Effect of Penetration Rate on Penetrometer

Resistance in Clay
Shin Fun Chung1; Mark F. Randolph2; and James A. Schneider3

Abstract: This paper discusses the effects of penetration rate on the penetration resistance in soft clay for various shaped penetrometers
共cone, T-bar, ball, and plate兲 and for T-bars with different aspect ratios. Constant rate 共“normal”兲 and variable rate 共“twitch”兲 penetration
tests, where the penetration rate was successively halved over eight steps with the penetrometer advanced by one or two diameters in each
step, were undertaken in the beam centrifuge at the University of Western Australia. The tests were conducted on samples reconstituted
from clay collected from the Burswood site in Western Australia. The twitch tests showed higher penetration resistance than the corre-
sponding normal tests after the penetration rate had been reduced by a factor of 16 due to cumulative effects of partial consolidation. The
penetration rate at which the resistance started to increase due to partial consolidation was used to estimate the consolidation coefficient,
cv, of the reconstituted clay. The interpreted cv values were similar to values estimated from other consolidation data, both in the
centrifuge and from laboratory Rowe cell and constant rate of strain consolidation tests. In addition, results from in situ twitch tests at the
Burswood site were examined to evaluate viscous effects on the penetration resistance.
DOI: 10.1061/共ASCE兲1090-0241共2006兲132:9共1188兲
CE Database subject headings: Centrifuge model; Clays; Loading rate; Penetration resistance; Penetration tests; Viscosity.

Introduction may be compared by normalizing the rate of penetration, v, as

The tip resistance of an advancing penetrometer decreases as the vd

V= 共1兲
rate of penetration is decreased—provided the conditions around cv
the advancing penetrometer are undrained—due to viscous ef-
fects. On the other hand, once the penetration rate has reduced where d = penetrometer diameter; and cv = coefficient of consoli-
sufficiently for partial consolidation to occur, with local strength- dation 共Finnie and Randolph 1994; Randolph and House 2001兲.
ening of the soil around the penetrometer, the tip resistance will In natural clays, consolidation may be dominated by radial flow
increase as the penetration rate is decreased. Therefore, there is a of pore water, and a horizontal consolidation coefficient, ch,
transition point from undrained to partially drained response may be more appropriate. A further consideration is whether cv
where the viscous and partial consolidation effects balance out, should be the value associated with normal consolidation or with
leading to a minimum resistance. This phenomenon is clearly swelling. At present, only empirical correlations have been devel-
illustrated in the results presented by Bemben and Myers 共1974兲, oped for the variation of penetration resistance with normalized
who carried out in situ mechanical cone penetration tests with velocity, and those have been based on a cv value for normal
penetration rates ranging from 0.2 to 200 mm/ s in varved clay, consolidation. The logic for that choice was that the increase in
and by Roy et al. 共1982兲 who performed piezocone tests with resistance due to partial consolidation involves increasing effec-
rates varying from 0.5 to 40 mm/ s in sensitive clay. Data from tive stress levels above the in situ values 共House et al. 2001兲. In
penetrometer tests of various diameters, and in different clays, this paper, the symbol cv, corresponding to normal consolidation
as measured in a conventional consolidation test, will be used
1 throughout without further consideration of the precise nature of
Geotechnical Engineer, Advanced Geomechanics, 4 Leura St,
the flow regime.
Nedlands, Western Australia 6009; formerly, Ph.D. Student, Univ. of
Western Australia. E-mail: [email protected] In order to verify the transition point from undrained to par-
2
ARC Federation Fellow, Centre for Offshore Foundation Systems, tially drained response, a series of constant-rate T-bar penetration
Univ. of Western Australia, 35 Stirling Highway, Crawley, Western tests at various penetration rates, covering three orders of magni-
Australia 6009 共corresponding author兲. E-mail: [email protected]. tude, were conducted in the centrifuge independently by Watson
edu.au and Suemasa 共unpublished data, 2000兲; House et al. 共2001兲; and
3
Ph.D. Student, School of Civil and Resource Engineering, Univ. of Randolph and Hope 共2004兲, who also carried out similar tests
Western Australia, 35 Stirling Highway, Crawley, Western Australia 6009. using a piezocone penetrometer. Data obtained from these pen-
E-mail: [email protected] etration tests were plotted in the form of normalized resistance
Note. Discussion open until February 1, 2007. Separate discussions against V, and “backbone” curves giving the relationship between
must be submitted for individual papers. To extend the closing date by
the two parameters for the T-bar and cone penetrometers were
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this paper was submitted for review and pos- subsequently derived.
sible publication on December 21, 2005; approved on February 2, 2006. The backbone curves may be used to assess whether a given
This paper is part of the Journal of Geotechnical and Geoenvironmental penetration test in a particular soil is partially drained or not, or to
Engineering, Vol. 132, No. 9, September 1, 2006. ©ASCE, ISSN 1090- deduce the consolidation coefficient, cv, of the soil. For the
0241/2006/9-1188–1196/$25.00. former objective, where a single penetration test is performed,

1188 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2006

Table 1. Soil Properties for Reconstituted Burswood Clay
Soil property Value
Effective unit weight, ␥⬘ 共kN/ m 兲 3
4.4–5.1
⬘
Reconstituted friction angle, ␾centrifuge共°兲 29–32
Moisture content, ␻ 共%兲 55–86
Liquid limit, ␻L 共%兲 120
Plastic limit, ␻ P 共%兲 51
Plasticity index, I P 共%兲 69
Compression index, Cc 0.66–0.82
Coefficient of consolidation, cv 共m2 / yr兲 0.5–0.75
Particle size ⬍0.02 mm 共%兲 75
Particle size ⬍0.006 mm 共%兲 40
Clay content 共⬍0.002 mm兲 共%兲 10

some independent assessment of cv is needed in order to calculate

V for the test; for the latter objective, multiple tests performed at
different penetration rates are needed to assess a transition in the
measured penetration resistance. A value of cv can then be de-
duced by matching the transition point to an appropriate backbone
curve. The transition point, and hence estimate of cv, may also be
assessed from a single variable rate penetration test 共referred to as
a “twitch” test兲, where the penetration rate is successively halved
with the penetrometer advanced by one to two diameters of the
probe at each new rate 共Randolph and House 2001兲.
This paper presents the results of twitch penetration tests for
different shaped model penetrometers conducted in a centrifuge.
The model penetrometers tested included cone, T-bar, ball, and
plate. The effect of partial consolidation on the penetration resis-
tance has been used to estimate the consolidation coefficients of Fig. 1. Consolidation of samples in the centrifuge at 100 g: 共a兲
the reconstituted samples. The interpreted cv is compared with first centrifuge strongbox sample 共Box 1兲; 共b兲 second centrifuge
values deduced from other consolidation data including bulk con- strongbox sample 共Box 2兲
solidation of the sample in the centrifuge, and laboratory Rowe
cell and constant rate of strain consolidation 共CRSC兲 tests. Vis-
cous effects on the tip resistance are also examined in light of in in Figs. 2 and 3, respectively. In the CRSC tests 共Fig. 2兲, the
situ twitch test results reported by Schneider et al. 共2004兲. values of cv corresponding to the yield 共or preconsolidation兲
stress are in the range 1 – 1.2 m2 / year, but reduce to between 0.5
and 0.7 m2 / year at stress levels beyond three times the yield
stress. The Rowe cell tests performed on Burswood clay material
Soil Properties
reconsolidated from a slurry gave rather low values of cv, be-
Clay samples for the centrifuge testing were reconstituted using tween 0.35 and 0.54 m2 / year 共Fig. 3兲.
bulk material collected from a depth of about 6 m at the Bur-
swood site, located a few kilometers from Perth, Western Austra-
lia. Soil properties of the reconstituted samples are summarized in Centrifuge Testing
Table 1. The properties are mostly based on results from labora-
tory tests conducted on tube samples retrieved after completion of The model penetrometer tests were carried out “in-flight” in the
the centrifuge tests. beam centrifuge at the University of Western Australia, a detailed
The values of cv shown in Table 1 have been estimated using description of which has been given by Randolph et al. 共1991兲.
data during consolidation of the strong-box samples in the centri- Two centrifuge strongbox samples were prepared using bulk clay
fuge, which are presented in term of surface settlement against from the Burswood site, reconstituted at a water content of 125%.
root time 共in minutes兲 in Fig. 1. The time for 90% degree of Both samples were consolidated in a press under a vertical stress
consolidation 共t90兲 has been estimated using Taylor’s root time of 35 kPa, and then transferred to the centrifuge for consolidation
method, as indicated in the figure. The values of cv for the at a centrifugal acceleration of 100 g. A final sample thickness of
samples are then calculated as between 200 and 230 mm was targeted, which is equivalent to a
prototype depth of 20– 23 m.
T90h2 In the first strongbox sample 共Box 1兲, various model pen-
cv = 共2兲 etrometers were employed, comprising cone, ball, and plate pen-
t90
etrometers with respective diameters of 10, 11.9, and 11.2 mm,
where T90 = time factor 共taken as 0.848兲; and h = length of the and a T-bar penetrometer 共Stewart and Randolph 1991兲 of
drainage path. 20⫻ 5 mm. In the second sample 共Box 2兲, four T-bars of different
Independent measurements of cv were also obtained from lengths—20, 30, 40 and 50 mm, all with the same diameter of
CRSC and Rowe cell tests, the results of which are presented 5 mm—were tested.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2006 / 1189

 Fig. 4. Penetration resistance profiles of constant rate 共1 mm/ s兲 tests
for various shaped model penetrometers

depth of travel only allowed a one-diameter advance at each step,

which may be insufficient to achieve the steady-state resistance.
However, it was found sufficient to identify the point at which
Fig. 2. CRSC tests: relationship between cv and ␴⬘v: 共a兲 tube samples partial consolidation effects become apparent and, as the results
extracted from Box 1; 共b兲 tube samples extracted from Box 2 presented later will show, the penetration resistances at the end of
each step stay relatively close to those for the T-bar.
During completion of the final step of the twitch tests, the
Both constant rate 共“normal”兲 and variable rate 共“twitch”兲 pen- penetrometer was either pulled out immediately at a rate of
etration tests were conducted for each penetrometer. In a normal 1 mm/ s, or pushed in further at the original rate of 1 mm/ s be-
test, the penetrometer was pushed in and pulled out at a rate of fore being extracted at the same rate.
1 mm/ s. The use of a nondimensional velocity, as in Eq. 共1兲, overcomes
A twitch test commenced by first pushing the penetrometer scale effects in respect of consolidation at model and prototype
through the upper lightly overconsolidated layer at 1 mm/ s down scales, but it is also necessary to consider scale effects due to
to a depth of 60– 110 mm; the rate was then successively halved shear strain rates. As will be discussed later, the shear strain rate
over seven steps to a final value of 0.0078 mm/ s. Each succeed- is determined by the ratio of velocity to diameter 共v / d兲. The
ing step was triggered after the penetrometer had advanced by model penetrometers were advanced at a maximum rate of be-
either 1 or 2 diameters. Discrete steps for the penetration rate tween 0.1 共ball, plate兲 and 0.2 共T-bar兲 diameters per second, com-
reduction were used, rather than a continuous and smooth de- pared with typical field rates of 0.2 to 0.5 diameters per second.
crease in penetration rate, so as to overcome the effects of creep The effect of shear strain rate on the penetration resistance will
and to allow the penetration resistance to develop fully and therefore be slightly lower in the model tests compared with a
achieve an approximate steady state 共Randolph and House 2001兲. field penetration test.
It has been found that advancing the T-bar by two diameters is
sufficient to reach a steady-state resistance 共House et al. 2001兲.
For the larger-diameter ball and plate penetrometers, the available Resistance Profiles for Various Penetrometers

Normal Penetration Tests

Fig. 4 presents the resistance profiles from the constant rate 共nor-
mal兲 penetration tests for the different model penetrometers. Note
that the resistance profiles are plotted against the equivalent pro-
totype depth, where 1 mm in the centrifuge represents 0.1 m at
prototype scale.
The measured cone resistance, qc, should be corrected appro-
priately for the effects of unequal pore pressure 共Baligh et al.
1981; Campanella et al. 1982兲. Although the pore pressure at
the cone shoulder was not measured, since there was no pore
pressure transducer in the 10-mm diameter model cone, the net
cone resistance, qcnet, may be estimated using the following equa-
Fig. 3. Rowe cell tests: relationship between cv and ␴⬘v tion 共Robertson and Campanella 1983; Watson et al. 1998兲

1190 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2006

 Fig. 6. Penetration resistance profiles of normal and twitch tests for
Fig. 5. Penetration resistance profiles of normal 共1 mm/ s兲 tests for T-bar 20⫻ 5 mm
Model T-bars with various aspect ratios

normal test. The end of each step of the twitch test is also indi-
cated in the figure.
qc − 共␴⬘v + uo␣兲
qcnet = 共3兲 It may be seen in Fig. 6 that the T-bar twitch test initially
1 − 共1 − ␣兲Bq shows similar 共but slightly lower兲 penetration resistance to the
where ␴⬘v⫽estimated effective vertical stress; uo⫽estimated hy- normal tests at 1 mm/ s down to a rate of 0.062 mm/ s from where
drostatic water pressure; ␣ = unequal area ratio; and Bq⫽ratio the twitch test starts to show resistance increasing more strongly
of the excess pore pressure to the net bearing pressure. The value with depth than the normal test. At the end of the final step, where
of Bq was estimated to lie in the range 0.45–0.6, with the lower the rate is 0.0078 mm/ s, the resistance of the twitch test is ap-
limit corresponding to field measurements in natural clay at the proximately 78% higher than that of the normal test. As the pen-
Burswood site and the upper limit being typical for normally etration rate of the twitch test reverted to 1 mm/ s at completion
consolidated reconstituted clay 共Chung and Randolph 2004; of the final step, the resistance continues to increase over a short
Chung 2005兲. In practice, however, the net cone resistance is distance before reducing rapidly, forming a peak of resistance
relatively insensitive to Bq in this case, since the calibrated area between depths of 14 and 16 m. The resistance then slowly
ratio for the model cone penetrometer is relatively high, with merges with the normal test result. This phenomenon has also
␣ = 0.86 共Chung 2005兲, and a value of Bq = 0.45 was adopted. been reported in centrifuge tests by House et al. 共2001兲 and field
Tip resistances for the model T-bar, ball, and plate 共“full- data 共Schneider et al. 2004; Randolph 2004兲.
flow”兲 penetrometers were not corrected, since the area of these Fig. 7 shows the resistance profiles of the twitch tests con-
penetrometers is much larger than the area of the penetrometer ducted for the various shaped model penetrometers. The profile of
shaft so that the correction becomes insignificant 共Chung and
Randolph 2004; Chung 2005兲.
As may be seen in Fig. 4, the T-bar and ball penetrometers
demonstrate very similar resistance profiles. The plate shows
共unexpectedly兲 lower resistance than the T-bar and ball, while the
cone resistance appears to be the lowest.
Fig. 5 presents the resistance profiles of the normal tests for
the T-bars with aspect ratios 共L / d兲 from 4 to 10. It is evident that
all the T-bars show very similar resistances, suggesting that the
aspect ratio does not have any obvious effect on the T-bar resis-
tance, at least for L / d of 4–10.
Results for the “normal” constant rate tests have been dis-
cussed in detail by Chung and Randolph 共2004兲 and Chung
共2005兲. This paper concentrates on interpretation of the twitch test
results and discussion of partial consolidation and viscous effects
in clayey soils. Penetration resistance profiles presented in Figs. 4
and 5 have been used as reference values for normalizing corre-
sponding twitch tests.

“Twitch” Penetration Tests

The resistance profile of the variable rate 共twitch兲 penetration test Fig. 7. Penetration resistance profiles of twitch tests for various
for the 20⫻ 5 mm T-bar is presented in Fig. 6, together with its shaped model penetrometers

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2006 / 1191

Fig. 8. Penetration resistance profiles of twitch tests for Model T-bars
with various aspect ratios Fig. 9. Evaluation of cv using twitch test data of T-bar 20⫻ 5 mm

bunched, suggesting that the aspect ratio in the range 4–10 does
the normal T-bar test is also plotted in the figure as a broad ref- not affect the T-bar tip resistance, even at low penetration rates
erence for the tip resistance under undrained conditions. where the conditions are partially drained.
As may be seen in Fig. 7, the T-bar and ball penetrometers
exhibit very similar resistances even at different penetration rates
共at different stages of the twitch tests兲, implying that the two
Interpretation of Twitch Test Data
penetrometers experienced similar degrees of local strengthening
of the surrounding soil. The reasons may be attributed to the
relative size and flow pattern of the two penetrometers. The di- Evaluation of Consolidation Coefficient
ameter of the model ball 共11.9 mm兲 is considerably larger than Fig. 9 plots the “backbone” curves for T-bar penetrometer derived
that of the Model T-bar 共5 mm兲, so a given penetration rate gives independently by Watson and Suemasa 共unpublished data, 2000兲
a larger normalized velocity, V = vd / cv, suggesting lower veloci- and based on results presented by Randolph and Hope 共2004兲.
ties for a given degree of partial consolidation. However, consoli- The backbone curves can be expressed as
dation under conditions of axial symmetry occurs more rapidly
than for plane strain 共Davis and Poulos 1972兲, and this appears to q b
compensate for the larger diameter in this case. The consequence =a+ 共4兲
qref 1 + cVm
is that both of the penetrometers show similar tip resistances re-
gardless of the penetration rate, although this would only be the where q = penetration resistance at any rate; qref = reference 共un-
case for this particular relative size 共or diameter ratio兲 between drained兲 resistance; V = nondimensional velocity; and a, b, c, and
the T-bar and ball penetrometers. m = constants for the backbone curves 共see Table 2兲. As V ap-
The plate twitch test gave lower resistance than the T-bar and proaches zero, the penetration resistance will approach a value
ball twitch tests, which is consistent with the normal tests. It relevant for drained conditions. Therefore, the sum of a and b
should be noted that the peak of resistance, which is seen in the can be considered as the ratio of drained to undrained penetration
T-bar resistance profile, is not observed in profiles for the plate resistance. A relatively constant ratio of drained to undrained
and ball. This is because they were pulled out immediately 共with- penetration resistance has been observed for clayey soils in the
out further penetration兲 at completion of the final step of the centrifuge, but it is noted that this ratio is likely to be influenced
twitch tests. by friction angle and other soil parameters.
The net resistance for the cone twitch test has been calculated Also plotted in Fig. 9 are the normalized resistances of the
using Eq. 共3兲, with an initial Bq of 0.45. As the penetration rate is twitch tests for the 20⫻ 5 mm T-bar. The normalized resistances
decreased during the twitch test, the conditions change from
undrained to partially drained, leading to a reduction in Bq. In
correcting for the pore pressure at the cone shoulder, the value of Table 2. Summary of Constants Derived for the Backbone Curves
Bq has been assumed to decrease linearly with each reduction in Parameters
penetration rate, to 50% of its initial value in the final step. This
Type of test and source a b c m
assumption is based on the Bq results presented by Randolph and
Hope 共2004兲, where Bq decreases to half its initial value as the T-bar: Watson and Suemasa, 1 2.77 0.57 1.45
rate is decreased by about two orders of magnitude. As seen pre- unpublished data, 共2000兲
viously in the normal test results, the cone twitch test also gave T-bar: after Randolph and Hope 共2004兲a 1 2.1 1.1 1.6
the lowest resistance 共Fig. 7兲. Cone: after Randolph and Hope 共2004兲a 1 2.65 1.5 0.83
a
Fig. 8 presents twitch test results for T-bars with different as- Constants derived based on data from Randolph and Hope 共2004兲,
pect ratios. The resistance profiles of the various T-bars are tightly without correction for viscous effects.

1192 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2006

 Fig. 10. Evaluation of cv using data from cone twitch test Fig. 11. Normalized resistance profiles for various shaped
penetrometers

are computed as the ratio of the resistance developed at the end of

each stage of the twitch test to the corresponding reference resis- the rate of local consolidation is relatively higher around the ball
tance at 1 mm/ s from normal T-bar tests. Since the penetration than around the T-bar 共for a given diameter兲 by a ratio of about
rate and the probe’s diameter are known, the only variable in V is 2.5. Therefore, if the ball diameter is two to three times the T-bar
cv 关see Eq. 共1兲兴. The value of cv can be obtained by adjusting it to diameter 共the actual ratio dball / dT−bar being 2.38兲, the absolute
fit the twitch test data onto a backbone curve, with the value that consolidation rates for the two penetrometers will be similar, as
gives the best fit corresponding to the average cv for the depth reflected by the similar resistance profiles previously noted in the
over which the twitch test is conducted. The best fit has been twitch tests for the two penetrometers 共Fig. 7兲.
obtained using a least-squares approach, based on the difference Although, to the authors’ knowledge, no analysis has yet been
between the normalized resistance and the backbone curves for undertaken to study consolidation rates for the penetration prob-
each data point. lems, the rates of consolidation under rectangular and circular
As illustrated in Fig. 9, the cv values deduced from the curve surface footings may be used as an analogy to the rates of con-
fitting method using the T-bar backbone curve from Watson and solidation around the T-bar and ball penetrometers during pen-
Suemasa 共unpublished data, 2000兲 are 0.45 and 0.4 m2 / year for etration. Theoretical solutions presented by Davis and Poulos
the first 共Box 1兲 and the second 共Box 2兲 centrifuge samples, re- 共1972兲 showed that, for an infinitely deep layer, the rate of con-
spectively. However, using the curve from Randolph and Hope solidation under a surface footing is determined by the area of the
共2004兲, higher cv values of 1 and 0.9 m2 / year are obtained for footing and is virtually independent of the shape. Likewise, the
sample Boxes 1 and 2, respectively. similar resistance profiles for the T-bar and ball seen earlier in
Randolph and Hope 共2004兲 also derived a backbone curve for Fig. 7 may be owing to the similar projected areas of the two
the cone penetrometer, as shown in Fig. 10. Similarly, normalized penetrometers 共100 mm2 for the T-bar compared to 111.2 mm2
resistances from the cone twitch test have been fitted to the curve, for the ball兲. With that reasoning, it may be more appropriate to
resulting in a cv value of 0.6 m2 / year for sample Box 1. plot the normalized resistances against V⬘ = vde / cv, where de is
The values of cv deduced from the twitch test results, based on the diameter for a circle with area equivalent to the penetrometer,
the T-bar backbone curve from Watson and Suemasa 共unpublished rather than against V based on the actual diameter. Given the deep
data, 2000兲 and the cone backbone curve from Randolph and embedment of the penetrometers, it might also be argued that an
Hope 共2004兲, appear to be reasonably close to the cv ranges esti- equivalent diameter based on the volume of the T-bar or ball
mated using data from consolidation of the samples in the centri- should be adopted, which would weight the equivalent diameter
fuge, Rowe cell tests, and CRSC tests at high stress levels, as of the T-bar more toward that of the cylinder. However, for the
presented previously. present, given the similar resistance profiles for the T-bar and ball
in Fig. 7, an equivalent diameter based on the projected area
rather than the volume will be adopted.
Effect of Partial Consolidation for Various
Normalized resistances for the ball and various T-bars with
Penetrometers
projected areas from 100 to 250 mm2 are plotted in Fig. 12 nor-
Fig. 11 plots the normalized resistances against nondimensional malized resistance profiles for ball and T-bars with various aspect
velocity, V, for the different shaped penetrometers. For the pur- ratios 共a兲 against V; and 共b兲 against V⬘. Note that the backbone
poses of comparison, a single value of cv = 0.45 m2 / year has been curve of Watson and Suemasa 共unpublished data, 2000兲 must be
adopted for all penetrometers. adjusted accordingly in the plot against V⬘. The original profile
As may be seen in Fig. 11, the normalized resistance profile for the T-bar will be shifted horizontally to the right by a factor of
for the ball penetrometer plots horizontally to the right of the 2.26 共=de / d for the T-bar兲 and will almost coincide with the ball
profile for the T-bar by a factor of about 2.5. This suggests that profile, which is not affected by V⬘. It may be seen that the nor-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2006 / 1193

 Fig. 13. Rate effects in the undrained 共viscous兲 region

reason may be due to the fact that soil is displaced by the cone but
is able to flow around the ball and plate, thus resulting in a more
extensive field of excess pore pressure generated during penetra-
tion around the cone than around the ball and plate. This results in
the cone requiring lower penetration rates for a given effect of
partial consolidation compared to the ball and plate.

Effect of Penetration Rate in Viscous

„Undrained… Region
Mitchell 共1976兲 suggested a hypothetical relationship between
undrained shear strength and strain rate expressed in the form of
an inverse hyperbolic sine function. Based on the suggested rela-
tionship, the following expression is derived to define viscous
effects on penetration resistance

␭
1+ · sinh−1共␥˙ /␥˙ o兲
q ln共10兲
= 共5兲
qref ␭
Fig. 12. Normalized resistance profiles for ball and T-bars with 1+ · sinh−1共␥˙ ref /␥˙ o兲
various aspect ratios: 共a兲 against nondimensional velocity, V; 共b兲 ln共10兲
against V⬘ = vde / cv where ␭ = rate parameter 共typically 0.1—0.2: Randolph 2004;
Einav and Randolph 2005兲, ␥˙ = strain rate, ␥˙ o = strain rate for
which the viscous effects start to decay; and ␥˙ ref = reference strain
malized data appear to fit relatively better to the adjusted back- rate corresponding to qref .
bone curve in the plot against V⬘ 关Fig. 12共b兲兴 compared to the Note that ␥˙ will be a function of the ratio v / d of penetration
original backbone curve in the plot against V 关Fig. 12共a兲兴. This rate to diameter of the penetrometer. It is also the only variable
implies dependency of the consolidation rate on the projected area in Eq. 共5兲. Although cv will influence the value of v 共perhaps also
of the penetrometer, rather than its physical diameter, although ␥˙ 兲 at the transition point from undrained to partially drained,
more experimental results are necessary to verify this observation. within the undrained region, ␥˙ 共or v / d兲 is the primary governing
It is interesting to note that the plate penetrometer demon- parameter for viscous effects.
strates very similar normalized resistance response to the ball Fig. 13 plots the data from all model T-bar twitch tests in terms
共Fig. 11兲, even though its “actual” penetration resistances are of normalized resistance against v / d. Data from the field T-bar
lower than for the T-bar and ball both in the normal and twitch twitch tests reported by Schneider et al. 共2004兲 are also plotted in
tests 共Figs. 4 and 7兲. This suggests that the ball and plate have the figure for comparison. It may be seen that the model T-bar test
similar rates of local consolidation around them, despite the round data are rather scattered in the undrained region; some model tests
base for the ball compared to the flat base for the plate. demonstrate very marginal decrease in normalized resistance as
In Fig. 11, the normalized cone resistance profile plots to the the penetration rate is decreased. Perhaps this is partly due to the
left of the ball and plate profiles, suggesting a lower rate of con- reference v 共or reference V兲 corresponding to qref for the model
solidation around the cone than around the ball and plate. The tests being rather close to the transition point.

1194 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2006

 of d and de, varying the value of cv will also influence the viscous
portion of the “modified” backbone curve in the plot with a hori-
zontal axis of V⬘.
However, the value of cv may be estimated by fitting the field
data with the modified backbone curve. This results in a cv value
of 1.5 m2 / year 共Fig. 14兲. This value is slightly higher than the
values measured from oedometer tests on undisturbed field
samples reported by Chung 共2005兲, which were 1 to 1.25 m2 / year
at the yield stress. A similar procedure can be performed using the
backbone curve from Randolph and Hope 共2004兲, although a
higher cv will be obtained.
It should also be pointed out that the partially drained portion
of the modified backbone curve of Watson and Suemasa 共unpub-
lished data, 2000兲 has been shifted to the left from its original
position 共Fig 14兲. This is due to the influence of the multiplier for
viscous effects 关Eq. 共6兲兴 extending to the partially drained portion
of the curve. One option to avoid this would be to derive the
constants of the backbone curves after incorporating viscous
effects.

Fig. 14. Fitting results of field T-bar twitch tests

Conclusions

By contrast, the field T-bar twitch tests show clear reduction in This paper has presented results of model twitch tests performed
normalized resistance as the penetration rate is reduced. This is in the centrifuge for various-shaped penetrometers. The effects of
due to the fact that the field T-bar tests have two-orders-of- penetration rate on the tip resistance both in partially drained and
magnitude greater “undrained” range than the model T-bar tests, undrained regions have been discussed.
since the product 共vd兲field corresponding to qref for the field T-bar It has been shown that, by fitting data of the twitch tests to
tests is 160 times greater than 共vd兲centrifuge for the model T-bar backbone curves derived by Watson and Suemasa 共unpublished
tests. Therefore, the field tests can define viscous effects much data, 2000兲 and Randolph and Hope 共2004兲, it is possible to de-
better than the centrifuge tests. duce a cv value reasonably close to the values estimated from the
The contrasting ratios of the product vd for field and centri- data from sample consolidation in the centrifuge, Rowe cell tests,
fuge tests accounts for the fact that the value of v / d correspond- and oedometer tests at high stress levels. There is some variation
ing to the minimum normalized resistance for the field T-bar tests among the experimentally derived backbone curves, and this
is significantly lower than that for the model T-bar tests, by a leads to a range of a factor of about 2 in the deduced cv values.
factor of between 10 and 30. Analytical or numerical confirmation of the position of the back-
Both the field and centrifuge test data in the undrained region bone curves would be beneficial in reducing this uncertainty.
have been fitted with the curves computed using The rate of consolidation around a “full-flow” penetrometer
共T-bar, ball, or plate兲 appears to be determined solely by the size
␭ or projected area of the probe, but virtually independent of its
1+ · sinh−1共v/d␥˙ o兲
q ln共10兲 shape. Hence, it may be argued that the rate of penetration should
= 共6兲 be normalized to nondimensional velocity using the diameter of a
qref ␭
1+ · sinh−1共vref /d␥˙ o兲 circle with projected area equivalent to the penetrometer, instead
ln共10兲 of the physical diameter of the penetrometer.
as shown in Fig. 13. Following a least-squares approach, it is There is still limited understanding of the parameters govern-
found that, taking ␭ = 0.18 and ␥˙ o = 0.061 s−1, the curve fits very ing viscous effects. Nevertheless, an expression based on an in-
well to the field data. However, the centrifuge data are fitted bet- verse hyperbolic sine function was used to quantify viscous
ter with ␭ = 0.2 and ␥˙ o = 0.23 s−1, although the quality of fit is effects on the penetrometer tip resistance. The expression was
uncertain due to the scatter observed in the limited undrained calibrated using data from field T-bar twitch tests, taking the
region. strain rate as proportional to the ratio, v / d, of penetration rate to
Viscous effects may be incorporated in the backbone curves of diameter. Further study is required on the effects of strain rate on
Watson and Suemasa, unpublished data, 共2000兲 and Randolph and penetration resistance, and how the rate for minimum resistance is
Hope 共2004兲 by multiplying the inverse hyperbolic function of affected for materials of differing rate dependency.
Eq. 共6兲 by the right-hand side of Eq. 共4兲. This has been performed
using the backbone curve from Watson and Suemasa 共unpub-
lished data, 2000兲 and the parameters for viscous effects 共␭ and Acknowledgments
␥˙ o兲 calibrated based on the data from the field T-bar twitch tests.
The resulting curve is plotted in term of normalized resistance This work forms part of the activities of the Center for Offshore
against “alternative” nondimensional velocity, V⬘共=vde / cv兲, as Foundation Systems 共COFS兲, established and supported under the
shown in Fig. 14. Note that the physical diameter, d, rather than Australian Research Council’s Research Centers Program. The
de, is used in the viscous part of the “combined” equation, since first and third authors hold International Postgraduate Research
v / d determines the average strain rate. Also note that, since the Scholarships and University Postgraduate Awards from the Uni-
relationship between V⬘ and v / d is a function of cv for a given set versity of Western Australia. Acknowledgment is made of the

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2006 / 1195

excellent technical support for the centrifuge and field penet- 217–230.
rometer tests, principally Mr. Don Herley and Mr. Alex Duff, and House, A. R., Oliveira, J. R. M. S., and Randolph, M. F. 共2001兲. “Evalu-
also of the advice provided by Professor Marcio Almeida of the ating the coefficient of consolidation using penetration tests.” Int. J. of
Federal Univ. of Rio de Janeiro in the planning and execution of Physical Modelling in Geotechnics, 1共3兲, 17–25.
the centrifuge tests. Mitchell, J. K. 共1976兲. Fundamentals of soil behavior, Wiley, New York.
Randolph, M. F. 共2004兲. “Characterization of soft sediments for offshore
applications.” Proc., ISC-2 on Geotechnical and Geophysical Site
Characterization, A. Viana da Fonseca and P. W. Mayne, eds.,
References
Millpress, Rotterdam, 209–232.
Randolph, M. F., and Hope, S. 共2004兲. “Effect of cone velocity on cone
Baligh, M. M., Azzouz, A. S., Wissa, A. Z. E., Matyin, R. T., and resistance and excess pore pressures.” Proc., Int. Symp. on Engineer-
Morrison, M. H. 共1981兲. “The piezocone penetrometer.” Proc., Cone ing Practice and Performance of Soft Deposits, Osaka, Japan,
penetration testing and experience, ASCE, New York, 247–263. 147–152.
Bemben, S. M., and Myers, H. J. 共1974兲. “The influence of rate of pen- Randolph, M. F., and House, A. R. 共2001兲. “The complementary roles of
etration on static cone resistance in Connecticut River Valley varved physical and computational modelling.” Int. J. of Physical Modelling
clay.” Proc., European Symp. on Penetration Testing, ESOPT, in Geotechnics, 1共1兲, 1–8.
National Swedish Building Research, Stockholm, 2.2, 33–34. Randolph, M. F., Jewell, R. J., Stone, K. J. L., and Brown, T. A. 共1991兲.
Campanella, R. G., Robertson, P. K., and Gillespie, D. 共1982兲. “Pore “Establishing a new centrifuge facility.” Proc., Int. Conf. on Centri-
pressure during cone penetration.” Proc., 34th Canadian Geotechnical fuge Modelling, Boulder, Colo., 2–9.
Conf., Canadian Geotechnical Society, 5.2.1-14 . Robertson, P. K., and Campanella, R. G. 共1983兲. “Interpretation of cone
Chung, S. F., 共2005兲. “Characterisation of soft soils for deep water devel- penetrometer tests. Part II: Clay.” Can. Geotech. J., 20, 734–745.
opments.” Ph.D. thesis, Univ. of Western Australia, Crawley, Roy, M., Tremblay, M., Tavenas, F., and La Rochelle, P. 共1982兲. “Devel-
Australia. opment of pore pressures in quasi-static penetration tests in sensitive
Chung, S. F., and Randolph, M. F. 共2004兲. “Penetration resistance in soft clay.” Can. Geotech. J., 19共2兲, 124–138.
clay for different shaped penetrometers.” Proc., ISC-2 on Geotechni- Schneider, J. A., Randolph, M. F., and Chung, S. F. 共2004兲. “Character-
cal and Geophysical Site Characterization, A. Viana da Fonseca and ization of soft soils for deep water developments.” Report on variable
P. W. Mayne, eds., Millpress, Rotterdam, 671–677. rate penetration tests at Burswood site, Centre for Offshore Founda-
Davis, E. H., and Poulos, H. G. 共1972兲. “Rate of settlement under two- tion Systems, Univ. of Western Australia, Crawley, Australia.
and three-dimensional conditions.” Geotechnique, 22共1兲, 95–114. Stewart, D. P., and Randolph, M. F. 共1991兲. “A new site investigation tool
Einav, I., and Randolph, M. F. 共2005兲. “Combining upper bound and for the centrifuge.” Proc., Int. Conf. on Centrifuge Modelling,
strain path methods for evaluating penetration resistance.” Int. J. 531–538, Balkema, Rotterdam.
Numer. Methods Eng., 63共14兲, 1991–2016. Watson, P. G., Newson, T. A., and Randolph, M. F. 共1998兲. “Strength
Finnie, I. M. S., and Randolph, M. F. 共1994兲. “Punch-through and lique- profiling in soft offshore soils.” Geotechnical site characterization,
faction induced failure of shallow foundations on calcareous sedi- P. K. Robertson and P. W. Mayne, eds., Balkema, Rotterdam,
ments.” Proc., Int. Conf. on Behavior of Offshore Structures, Boston, 1389–1394.

1196 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2006