See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/312328352

Reliability measures for consolidation settlement by means of CPT data

Conference Paper · September 2013

CITATIONS READS

0 319

2 authors:

Pouya Pishgah Reza Jamshidi Chenari

University of Guilan University of Guilan
4 PUBLICATIONS   12 CITATIONS    168 PUBLICATIONS   557 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Evaluation of earth pressures using FELA method View project

Reliability assessment of slopes View project

All content following this page was uploaded by Reza Jamshidi Chenari on 14 January 2017.

The user has requested enhancement of the downloaded file.

Reliability measures for consolidation
settlement by means of CPT data
Pouya Pishgah
SNC-Lavalin, Toronto, Ontario, Canada
Reza Jamshidi Chenari
The University of Guilan, Rasht, Guilan, Iran

ABSTRACT
A methodology is introduced for reliability calculation of consolidation settlement based on cone penetration test (CPT)
data. The measured data of corrected cone tip resistance ( ) is detrended using a quadratic trend and the residuals are
assumed to be lognormally distributed random field. Realizations of is generated by using statistical parameters of
residuals including standard deviation and the scale of fluctuation. The quadratic trend and the generated residuals are
then combined to correlate shear and bulk moduli as input consolidation properties for coupled analysis and
subsequently consolidation settlement was calculated by using finite difference method adopted in Monte Carlo
simulations. The results of reliability analysis are presented describing the range of possible settlements by considering
characteristics of uncertainties involved at the particular site. Maximum possible settlements must be smaller than the
allowable settlement to guarantee proper performance.

RÉSUMÉ
Une méthodologie est introduit pour le calcul de la fiabilité d'établissement de consolidation sur la base de données de
test de pénétration de cône (CPT). Les données mesurées corrigées résistance à la pointe du cône ( ) est redressée à
l'aide d'une tendance quadratique et les résidus sont supposés être distribués lognormale champ aléatoire. Réalisations
de est généré en utilisant des paramètres statistiques des résidus dont l'écart-type et l'ampleur des fluctuations. La
tendance quadratique et les résidus générés sont ensuite combinées pour corréler les modules de cisaillement et en
vrac en tant que propriétés de consolidation d'entrée pour l'analyse couplée puis tassement de consolidation a été
calculé en utilisant la méthode des différences finies adoptée en simulations de Monte Carlo. Les résultats de l'analyse
de fiabilité sont présentés pour décrire la gamme des implantations possibles en tenant compte des caractéristiques de
incertitudes liées à un site particulier. Colonies maximales possibles doivent être plus petit que le règlement admissible
pour garantir un bon fonctionnement.

1 Introduction be more work on how much effort is required to improve

the estimates for a particular site. However calculating the
The importance of variability analysis is increasingly actual variability of soil properties is one of the unresolved
recognized in geotechnical engineering as reliability problems in geotechnical risk and reliability (Christian and
based design (RBD) methods are assuming a prominent Baecher 2011).
role for the calibration of new design codes. However, in Among in situ testing methods, cone penetration
geotechnical practice, there has been rather slow (CPT) measurements are ideal for assessing soil
progress in employing RBD, compared to other variability because a large volume of near continuous
engineering disciplines. A primary reason for this slow data can be collected in a cost-effective way, the test has
progress is difficulty in estimating the variability of the good repeatability, the equipment is highly standardized,
design properties of geo-material, which is essential for and the procedure is well-defined and almost operator-
any RBD procedure. A fundamental problem is the lack of independent. Moreover, CPT has strong theoretical basis
data. There is a tradition of basing conclusions in for interpretation and numerous semi-empirical
laboratory testing programs on small numbers of tests. It correlations have been developed to estimate
is quite common to encounter laboratory programs geotechnical parameters from the CPT for a wide range of
consisting of three or four consolidation tests. Therefore soils (e.g, Robertson 2009). A number of researches on
there are seldom enough data to support broad the application of CPT measurement in reliability analysis
conclusions about the statistics of soil properties. for different geotechnical problems have recently been
Several studies have been published on the published (e.g., Uzielli 2004 ; Haldar and Babu 2008 ;
variability of the properties of soils (e.g., Phoon and Doherty and Gavin 2010 ; Kenarsari et al. 2011 and
Kulhawy 1995). Sometimes the values in these 2012).
publications are simply adopted into reliability calculations In this article, a methodology is introduced for
without further efforts to establish the variability for the reliability calculation of consolidation settlement based on
particular project at hand. Despite the excellent work that statistical review of CPT data. This general approach can
has been done on this issue, it is not a closed matter, and be an effective technique to predict consolidation
more work needs to be done. In particular there needs to settlement using probabilistic analyses considering
uncertainties involved at a specific site. A deterministic In soft clays and silts the measured must be corrected
analysis based on simple uncoupled (Terzaghi) approach for pore water pressures acting on the cone geometry,
is also conducted to compare with proposed reliability thus obtaining the corrected cone resistance, :
method.
1 [1]

2 CPT data and statistics Where “a” is the net area ratio determined from laboratory
calibration and in this case this value is 0.75.The data set
In present study, a typical CPT sounding from the consists of 368 measurements of corrected cone tip
Adapazari area, Turkey was selected for stochastic resistance, , digitized at a 0.02 m interval
analyses. To exclude lithological heterogeneity a clay - To guarantee stationary, a quadratic trend was
silty clay layer with 7.36 meter thickness was identified sought to remove from original data set. The original
from the CPT profile for variability analysis by adopting measured data, quadratic trend and the residual
the classification charts proposed by Robertson (1990), component are illustrated in Figure 3. Pishgah and
as illustrated in Figures 1 to 2. Jamshidi (2011 a,b) have shown that the deterministic
component of the inherent variability plays a prominent
role in consolidation behaviour of natural alluvial deposits.
The stochastic properties of the residual component can
be quantified thorough calculation of the standard
deviation and the correlation length.

qt (MPa)

-3 -1 1 3 5 7 9
0

Original CPT data

2
Depth (m)

3
Quadratic trend

Residual component
Figure 1. Normalized soil behavior type with Adapazari 5
data
6

Figure 3. Detrending of CPT data

Standard deviation is translated into a more

recognized parameter, coefficient of variation (COV)
which is defined as the ratio of the standard deviation to
the deterministic mean value. Correlation length or the
scale of fluctuation which is defined as the distance over
which the data points have meaningful correlation. The
procedure proposed by Jones et al. (2002) was employed
to calculate the scale of fluctuation of the CPT residual
profile. Figure 4 shows how the lag distance varies with
the window size used in moving average scheme. The
plot peaks at 0.16 m; therefore the scale of fluctuation is
taken to be 0.16 m.
Figure 2. Normalized soil behavior type with Adapazari
data
 0.18 (e.g., groundwater) through a soil. The saturated fast-flow
0.16
logic is used for simulating coupled fluid flow/deformation
Function of fluctuation (m)

of a porous medium (i.e., consolidation) by simply adding

0.14 the command “SET fastflow on”. This numerical scheme
0.12 provides faster solutions for consolidation calculations
which can be very time-consuming. The FLAC model for
0.1
one-dimensional consolidation problem is a column of 368
0.08 zones representing adjacent CPT readings and 100 kPa
pressure is applied at the top of the column to model the
0.06
surcharge effect.
0.04

0.02
4 Reliability analysis
0
0 0.5 1 1.5 2
CPT profiles were virtually reproduced using random field
Window size (m) theory and the stochastic properties acquired from the
Figure 4. Calculation of the correlation length original CPT profiles. Cholesky decomposition technique
was used to produce correlated random fields of .The
generated realizations of were then employed to
correlate shear modulus and produce equivalent profiles
Kulhawy and Phoon (1996) suggest the values of the shear modulus. The small strain shear modulus, ,
of scale of fluctuation for CPT parameters presented in for young, uncemented soils can be estimated using the
Table 1. In our case study, computed scale of fluctuation following relationship proposed by Robertson (2009) :
for vertical direction for corrected cone tip resistance is
. .
close to minimum of reported range. 0.0188 10 [2]

Table 1. Scale of fluctuation values for CPT (Kulhawy and in which is soil behavior type index and is vertical
Phoon (1996)) total stress.
The above correlation was used to interpret input
Scale of Fluctuation shear modulus values for numerical modeling. A constant
Fluctuation Soil No. of (m) poison ratio of 0.35 was considered for correlation of the
Property
Direction Type Studies
Range Mean bulk modulus, K as an input parameter for numerical
Sand, modeling of coupled consolidation. Ultimate consolidation
7 0.1-2.2 0.9
Vertical Clay settlement is assumed to be independent of the hydraulic
Clay 10 0.2-0.5 0.3 conductivity of alluvial deposits, k. Therefore the variability
Sand,
11 3.0-80.0 47.9 of the permeability coefficient was neglected. Near 100
Horizontal Clay percent average degree of consolidation was sought as
Clay 2 23.0-66.0 44.5 an index for full consolidation settlement.
2000 Monte Carlo simulations were conducted to
investigate the effect of spatial variability of on
3 Consolidation problem consolidation settlement results as shown in Figure 5. The
results show that the consolidation settlement fluctuates
Soils are naturally, inherently variable because of the way between 8 and 11.5 mm. The mean settlement of 9.6 mm
they are formed and the continuous processes of the is considered as the most likely settlement value.
environment which work to alter them. Taking the
coefficient of consolidation as a constant is a major 12
shortcoming of conventional theory. It is well known that 11.5
consolidation properties must be treated independently
11
Settlement (mm)

and cannot be embodied into a single coefficient of

consolidation. Therefore it is very important to use a 10.5
coupled approach for proper modeling of consolidation 10
behavior of heterogeneous soils (Huang et al. 2010). 9.5
Consolidation is one type of fluid/solid
9
interaction, in which the slow dissipation of pore pressure
causes displacements to occur in the soil. This type of 8.5
behavior involves two mechanical effects. First, changes 8
in pore pressure cause changes in effective stress, which 7.5
affect the response of the solid. Second, the fluid in a
7
zone reacts to mechanical volume change by a change in 0 500 1000 1500 2000
pore pressure. For the coupled consolidation analysis in
present study, commercially finite difference code, FLAC Number of realizations
5.0 (Itasca 2006) is used that models the flow of fluid
Figure 5. Settlement results of Monte Carlo simulations
 Figure 6 shows the convergence of the mean 5 Deterministic analysis
settlement as the number of simulations increases. It can
be seen from Figure 6 that 2000 simulations were enough CPTu data can be used to directly estimate induced
to give reliable and reproducible estimates. settlements due to an external load. To validate reliability
results with a deterministic estimate of settlements,
9.8 following simple formula (based 1-D consolidation) is
used to estimate vertical settlements:
9.7
∑ [3]
Mean settlemeent (mm)

9.6
Where, and are applied footing pressure, layer
9.5 thickness and stress reduction factor according to
Boussinesq respectively. As shown in Figure 8. ,
9.4 Most likely settlement : 9.6 mm constrained modulus of soil layer can be estimated from
CPT results using the CPTet-IT software based on the
9.3 correlation suggested by Robertson (2009). For a
constant surface load, 100 , maximum settlement
result accumulated on top of the layer is 9.2 mm.
9.2
The computed consolidation settlement for deterministic
analysis (9.2 mm) is in the range of proposed settlement
9.1 estimation based on reliability analysis (8 and 11.5mm).
0 500 1000 1500 2000
Number of realizations

Figure 6. Influence of the number of Monte Carlo

simulations on mean settlement

Reliability theory can be employed to evaluate

the effects of uncertainties embedded in settlements
calculations and the probability that the computed most
likely settlement will be larger than allowable settlements
can be defined as probability of failure, P . A more
convenient measure of design risk is the reliability index,
β which is defined as Φ P in which Φ function is
the inverse standard normal cumulative function. Target
reliability index should be more than 3 for proper
performance. In Figure 7 reliability index is calculated for
different allowable settlements and it shows that target
Figure 8. Consolidation settlement for deterministic
reliability index will satisfy when the allowable settlements
analysis
is around the maximum possible settlement.
11.6
11.4 6 Conclusions
Allowable Settlemenet (mm)

11.2
CPT data can be used to evaluate uncertainties involved
11 at the specific site by generating corrected cone tip
resistance, based on statistical parameters of residual
10.8
components including the standard deviation and the
10.6 scale of fluctuation. The quadratic trend and the
generated residuals were superposed to realize profiles of
10.4
employed in subsequent coupled Monte Carlo
10.2 simulations. Realizations of were utilized for correlation
10 of shear modulus which are finally fed to the finite
difference formulation to calculate ultimate consolidation.
9.8 The results of 2000 Monte-Carlo simulation shows that
9.6 variation of has significant effect on settlement values.
0 1 2 3 4 5 Range of ultimate consolidation settlements was
computed and provides a means of estimating most likely
Reliability Index (β) settlement which is essential for reliability analysis.
Reliability concepts can be applied to settlements
Figure 7. Evaluation of reliability index for different analyses by comparing the most likely settlement and
allowable settlements allowable settlement. The probability of failure occurs
 when the most likely settlement is larger than allowable Pishgah Gilani, P., Jamshidi Chenari, R. 2011a.
settlement. The results of reliability index for different Discussion of “Probabilistic Analysis of Coupled Soil
allowable settlements demonstrate that for satisfactory Consolidation” by Jinsong Huang, D.V. Griffiths and
performance it is important that the allowable settlement Gordon A. Fenton, ASCE Journal of Geotechnical and
be larger than maximum possible settlements. Geoenvironmental Engineering. Vol. 137, No. 9,857-
858.

7 References Pishgah Gilani, P., Jamshidi Chenari, R. 2011b.

Stochastic vs. Deterministic Analyses of Consolidation
Christian J. T., Baecher, G. B. 2011. Unresolved in Natural Deposits. Proceedings of GeoRisk2011
Problems in Geotechnical Risk and Reliability, Conference - Risk Assessment and Management in
GeoRisk2011, ASCE, Atlanta, GA, USA, 50-63. Geoengineering , Atlanta, GA. 271-278.

Doherty P., Gavin. K. 2010. Statistical Review of CPT Pishgah Gilani, P. 2012. Consolidation Behaviour of
Data and Implications for Pile Design, 2nd Heterogeneous Alluvial Deposits”, In partial fulfillment
International Symposium on Cone Penetration of the requirements the degree of master of science,
Testing, Huntington Beach, CA, USA. Volume 2&3: faculty of engineering, University of Guilan, Guilan,
Technical Papers, Session 3: Applications, Paper No. Iran.
3-05.
Robertson, P.K., 2009. CPT Interpretation – A Unified
Haldar S., Sivakumar Babu G.L. 2008. Reliability Approach, Canadian Geotechnical Journal,46: 1-19.
Measures for Pile Foundations Based on Cone
Penetration Data, Canadian Geotechnical Journal, 45, Robertson, P.K. 1990. Soil Classification Using the Cone
1699-1714. Penetration Test.” Canadian Geotechnical Journal,
27(1): 151-158.
Huang, J., Griffiths, D.V. and Fenton, G.A. (2010).
“Probabilistic analysis of coupled soil consolidation.” Uzielli, M. 2004. Variability of Stress-normalized CPT
ASCE Journal of Geotechnical and Geoenvironmental Parameters and Application to Seismic Liquefaction
Engineering, Vol. 136, No. 3, 417-430. Initiation Analysis. Ph.D. dissertation, University of
Florence, Italy.
Itasca Consulting Group Inc. 2006. FLAC 5.00-Reference
Manual. Minneapolis.

Jones, A.L., Kramer. S.L. and Arduino, P. 2002,

Estimation of Uncertainty in Geotechnical Properties
for Performance-Based Earthquake Engineering,”
PEER report 2002/16.

Kenarsari, E. A., Oloomi, R., Jamshidi Chenari, R. ,

Eslami A. 2011. Effect of vertical heterogeneity in soil
strength on pile bearing capacity prediction from CPT
data. Proceedings of the 36th Annual Conference on
Deep Foundations, Boston, MA, USA.

Kenarsari, E. A., Jamshidi Chenari, R. , Eslami A. 2012.

Characterization of the Correlation Structure of CPT
Data in Sand Deposits”. International Journal of Civil
Engineering.

Kulhawy, F. H., Phoon, K. 1996. Engineering judgment in

the evolution from deterministic to reliability-based
foundation design (plenary), Uncertainty in the
Geologic Enviroment, From Theory to Practice,
Proceeding of Uncertainty ’96, Geotechnical Special
Publication No. 58.

Phoon, K-K., Kulhawy, F. H. & Grigoriu, M. D. 1995

Reliability-Based Design of Foundations for
Transmission Line Structure, Report TR-105000,
Electric Power Research Institute, Palo Alto , 380 p.

View publication stats