Professional Documents
Culture Documents
net/publication/233013578
Article in Georisk Assessment and Management of Risk for Engineered Systems and Geohazards · March 2010
DOI: 10.1080/17499510902896612
CITATIONS READS
3 2,334
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Gil-Lim Yoon on 17 July 2014.
To cite this Article Yoon, Gil Lim, Yoon, Yeo Won and Kim, Hong Yeon(2010) 'Determination of geotechnical
characteristic values of marine clay', Georisk: Assessment and Management of Risk for Engineered Systems and
Geohazards, 4: 1, 51 — 61
To link to this Article: DOI: 10.1080/17499510902896612
URL: http://dx.doi.org/10.1080/17499510902896612
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Georisk
Vol. 4, No. 1, March 2010, 5161
a
Coastal Harbor Engineering Lab., Korea Ocean Research & Development Institute, Ansan, Korea; bDepartment of Civil
Engineering, Inha University, Incheon, Korea; cCoastal Harbor Engineering Lab., Korea Ocean Research & Development
Institute (Ph.D candidate at Inha Univ.), Ansan, Korea
(Received 19 May 2008; final version received 14 March 2009)
In this paper, methods of determining characteristic values for geotechnical reliability-based design were analysed
for Korean marine clays. The characteristic values using the Student approach were closest to the mean value of
parameters followed by the characteristic values using Ovesen, Schneider and EN 1990’s approaches. Four
different approaches showed a trend of evaluating characteristic values conservatively with increasing soil
variability. Geotechnical stability and settlement of a breakwater, subjected to nominal stress on unimproved soft
ground, were studied to investigate the effects of the estimated different characteristic values. Using Schneider’s
approach, the ratio of allowable bearing capacity to acting loads was 65% of that based on the arithmetic mean
values, and the settlement was underestimated by 13.6%. It was also found that determining a representative
value using an arithmetic mean value significantly overestimated the ratio comparing the values using the
proposed approaches, which may induce an uneconomic design of structure if not considering serious soil
Downloaded By: [Yoon, Gil Lim] At: 13:53 2 March 2010
difficulty in defining the methods to estimate char- this study, the mean value should not represent
acteristic value. Eurocode 7 offers four levels of directly the arithmetic mean, but should be consid-
values as a procedure to gain design value of ered using statistical estimation error to obtain a
geotechnical parameters, which are measured value, mean value. Also, this careful evaluation of the mean
derived value, characteristic value and design values. value should be considered by not only geotechnical
Figure 1 shows basic concepts of these values. theory and experience of similar sites, but also
consistent cross-validation of some other results
Measured value based on ground investigation.
‘Measured value’ is defined as a value measured from
many kinds of laboratory and in-situ tests. For Design value
example, ground water level, N-value from standard Design value is a parameter which is used by design
penetration test (SPT), stress and strain from triaxial calculation. It is determined either from the char-
compressive test and so on. acteristic value by applying partial factor or by direct
assessment from the derived value.
Derived value
‘Derived value’ is defined as a value presenting a Determination of geotechnical characteristic values
property of geomaterials which is estimated from the
A characteristic value of geotechnical parameter is
measured values using theoretical, empirical or sta-
defined using a bias factor, kr, that is, mean value of
Downloaded By: [Yoon, Gil Lim] At: 13:53 2 March 2010
Figure 1. Procedure to determine design value (Orr and Figure 2. Characteristic values of parameters for safety
Farrell 1999). side (Becker 2006).
Georisk 53
field and is not 5% fractile of derived value obtained the result of Equation (3) where the t-parameter is
from test results, but the value affects the occurrence applied in case of an infinite number of N.
of the limit state. Therefore, the characteristic value
1:645
corresponds to a 95% confidence level so that the Xk m(X) pffiffiffiffi s(X) (4)
mean value, m(X), becomes greater than the deter- N
mined characteristic value which is represented as On the other hand, Schneider (1997) presented a
follows using statistical theory more simple equation.
Xk m(X)[1kn ×V(X)] (1) Xk m(X)0:5s(X) (5)
where kn is a factor which depends on statistical when kr equals 0.5 in Equation (1), that is, when N
probability distribution and the number of test becomes 11 in Equation (4), Schneider showed a good
results, and V(X) is coefficient of variation (COV). approximate value for Xk, through comparative
The term [1kn ×V(X)] is used as a correction studies. Equation (5) represents the case of determin-
factor. ing the lower 50% of the SD for the mean value.
According to EN 1990, when a value of either In the Japanese design code for port facilities, the
material or material property has a prescribed prob- correction factors are slightly different depending on
ability which is not obtained from hypothetical the amount of data. They are denoted as b1 and b2
unlimited tests, it is defined as a characteristic value respectively and are represented as the following
with any prescribed confidence level in the statistical
distribution assumed as a specific property. Also, it b1 10:5COV (6)
Downloaded By: [Yoon, Gil Lim] At: 13:53 2 March 2010
Table 2. Maximum number of test data with geographical locations of the site in Korea.
and were analysed statistically. Figure 3 shows the Effects of characteristic values on practical design
results for the comparison of the mean value obtained The effects of characteristic values on geotechnical
Downloaded By: [Yoon, Gil Lim] At: 13:53 2 March 2010
by using each approach. For the estimation results stability and settlement were investigated using
for characteristic values of soil parameters in physical arithmetic mean and the characteristic values esti-
characteristics, Of the four approaches, Student’s mated using four different approaches. In order to
method presented results closest to the mean value. find out their effects, the design values were deter-
However, when sufficient test data were available, mined using partial material factors proposed by
Student’s and Ovesen’s approaches showed similar Eurocode 7, and then geotechnical stability and
results. This is because the t-value is convergent to settlement problems were analysed for the marine
1.645 within a 95% confidence level when the data clays of Gwangyang Port and Busan Port subjected
exceed a certain number of data, even though the t- to breakwater loading. Also, the results were com-
value is affected by the number of test data. Thus, pared with representative values, which were recently
Ovesen’s equation is equal to Student’s. However, used for practical designs with the design values
EN 1990’s approach gives very small values in determined by the characteristic values considering
comparison with the mean values, even though this soil uncertainties.
approach depends on the dispersion of a population.
Determination of design values
Strength characteristics In the material factor approach (MFA), the design
value (Xd) of a geotechnical parameter is soil
Comparison of the mean values with characteristic
property which is substituted for design calculation.
values for certain parameters of strength character-
This value is taken as a characteristic value (Xk)
istic is presented in Figure 4. In the case of strength
divided by a partial factor (gm) as the following
characteristic parameters, Student’s and Ovesen’s
approaches gave a result closest to the mean value Xd Xk =gm (6)
followed by Schneider’s and EN 1990’s approaches. where the partial factor is determined at a confidence
Especially, EN 1990’s approach estimated the values level considering reliability of evaluation method to
less than zero with large COV. determine the characteristic value and variation of the
derived value. In choosing the partial factor value for
the design, it should be determined which side
Consolidation characteristics between the resistance and load sides is most affected
For parameters related to consolidation characteris- by. Table 3 only shows partial material factors for the
tics, the soil variability in Gwangyang Port was larger ultimate limit state adopted in Eurocode 7 (Orr and
than in Busan Port. As shown in Figure 5, Student’s Farrell 1999).
approach came the closest to the mean value in Case A shows the uncertainties with permanent
estimating characteristic values followed by Schnei- loading and variable loading when the strength of
der’s and EN 1990’s approaches. structure and ground is insufficient. The aims of Case
Georisk 55
Downloaded By: [Yoon, Gil Lim] At: 13:53 2 March 2010
Figure 5. Comparison of characteristic values for consolidation characteristic: (a) compression index from oedometer test
(Cc); (b) coefficient of consolidation for vertical direction (Cv); (c) coefficient of consolidation for horizontal direction (Ch); (d)
permeability for vertical direction (Kv).
1990’s approaches were excluded, because the former Case study 1: comparison of results using characteristic
is similar to Student’s approach owing to a relatively values with arithmetic means
large number of samples, and the latter are inap- A loading of rubble mound breakwater on the
propriate for geotechnical design as they are less than unimproved soft ground was considered to review
zero. Representative design values to calculate geo- the geotechnical stability and foundation settlement.
technical stability and ground settlement are shown in In general, marine structures are mostly constructed
Table 4. after treating the ground using sand compaction piles
Table 3. Partial material factors for ultimate limit state (Orr and Farrell 1999).
Values in bold are partial factors either given or implied in the ENV version of EC7.
Values in italics are proposed partial factors not in the ENV that may be in the EN version.
Georisk 57
Table 4. Summary of main design values. To calculate geotechnical stability and settlement,
Soil the modified Terzaghi method and the compression
parameter Port Mean Student Schneider index method were employed. The safety factor for
calculating geotechnical stability, 1.5, is used (Min-
/ e Gwangyang 2.27 2.25 2.01 istry of Maritime Affairs and Fisheries 2005). Figure
Busan 1.63 1.61 1.46
3 7 shows the results of allowable bearing capacity
/g (kN/m ) Gwangyang 15.23 15.18 14.69
t
using qa/Q, i.e. allowable bearing capacity divided by
Busan 16.45 16.38 15.56
/G Gwangyang 2.71 2.71 2.70 acting load per unit area. The results (qa/Q) using
s
Busan 2.70 2.69 2.68 Student’s approach were slightly smaller than the
qu (kPa) Gwangyang 18.81 18.13 11.25 mean values, 9798% of the mean values. However,
Busan 32.48 31.59 20.68 the results using Schneider’s approach gave very
/C
UU (kPa) Gwangyang 12.54 12.15 8.27 underestimated values, 6466% of the mean value.
Busan 20.40 19.86 13.06 Considering the results, the safety factor of
/C Gwangyang 1.13 1.11 0.92
c Busan clay is 1.6 times that of Gwangyang clay.
Busan 0.70 0.69 0.57
2 Figure 8 shows the changes of total settlements and
/C
v (cm /sec) Gwangyang 0.00260 0.00246 0.00140
residual settlements with time in Gwangyang Port.
Busan 0.00559 0.00535 0.00349
Total settlements depending on the compression
index and initial void ratio were 1.821 m (Mean),
1.803 m (Student), and 1.617 m (Schneider), respec-
Downloaded By: [Yoon, Gil Lim] At: 13:53 2 March 2010
or other soil improvement methods. To investigate tively. Thus, there were no great differences between
strength and compression characteristics of unim- Student’s approach and the mean value. But
proved marine clay with each mean value and Schneider’s approach underestimated the settlement
characteristic value of parameters, the effects on about 0.2 m (12.6%) in comparison with the mean
stability and settlement were only estimated and value.
compared. The cross-sectional diagram is presented Figure 9 shows the total settlement and residual
in Figure 6. settlement with time in Busan Port. The total
settlements are 1.40 m (Mean), 1.396 (Student) and design cases of harbour and port construction in the
1.222 m (Schneider), respectively. Using Student’s each area are obtained by intuitive judgement of
approach, there was no great difference, whereas designers, by ignoring statistical variability and by
Schneider’s approach underestimated the settlement insufficiently or locally tested data; secondly, design
by 0.18 m (14.6%) in comparison with the mean value. values were estimated through characteristic values
considering uncertainties; finally, arithmetic mean
Case study 2: Comparison of results using values were obtained. Those are the cases when
characteristic value, arithmetic mean and intuitive designs have been already finished or construction
representative value is being performed. Comparing the results of design
Case study 2 shows comparison with stability and with sufficient variable data is meaningful in aspects
settlement results using three kinds of values. First, of economic feasibility or safety margin. Table 5
representative values which were applied to recent shows the main design values which have been
recently applied to the designs of port and harbour comparing them with the values considering uncer-
constructions in Gwangyang and Busan areas. Acting tainties, the results obtained by using the representa-
Downloaded By: [Yoon, Gil Lim] At: 13:53 2 March 2010
load was applied with the same cross-sectional break- tive values overestimated qa/Q by 2.4% when qa/Q
water structure as shown above. was calculated with the mean value, 3.7% using
Figure 10 shows the value of qa/Qs ratios Student’s approach and 20.4% using Schneider’s
calculated by the representative values applied to approach. Thus, there were no great differences
each case and the calculated values. In the case of except with Schneider’s approach. In the case of
Gwangyang (Case A), comparing the design value Busan (Case B), the results obtained by using the
with the values considering uncertainties, the results representative values overestimated qa/Q by 24.4%
obtained by using the representative values over- when qa/Q was calculated using the mean value,
estimated qa/Q by 39.9% when qa/Q was calculated 25.1% using Student’s approach and 48.3% using
by using the mean value, 44.3% by using Student’s Schneider’s approach. These are more excessive
approach and 111.5% by using Schneider’s approach. results than the cases of considering uncertainties.
In the case of Busan (Case B), the results obtained by
using the representative values overestimated qa/Q by
Conclusions
47.2% when qa/Q was calculated by using the mean
value, 60.0% by using Student’s approach and Geotechnical characteristic values of main marine
130.0% by using Schneider’s approach. Thus, in clay in Korea were estimated using four statistical
estimation of the stability, the representative values methods. Characteristic values for the marine clays
applied to each design approach typically overesti- using Student’s approach are closest to the mean
mated up to twice the actual values because of not value followed by Ovesen’s, Schneider’s and EN
considering uncertainties. Also, the stability charac- 1990’s approaches. EN 1990’s approach gives the
teristics of Busan clay were better than those of smallest characteristic values and resulted in less than
Gwangyang clay up to of 1.61.7 times. This is zero value with statistical variability, which is in-
similar to the results of case study 1. appropriate for estimating geotechnical design value.
Figure 11 shows comparison with settlements of Comparing Student’s with Ovesen’s approach, the
unimproved ground under the same loading and soil latter gives slightly larger values with a small number
conditions. In the case of Gwangyang (Case A), of samples, but the values become similar with a large
Figure 10. Comparison from the calculated for qa/Q: (a) Gwangyang; (b) Busan.
number of samples. A larger number of samples samples. The characteristic value is always estimated
induces a smaller degree of decreasing, even though at less than the mean value and has trends to be
the t-value in Student’s approach decreases with an estimated more conservatively with increasing soil
increasing number of samples. In that case, t-value is variability.
convergent to any specific value. When N equals 11, When using the design values determined by using
Ovesen’s equation becomes equal to Schneider’s and each approach, qa/Qs ratios with Student’s approach
both estimate the same value. Schneider asserted that did differ greatly from those by the mean values. But,
an approximate value was well predicted for the qa/Qs ratios using Schneider’s approach underesti-
characteristic value. mated the mean values by 6466%. In the case of
All approaches are used to estimate characteristic total settlements for both areas, there were differences
values by deducting a constant rate of variability between Student’s approach and the mean value.
from a mean value. The soil variability is a value However, settlements using Schneider’s approach
(Kn), which is multiplied by SD and a factor varying were less than those using the mean value by 12.6
with probability distribution and the number of 14.6%. In comparison with them, the stability
Georisk 61
Acknowledgements
This paper was made possible with a financial support from
the Ministry of Land and Marine Affairs in Korea
(Research code; PM54100).
References
Downloaded By: [Yoon, Gil Lim] At: 13:53 2 March 2010