on the effect of soil spatial variability on liquefac- media has been studied thoroughly along the history of pe-
tion assessment using the results of cone penetration tests. troleum engineering and water resources research. One of
where the cone tip resistance. qe' and the cone index. Ie. the pioneer attempts to apply the principles of geostatistics
were treated as random variables. A simulation algorithm into the geotechnical engineering practice to study the effect
was developed in the study for the simulation of non- of soil spatial variability on seepage flow was made by
Gaussian multivariate random fields. A nonlinear regression Griffiths and Fenton (1993). In their study. the local average
algorithm was adopted to determine the probability distribu- subdivision simulation technique was used to generate lOOO
tion and the correlation structure of the random variables. realizations of spatially variable hydraulic conductivity be-
The problem was analysed using the DYNAFLOW software Iowa water retaining structure. The resulting field was then
implementing stochastic input soil parameters obtained us- mapped onto a finite element mesh to perform numerical
ing the correlations between soil properties and CPT data. analysis of the problem under deterministic boundary condi-
For comparison. a deterministic numerical analysis was car- tions. The hydraulic conductivity. k. was assumed to follow
ried out using the mean values of soil parameters. A charac- lognormal probability distribution. and the effect of spatial
teristic percentile of cone tip resistance was proposed for use correlation structure was accounted for through quantifying
in deterministic analyses to predict the same maximum pore the influence of the scale of fluctuation on different response
pressure obtained from stochastic analysis. The major limita- (output) variables. The limitations of the study lie in the use
tion of the study lies in the use of only four realizations to of an isotropic correlation structure where vertical and hori-
quantify the effect of soil variability. which may not be suffi- zontal ranges were assumed equal. and the effect of different
cient to sample the expected range of response. as discussed probability distributions and correlation structure models
earlier. In addition. the effect of spatial correlation range on were not accounted for. Furthermore. the sensitivity of the
the study outcomes was not accounted for. The authors be- analysis to the number of realizations of the random vari-
lieve. however. that this spatial range may have a profound able. hydraulic conductivity. was not considered.
effect on liquefaction susceptibility and needs to be consid-
ered in future stochastic liquefaction studies. In addition. the Decision making in geotechnical
strength percentile proposed for use in deterministic analysis engineering
to capture the effect of soil spatial variability was subjec-
tively assessed. One of the major challenges that faces geotechnical engi-
Popescu et al. (1998) extended the previous study to pro- neers is the need to make decisions regarding the soil param-
vide risk-based liquefaction potential measures through two- eter to be used in engineering analysis. These decisions have
dimensional stochastic analysis where cone tip resistance to be based on information that invariably has a certain de-
was treated as a spatially random variable. The effect of in- gree of uncertainty. Consequently. the decision making pro-
herent variability was assessed through 25 realizations of the cess is considered to be governed by two factors. the
spatial distribution of CPT data across the site. A series of uncertainty in the decision variables and the risk level of the
project. Several decision making algorithms have been used Table 2. Assessment of performance of earth slopes and the as-
throughout the history of geotechnical engineering practice, sociated failure probability as proposed by the US Army Corps
such as the worst case and quasi worst case approaches, reli- of Engineers (1995).
ability-based techniques, confidence interval approach, and
Bayesian decision analyses. Details of these algorithms will Expected level of Probability of
be discussed in the following paragraphs. performance ~ failure (%)
The worst case approach aims at achieving the absolute High 5 3xlO-s
safety of the project and relies on the notion of maximum Good 4 0.003
loss and maximum expected hazards, often referred to as the Above average 3 0.1
maxi-max criterion (Ang and Tang 1984). For example, if Below average 2.5 0.6
the range of the measured friction angle of a sandy deposit Poor 2 2.3
at a certain site ranges from 30-40, the design value will be Unsatisfactory 1.5 7
assessed as 30. This approach is overly conservative and Hazardous 1 16
rarely used in practice. On the other hand, the quasi worst
case approach (Pate-Cornell 1987) tries to apply some kind
of engineering judgment into the above approach to provide Fig. 11. Critical probabilities of failure for dam design in terms
an upper bound for the risk level. Revisiting the above ex- of expected number of fatalities (modified from Whitman 2000).
ample, the sandy soil at the site is classified (say medium ANCOLD, Australian National Committee on Large Dams;
dense sand) and the minimum value associated with such PMFJMDE, XXXXXXXXXXXX.
classification (say 33) will be used as the design value. A
common problem of the two approaches is that no informa-
tion can be obtained about the risk level associated with the
design value.
The reliability-based approach relies on selecting design
parameters that satisfy a desired degree of reliability or a
certain probability of failure. This approach has been com-
monly used in slope stability analysis. Wolff (1996) pro- ~
posed soil design parameters to be associated with a
reliability index, ~, of 3 for routine slopes and 4 for critical
slopes such as dams. The reliability index can be obtained
J 10-~~----~------~~-----r------~
'0
through ~
[14] ~ =mFS - L
1 1o'st-----+-.I!I,,-~~I\o.:~i,.._
(JFS iii
E SAFE
where mFS is the mean factor of safety; L is a limit state ~
value usually equal to 1; and (JFS is the standard deviation of
the factor of safety. 10" t------+------'~-It---~--
In a similar fashion, the US Army Corps of Engineers
(1995) proposed an assessment of the performance level of
embankments depending on the target reliability index and 107------__ ________ ____________
~ ~
Fig. 12. Selection of design parameters associated with a 90% Fig. 13. Linear loss functions to quantify the effect of making
confidence level. mistakes in estimating soil design parameters.
Loa Function
b
0.1
-4 -2 o 2 4
Parameter value Underestimation
selecting design parameters. These functions can take differ- discussed together with comments regarding their applicabil-
ent fonns, such as linear, quadratic, and exponential, as ity in geotechnical engineering.
shown in Fig. 13 for linear loss functions. For more details From this study it can be concluded that there is a need
of the application of this approach, the reader can refer to for a comprehensive study of soil heterogeneity that takes
Deutsch (2002). The main limitation of this approach is the into consideration different sources of nonhomogeneity and
difficulty associated with its application in cases where loss their implications on different geotechnical applications.
of human lives may be expected. Furthennore, there is a need to ascertain whether or not the
Throughout the history of geotechnical engineering few outcomes of stochastic geotechnical analyses are sensitive to
attempts have been made to implement the above approaches changes in the state of stresses in subsurface layers. Finally,
into field problems. One of the pioneer works in this area the risk level of geotechnical projects should be incorporated
was that by Folayan et a1. (1970), where the Bayesian deci- in a decision-making framework to provide estimates of rep-
sion analysis was applied to settlement prediction analysis. resentative soil parameters that honor the detailed ground
In their study, the compression index, Ce, was treated as a heterogeneity.
random variable and the results of 27 one-dimensional con-
solidation tests were used to obtain a histogram for Ce. In
addition, an exponential loss function was adopted to assess
a value of Ce that produced the minimum expected loss. A References
main limitation of the study was that it ignored the effect of
Ang, A.H.S., and Tang, W.H. 1984. Probability concepts in engi-
the spatial correlation characteristics of Ce on the analysis
neering planning and design. Vol. 2: Decision, risk, and reliabil-
results.
ity. John Wiley & Sons Inc., New York, N.Y.
Auvinet, G., Bouayed, A., Orlandi, S., and Lopez, A. 1996. Sto-
Conclusions chastic finite element method in geomechanics. In Proceeding of
the 1996 Conference on Uncertainty in the Geologic Environ-
A thorough review of the different techniques developed
ment, Uncertainty 96, Vol. 2, Madison, Wis. pp. 1239-1253.
to deal with soil heterogeneity has been presented in this
Baecher, G. 1987. Statistical analysis of geotechnical data. Con-
study. Different approaches developed to handle the tract Report GL-87-1. Prepared for Department of the Army,
lithological heterogeneity of the ground in the petroleum en- US Army Corps of Engineers, Washington, D.C.
gineering field were briefly addressed. Different elements of Baecher, G.B., and Ingra, T.S. 1981. Stochastic FEM in settlement
soil inherent spatial variability, such as mean, variance, spa- predictions. Journal of the Geotechnical Engineering Division,
tial correlation characteristics, and volume-variance relation- ASCE, 107(GT4): 449-463.
ships were thoroughly discussed together with their Benjamin, J.R., and Cornell, C.A. 1970. Probability, statistics, and
implications in geotechnical design. Different approaches decision for civil engineers. McGraw-Hili Book Company, New
adopted throughout the history of geotechnical engineering York, N.Y.
to perfonn stochastic analyses of different geotechnical ap- Brzakala, W., and Pula, W. 1996. A probabilistic analysis of foun-
plications were thoroughly reviewed and criticized. In addi- dation settlement. Computers and Geotechnics, 18(4): 291-309.
tion, an expression for the variance reduction factor of Christian, J.T., and Baecher, G.B. 1999. Point estimate method as
spherical spatial correlation structures (variograms) was de- numerical quadrature. Journal of the Geotechnical and
veloped in this study. Examples of the applications of sto- Geoenvironmental Engineering Division, ASCE, 125(GT9):
chastic analysis to field problems such as shallow foundation 779-786.
settlement, liquefaction assessment, and seepage flow were Christian, J.T., Ladd, C.C., and Baecher, G.B. 1994. Reliability and
presented with emphasis on the limitations of the current probability in stability analysis. Journal of the Geotechnical En-
practice. Finally, different decision making algorithms were gineering Division, ASCE, 12O(GTI): 1071-1111.
DeGroot. OJ. and Baecher. G.B. 1993. Estimating autocovariance tion of Statistics and Probability in Soil and Structural
of in-situ soil properties. Journal of the Geotechnical Engi- Engineering. Vancouver. B.C. Vol. I. pp. 538-574.
neering Division. ASCE. 119(GTI): 147-166. Phoon, K.-K . and Kulhawy. EH. 1999. Characterization of
Deutsch. C. 1989. Calculating effective absolute penneability in geotechnical variability. Canadian Geotechnical Journal. 36(4):
sandstone/shale sequences. SPE Formation Evaluation. 4: 343- 612-624.
348. Popescu. R.. Prevost. J.H . and Deodatis. G. 1996. Influence of
Deutsch. C.V. 2002. Geostatistical reservoir modeling. Oxford Uni- spatial variability of soil properties on seismically induced liq-
versity Press. N.Y. uefaction.ln Proceeding of the 1996 Conference on Uncertainty
Deutsch. C.V. and Iournel. A.G. 1998. GSLIB geostatistical soft- in the Geologic Environment, Uncenainty 96. Pan 2 (of 2).
ware library. Oxford University Press. Oxford, New York. Madison. Wis. pp. 1098-1112.
Elkateb. T. 2002. Quantification of soil heterogeneity. Ph.D. thesis, Popescu. R . Prevost. I. H. and Deodatis. G. 1998. Characteristic
University of Alberta. Edmonton. Alta. (In progress.) percentile of soil strength for dynamic analysis. In Proceeding
EI-Ramly. H. 2001. Probabilistic and quantitative risk analysis for of the 1998 Conference on Geotechnical Earthquake Engi-
earth slopes. Ph.D. thesis. University of Alberta. Edmonton. neering and Soil Dynamics III. Pan 2 (of 2). Seattle. Wash.
Alta. pp. 1461-1471.
Fenton. G.A . and Vanmarcke. E. 1990. Simulation of random Resendiz. D., and Herrera, I. 1969. A probabilistic fonnulation of
fields via local average subdivision. Journal of Engineering Me- settlement control design. In Proceedings of the 6th Interna-
chanics. 116(8): 1733-1749. tional Conference on Soil Mechanics and Foundation Engi-
Fenton. G.A.. and Vanmarcke, E.H. 1991. Spatial variation in liq- neering. Mexico City. Mexico. pp. 217-225.
uefaction risk assessment. In Proceedings of the Geotechnical Righetti. G. and Harrop-Williams. K. 1988. Finite element analy-
Engineering Congress. Boulder. Colo. Geotechnical Special sis for random soil media. Journal of the Geotechnical Engi-
Publications. No. 27. Vol. 1. pp. 594-607. neering Division. ASCE 114(GTl): 59-75.
Folayan. J.I.. Hoeg. K. and Benjamin. J.R. 1970. Decision theory Robinsky. E.I. 1999. Thickened tailings disposal in the mining in-
applied to settlement predictions. Journal of the Soil Mechanics dustry. E.!. Robinsky Associates. Toronto. Ont.
and Foundation Division. ASCE. 96(SM4): 1127-1141.
Rosenblueth. E. 1975. Point estimate for probability moments. In
Griffiths. D.V. and Fenton. A. 1993. Seepage beneath water retain- Proceedings of the National Academy of Sciences of the United
ing structures founded on spatially random soil. Geotechnique. States of America, 72(10): 3812-3814.
43(4): 577-587.
Rosenblueth. E. 1981. Two-point estimates in probabilities. Ap-
Harr. M.E. 1987. Reliability-based design in civil engineering.
plied Mathematical Modeling. 5: 329-335.
McGraw-Hill Book Company. New York, N.Y.
Schultze. E. 1975. Some aspects concerning the application of sta-
Hutcheson. H. 2000. Depositional and geotechnical characteristics
tistics and probability to foundation structures. In Proceeding of
of mineral sands thickened/paste tailings. In Proceedings of the
the 2nd International Conference on the Applications of Statis-
Transponation and Deposition of ThickenedlPaste Tailings
tics and Probability in Soil and Structure Engineering. Aachen.
Learning Seminar. Edmonton. Alta.
Gennany. pp. 457-494.
Journel. A.G . and Huijbregts. C.J. 1978. Mining geostatistics. Ac-
Tobutt, D.C . and Richards. E. 1979. The reliability of earth slopes.
ademic Press. New York. N.Y.
International Journal for Numerical and Analytical Methods in
Kim. H. and Major. G. 1978. Application of Monte Carlo tech-
Geomechanics. 3: 323-354.
niques to slope stability analysis. In Proceedings of the 19th
US Anny Corps of Engineers. 1995. Introduction to probability
U.S. Symposium on Rock Mechanics. Nev. pp. 28-39.
and reliability methods for use in geotechnical engineering. En-
King. P.R. 1989. The use of renonnalization for calculating effec-
gineering Technical Letter No.llI0-2-547. Washington D.C.
tive permeability. Transpon in Porous Media, 4: 37-58.
Lumb, P. 1970. Safety factors and the probability distribution of Vanmarcke. E. 1977. Probabilistic modeling of soil profiles. Jour-
soil strength. Canadian Geotechnical Journal, 7(3): 225-242. nal of the Geotechnical Engineering Division. ASCE.
Morgenstern. N .R. 2000. Perfonnance in geotechnical practice. l03(GTll): 1227-1245.
The inaugural Lumb lecture. Hong Kong Institution of Engi- Vanmarcke, E.H. 1984. Random fields. analysis and synthesis.
neers. May 2000. 59 pp. MIT Press. Cambridge, MA.
Neter. 1., Kutner. M.H . Nachtsheim. C.I . and Wassennan. W. Warren, 1.E., and Price, H.S. 1961. Flow in heterogeneous porous
1996. Applied linear statistical models. McGraw-Hill Book media. Society of Petroleum Engineering Journal, XX: 153-169.
Company. New York. N.Y. Whitman, R.V. 2000. Organizing and evaluating uncenainty in
Norris. R.I . Lewis. J.M . and Heriot-Watt. U. 1991. The geological geotechnical engineering. Journal of the Geotechnical Engi-
modeling of effective penneability in complex heterolithic fa- neering Division, ASCE, 126(GT7): 583-593.
cies. In Proceedings of the 66th Annual Technical Conference Wolff, T.F. 1996. Probabilistic slope stability in theory and prac-
and Exhibition. SPE 22692. Dallas. Tex. Vol. W. pp. 359-374. tice. In Proceedings of the 1996 Conference on Uncenainty in
Paice. G.M . Griffiths. D.V. and Fenton. G.A. 1994. Influence of the Geologic Environment, Uncenainty 96, Pan I (of 2), Madi-
spatially random soil stiffness on foundation settlement. In Pro- son'-Wis. pp. 419-433.
ceedings of the Conference on Venical and Horizontal Deforma- Wu, T.H., and Kraft, L.M. 1967. The probability of foundation
tion of Foundations and Embankments. Pan I (of 2). College safety. Journal of the Soil Mechanics and Foundation Division,
Station, Tex. pp. 62&-639. ASCE. 93(SM5): 213-231.
Palisade Corporation. 1996. @Risk: Risk analysis and simulation Yegian, M.K., and Whitman, R.V. 1978. Risk analysis for ground
add-in for Microsoft Excel or Lotus 1-2-3. Palisade Corpora- failure by liquefaction. Journal of the Geotechnical Engineering
tion. N.Y. Division, ASCE, 104(GT7): 921-937.
Pate-Cornell, M.E. 1987. Risk uncertainties in safety decisions. Zeitoun, D.G., and Baker, R. 1992. A stochastic approach for set-
Reliability and risk analysis in civil engineering. In Proceedings tlement predictions of shallow foundations. Geotechnique.
of the ICASP5. the 5th International Conference on the Applica- 42(4): 617-629.
Appendix A
[A2] C(h)=1+0.S(;r -1.5(;)
Derivation of variance reduction factor for spherical
correlation structure Substituting in eq. [AI] provides
The variance reduction factor, r2, can be detennined ac-
cording to the following relation proposed by Vanmarcke
(1984): [A3) ri 1(1- ~ll+O.s(;)' -1.5(;)]dh
=:
[AI] 2T(1--Th)c(h)dh
rf =-J
To Integrating and rearranging results provides the following
expression for the variance reduction factor for spherical
where h is the separation distance and C(h) is the standard correlation structures:
covariance, i.e., covariance with a unit variance.
The standard covariance of spherical correlation structure [A4] rf = 1 -T - +T3
-- 3
can be expressed in the fonn 2a 20a