Professional Documents
Culture Documents
8, 597–602
Pore pressure dissipation tests were carried out employ- On a effectué des essais de dissipation de la pression des
ing two miniature piezocones, with areas of 1 cm2 and pores en employant deux piézocônes miniatures, avec une
5 cm2 , in a bed of reconstituted kaolin. By using indepen- surface comprise entre 1 cm2 et 5 cm2 , dans une couche
dently determined values of the coefficient of consolida- de kaolin reconstitué. En employant des valeurs, établies
tion, the results were interpreted to derive appropriate indépendamment, du coefficient de consolidation, on a
values of rigidity index for use with the theory of Teh & interprété les résultats pour en dériver des valeurs ap-
Houlsby. In accordance with the theory, different filter proximatives de l’indice de rigidité pouvant être utilisées
positions on a piezocone (tip, face and shoulder) were not avec la théorie de Teh et Houlsby. D’après cette théorie,
found to require different values of rigidity index. How- différentes positions du filtre, sur un piézocône (bout,
ever, the importance of allowing for initial load relaxation face et épaulement) ne nécessiteraient pas des valeurs
at the commencement of the dissipation test was demon- diverses de l’indice de rigidité. Toutefois, on a démontré
strated, and there appeared to be a systematic effect of l’importance de la prévision de la relaxation initiale de la
piezocone size. charge au début de l’essai de dissipation, et il semble
qu’il existe un effet systématique de la taille du piézo-
KEYWORDS: clays; consolidation; in situ testing; site investi- cône.
gation
INTRODUCTION 1·2
The piezocone has developed into the most widely used and, Shoulder
Normalised excess pore pressure, U
arguably, the most useful probing tool for ground investiga- 1·0 Shaft 5R from
tion. Not only does it provide information about soil type shoulder
and stratification, but it can also be used to determine a 0·8
range of in situ soil properties (Lunne et al., 1997). For Shaft 10R
clays, this includes the determination of the coefficient of 0·6
from shoulder
consolidation for horizontal water flow, ch , obtained from
analysis of the dissipation of transient pore pressures set up 0·4 Cone face
during the driving of the piezocone. It is the interpretation
of piezocone dissipation test data that forms the subject of 0·2 Cone tip
this paper.
The driving of a piezocone causes increases of total stress 0
0·0001 0·001 0·01 0·1 1 10 100
and large shear strains in the surrounding soil. In most clays,
Time factor, T *
except heavily overconsolidated ones, the resulting excess
pore pressures all around the piezocone are positive, and, as Fig. 1. Normalised excess pore pressure dissipation (after Teh &
they subsequently dissipate, consolidation occurs. In order to Houlsby, 1991)
describe this soil behaviour, a model is needed, and a widely
recognised model is that of an elastic-plastic continuum, in
which pore pressures dissipate according to the Terzaghi– used the strength and stiffness parameters for Boston Blue
Rendulic theory of consolidation. clay, effectively giving Ir a value of around 550. However,
The pore pressure dissipation rate depends partly on the with Teh & Houlsby’s theory, lower values of Ir are generally
initial distribution of pore pressure set up during driving. appropriate (Robertson et al., 1992).
This can be determined, for example, by applying the strain Despite its widespread use, there is a significant problem
path method (Baligh, 1985) with an assumed stress–strain with the model of an elastic-plastic continuum, in that the
behaviour. Teh & Houlsby (1991) presented a numerical shear modulus of real soil is strain level dependent. Near the
solution to the subsequent consolidation problem in terms of piezocone, and particularly around the tip, where shear strain
the rigidity index Ir ¼ G/su , where G is the shear modulus levels are high, the modulus will be very much lower than
and su is the undrained shear strength of the soil, as well as further away, where strain levels are lower, so it is difficult
ch . In order to estimate ch from a piezocone test, it is to know what single value should be taken for use in the
therefore necessary to use a value for Ir . Fig. 1 shows the model. In practice, this has been resolved empirically by
dissipation curves predicted for different locations on a using a triaxial test to provide the value of su , and taking
piezocone in terms of the time factor T ¼ ch t=I r0 5 R2,
:
the secant modulus at 50% of the undrained strength as the
where R is the radius of the piezocone and t is time. Baligh corresponding value of G. Danziger et al. (1997) and
& Levadoux (1986) presented results of similar analysis that Schnaid et al. (1996) showed that, when this approach was
employed with Teh & Houlsby’s theory to interpret various
piezocone test results, the calculated values of ch were of
Manuscript received 12 August 2004; revised manuscript accepted
5 August 2005.
the same order of magnitude as those measured indepen-
Discussion on this paper closes on 1 April 2006, for further details dently in oedometer tests in the laboratory.
see p. ii. Making the same assumption that the Teh & Houlsby
* Oxford University. model is a suitable one, the present paper reports experience
† University of Sheffield. of determining the Ir value that is needed to give the same
597
598 SILLS AND HIRD
coefficient of consolidation from a piezocone dissipation test 950 1500
Shoulder pore pressure
as from an oedometer test. Results are taken from laboratory
850
deployments of piezocones with various pore pressure meas- 1300
900 1400
850 Shoulder m11d4a
Face 1200
800
Tip resistance, qt: kPa
Tip resistance, qt
Pore pressure: kPa
750 1000
700 800
650
600 Sleeve mid-point 600
550 400
500 Sleeve top
200
450
400 0
0·01 0·1 1 10 100 1000 10000
Time: s
1·2 1·2
Normalised excess pore pressure
Normalised excess pore pressure
0·8 0·8
0·6 0·6
0·2 0·2
0 0
0·01 0·1 1 10 100 1000 10000 0·01 0·1 1 10 100 1000 10000
Time: s Time: s
Fig. 5. Summary of 1 cm2 excess pore pressure dissipations Fig. 7. Shoulder dissipation for 1cm2 and 5 cm2 piezocones
2·5 1·55
2
5 cm sleeve 2 1·50
Normalised excess pore pressure
2·0 1·45
5 cm2 sleeve 1
1·40
Void ratio, e
0·5 1·20
Fig. 6. Summary of 5 cm2 excess pore pressure dissipations Fig. 8. Results from two Rowe cell tests with radial drainage
600 SILLS AND HIRD
tion cells with horizontal drainage. The clay for these tests less than 2.5% if the hypothesis were true. Therefore the
was prepared with the same history of loading and drainage hypothesis must be rejected: that is, the variation is statisti-
as the clay in the soil model. After being consolidated in the cally significant.
Rowe cell under a stress of 250 kPa, it was subjected to two Inspection of Table 2 suggests that the instrument size
unloading and reloading cycles between 250 kPa and may have been responsible for at least some of this varia-
125 kPa before being loaded to 500 kPa and finally un- tion. Again, this can be checked statistically by comparing
loaded. It can be seen that the curves are similar, but not the two data sets obtained with a filter at the piezocone
identical, presumably as a consequence of small differences shoulder using the Mann–Whitney U test (e.g. Miller &
in sample preparation. The values of ch calculated at each Freund, 1977). The computed statistic, U ¼ 20, leads to a
stage are shown in Table 1. conclusion that the piezocone size probably had a significant
For most soils, the net increase in pore pressure during influence. Otherwise, this value would only have occurred
piezocone driving is larger than the mean increase in total with a probability of between 5% and 10% (Owen, 1962).
stress, so that the soil undergoes a reduction in effective For either the 5 cm2 or the 1 cm2 piezocone, the effect of
stress and, in effect, becomes more overconsolidated. This is filter location can similarly be examined using the Mann–1
followed by re-consolidation as the pore water drains. It is Whitney test, but although average values of ch /(Ir )2 are
therefore the unloading from 250 kPa in the Rowe cell and higher for the shoulder location than for either the face or
the subsequent reloading that provides the most relevant data tip location respectively, the test fails to demonstrate any
for comparison with the soil behaviour around the piezo- statistically significant differences.
cone. Table 1 shows good repeatability in each Rowe cell
test for ch values under these conditions, with average values
of 0.031 and 0.046 cm2 /s from the two tests. In order to DISCUSSION
have a single value for comparison with the piezocone It can reasonably be expected that the kaolin tested was
results, the average from the two tests, ch ¼ 0.038 cm2 /s, uniform, and the closely controlled nature of the experiments
was taken for the subsequent calculations.
1
and the favourable testing environment in the laboratory
Having obtained the ratio ch /(Ir )2 by matching the theor- should have led to highly repeatable results and interpreta-
etical and measured pore pressure dissipations, the above tions. Nevertheless, some variation is inevitable, and this can
value of ch was used to calculate Ir for each experiment. be seen in the Rowe cell tests to find the coefficient of
The results are presented in Table 2. The theoretical solution consolidation, where two sets of data were compared di-
does not predict the increased excess pore pressures that rectly. Such variation must also be expected in the dissipa-
were observed in the sleeve measurements, and it was there- tion results, as was observed in the spread of Ir values for a
fore decided to exclude the data collected at both sleeve given piezocone and filter position in Table 2. The advan-
locations. tage of analysing the data in statistical terms is that systema-
tic differences can be identified, irrespective of random
variations.
Although not the primary objective of this paper, it is of
STATISTICAL ANALYSIS
interest to compare the back-calculated Ir values with values
In Table 2 it can be seen that there is variation, not only
that can be obtained from triaxial tests. While no triaxial
within each set of data, but also between the average values
tests were conducted as part of the current investigation,
of each data set. The variation within the data sets is1
some Ir values can be derived from other research. For
expressed by the coefficient of variation, V, which for ch /(Ir )2
example, Hajj (1990) conducted undrained triaxial compres-
ranges between 11.8% and 23.1%. Because values of Ir are
1 sion and extension tests on anisotropically and normally
obtained after squaring the test result (i.e. c h /(Ir )2 ), values of
1 consolidated kaolin, albeit consolidated under different ver-
V are larger for Ir than for ch /(Ir )2 .
tical stresses from that used for the piezocone tests. Some of
The significance of the variation between the data sets can
these test data were reported by Hird & Hajj (1995). Parry
be analysed statistically by testing the hypothesis that all the
& Nadarajah (1974) presented data from similar tests as well
results are drawn from a common ‘population’: that is, the
as from tests on isotropically consolidated specimens. The
piezocone size and/or the filter position had no influence on
1 derived values of Ir , based on secant moduli, are shown in
the measured value of ch /(Ir )2 . Application of the Kruskal–
Table 3. As already indicated, the strain dependence of such
Wallis H test (e.g. Miller & Freund, 1977) yields the statistic
values poses a difficulty in selecting appropriate values to
H ¼ 9.99, a value that would occur with a probability of
use in the interpretation of piezocone measurements. At an
axial strain level of 1%, the values in Table 3 are clearly
Table 1. Horizontal coefficient of consolidation from Rowe cell lower than the back-calculated values in Table 2, while at
tests
0.1% axial strain they appear comparable. However, it
Vertical effective stress: kPa ch : cm2 /s should be noted that the back-calculated Ir values gave a
correct determination of ch in overconsolidated soil whereas
From To First test Second test the triaxial tests were conducted on normally consolidated
soil. As seen in Table 1, the values of ch for the normally
10 125 0.011 0.075 consolidated soil in the stress ranges 125–250 kPa and 250–
125 250 0.006 0.013 500 kPa are around four to five times smaller than those for
250 125 0.029* 0.043† the overconsolidated soil on which the Ir values in Table 2
125 250 0.029* 0.043† have been based. Therefore, if the normally consolidated
250 125 – 0.049† values of ch had been used to calculate Ir , the values would
125 250 0.035* 0.049† have been roughly half of those listed in Table 2.
250 500 0.007 0.012
500 250 0.032 0.043
One of the most striking findings of the present work was
250 125 0.012 0.020 the influence of the initial pore pressure on the interpretation
125 10 0.002 0.007 of measurements made on the face of the 5 cm2 piezocone.
Fig. 9 shows a typical best fit between the theoretical and
* Values averaged to give ch ¼ 0.031 cm2 /s. measured values when the latter were normalised by the
† Values averaged to give ch ¼ 0.046 cm2 /s. measurement at 0.01 s, and it can be seen that there is a
COEFFICIENT OF CONSOLIDATION FROM PIEZOCONE MEASUREMENTS 601
Table 2. Calculated values of I r for kaolin with ch 0:038cm2 =s
Table 3. Rigidity indices from consolidated undrained triaxial tests on normally consolidated kaolin
Consolidation mode Initial effective Shearing mode Ir Axial ˜q/˜qu :* Data source
vertical stress: strain: % %
kPa
Anisotropic (zero lateral strain) 350 Compression 105 0.1 51 Hajj (1990)
347 Extension 389 0.1 32
350 Compression 18 1.0 86
347 Extension 76 1.0 62
545 Compression 8 1.0 81 Parry & Nadarajah (1974)
540 Extension 60 1.0 45
Isotropic 550 Compression 41 1.0 63
550 Extension 43 1.0 64