Been 2010 CPT in Practice CPT10

2nd International Symposium on Cone Penetration Testing, Huntington Beach, CA, USA, May 2010
Interpretation of the CPT in engineering practice

K. Been & A. Quiñonez
Golder Associates Inc., Houston, TX, USA
R.B. Sancio
Geosyntec Consultants, Inc., Houston, TX, USA
ABSTRACT: Many empirical and theoretical CPT interpretation methods are broadly
accepted and used in practice. These approaches tend to consider whether the cone
penetration is drained or undrained, and then will consider the soil as either “sand” or
“clay,” respectively. Most fundamental research into the CPT and its interpretation
considers penetration through sands or clays separately, and includes verification tests
in materials with close to ideal sand or clay behavior. However, in our engineering
practice over the past five years, we have encountered several instances where the
soils are not typical and the usual methodologies are inadequate or lead to inconsis-
tent conclusions. This paper describes three of these case histories and our approach
to using the CPT in each case.
1 INTRODUCTION
Soil behavior is in general complex. In addition to physical interactions at the particu-

late or clay mineral scale, there is coupling of the pore fluid flow with the solid ma-
trix deformation; however, only macro-behavior is generally observed in soils testing.
In triaxial testing, for example, we impose well-controlled and known boundary con-
ditions on soil samples to measure the shear strength of the sample directly. Cone pe-
netration into soils imposes a completely different and complex set of boundary con-
ditions, and we take indirect measurements of the soils’ response to these imposed
boundary conditions (i.e. parameters such as tip resistance or sleeve friction) to de-
termine the shear strength of the soil. As a result, interpreting the cone penetration
test (CPT) is largely empirical, with the best empirical approaches having a sound
theoretical basis. Relatively simple soil and interpretation models tend to gain the
widest acceptance, but also fail to capture some of the complexities of soil behavior
and the imposed CPT boundary conditions. Robertson (2009) implicitly captures this
dilemma and presents an overview of currently available models for CPT interpreta-
tion. Robertson (2009) suggests that the empirical models are suitable as a screening
tool for “low risk” projects. However, in our geotechnical practice, we seldom en-
counter “low risk” projects where these screening methods are sufficient.
In general, our projects are large, involve capital expenditures in the hundreds of
millions of dollars, and tend to be in remote areas of the developing world, rather
than in developed urban areas of North America and Europe. Due to the scale and lo-
cation, the cost of site investigation for our projects is high, but contracting agencies
still have a tendency to squeeze the budget and minimize spending on geotechnical
testing. However, the most important consequences for CPT interpretation stem from
the location of our projects. First, because of the remote locations, it is difficult to get
high-quality laboratory testing on undisturbed samples. Ideally testing would be car-
ried out in field laboratories on site, but it is costly to exercise good quality control
and to provide geotechnical oversight of the testing on site. The preferred option is to
ship samples to permanent laboratories for advanced testing and to pay the (un-
known) penalty of sample disturbance during transport. Another significant issue in
remote sites is a complete lack of local experience and prior knowledge of what me-
thods and approaches are suitable for the soils in the region; for example, what typi-
cal values of Nkt, the coefficient relating tip resistance to undrained shear strength in
Equation 1, might be appropriate.
Our experience is therefore that site-specific relationships for the CPT interpreta-
tion are always needed to determine soil properties from the CPT. The CPT is a very
good tool to give an indication of soil behavior, continuous vertical profiles, variabili-
ty between locations, and of course undrained shear strength of clays and density or
state of sandy soils. However, the CPT can only be used with confidence when sup-
ported by all of the other tests and information at our disposal from the site investiga-
tion. There have been several instances where the soils are not typical sands or clays,
and the usual methodologies are inadequate or lead to inconsistent conclusions. In
this paper, we will discuss three examples where traditional CPT methodologies were
not sufficient for the location of the project or the type of soils encountered. We will
present the data and discuss our methodologies for using the CPT results in conjunc-
tion with the other tests.
2 RESIDUAL SOIL SITE
The first example is a site where the soils of interest were residual soils derived from
weathered basalt. The soil column is typically 15 m thick overlying the weathered ba-
salt bedrock profile. Near the ground surface, these soils lack texture or traces of the
parent rock and consist of brown to red and yellow, high plasticity silts. At a slightly
greater depth, the residual soils transition to weathered basalt and are typically dark
brown, black and dusky red, high plasticity silts. The liquid limit of all these soils is
generally greater than 50, and the plasticity index is such that they plot below the A-
line on the Casagrande plasticity chart. According to the Unified Soil Classification
System (USCS) in ASTM D2487, these soils have the abbreviated descriptor MH,
and are called elastic silts (or high plasticity silts outside the USA).
2.1 Index Tests

Figure 1 shows the index properties of the upper 15 m of the soil profile, including
the SPT N value and the undrained shear strength. Undrained shear strength was gen-
erally measured with unconsolidated undrained (UU) triaxial compression tests in a
field laboratory, but Figure 1 also presents results from several pressuremeter tests,
which are shown as hollow squares. The moisture content ranges between 19% and
79% and increases slightly with depth. The liquid limit (LL) ranges from 59 to 119,
and the plasticity index (PI) ranges from 9 to 47. The average undrained shear
strength (cu) measured in UU tests is approximately 120 kPa, with relatively wide
scatter and a standard deviation of 88 kPa. The high standard deviation is an indica-
tion of a large variability within this residual soil, which may be attributed to sam-
pling disturbance, failure along micro discontinuities, natural variability, and maybe
other factors. It is also noted that stiffer soils in the profile are difficult to sample, and
therefore UU tests are considered to be representative of the weaker parts of the soil
column. The groundwater table at this site was at a depth of about 8 m at the time of
drilling and CPT testing.
Water content (%) Liquid limit Plasticity index SPT N-value cu (kPa)
0 20 40 60 80 20 60 100 140 0 10 20 30 40 50 0 10 20 30 40 0 200 400 600
0
Depth below ground surface (m)
10
15
Figure 1. Soil properties at the residual soil site in West Africa (in undrained shear strength plot circles
indicate UU test results and squares indicate pressuremeter test results).
2.2 CPT Data

Several CPTs were carried out at the site, but Figure 2 shows only one CPT. This
CPT was situated close to the soil borings sampled to provide the index properties
and undrained shear strengths on Figure 1. Figure 2, shows the tip resistance (qt, cor-
rected for end area effects) as well as the common normalized parameters; friction ra-
tio (F), pore pressure parameter (Bq), normalized tip resistance (Q) and soil behavior
index (Ic). Both the Robertson and Wride (1998) and the Jefferies and Davies (1993)
versions of Ic are shown, illustrating that in this case there is very little difference be-
tween the two (because the main difference between the two equations for Ic is that
Jefferies and Davies include the pore pressure parameter Bq which is generally less
than about 0.1 for these soils). In the upper 5 m, the behavior type is between “clayey
silt to silty clay” and “clay”, and below 5 m the behavior type is solidly in the “clay”
zone, according to the Robertson and Wride (1998) classification. (The Robertson
and Wride classification is preferred in this instance because of the potential for de-
saturation of the pore pressure stone during penetration above the ground water lev-
el). Figure 3 shows the CPT measurements on a soil behavior chart, which further il-
lustrates the soil classification.
qt (MPa) F (%) Bq Q Ic (J&D) Ic (R&W)

0 1 2 3 4 5 0 2 4 6 8 10 -0.2 0.0 0.2 0 40 80 120 1 2 3 4 1 2 3 4
0
organic soil
sands: clean to silty
5
silty sand to sandy silt

Depth (m)
organic soil
silty sand to sandy

clayey silt to silty clay
10

clays
clays
15
Figure 2. CPT profile at residual soil site (soil behavior type Ic shown for both Jefferies and Davies,
1993, and Robertson and Wride, 1998, equations)
1000
8
7 1. Sensitive fine grained
9
2. Organic soils – peats
3. Clays – clay to silty clay
Normalized Tip Resistance, Q
100 4. Silt mixtures – clayey silt to silty

6 clay
5. Sand mixtures – silty sand to sandy
silt
5
6. Sands – clean sand to silty sand
10 4 7. Gravelly sand to sand
8. Very stiff sand to clayey sand
3 9. Very stiff fine grained
1
2
1
0.1 1 10
Friction Ratio, F (%)
Figure 3. Plot showing where data falls on the Robertson (1990) classification of soil behavior types.
The soil behavior type indicated by the CPT for this residual soil seems to be ap-
propriate despite being different from the soils used to develop the CPT charts. While
the material is classified as an elastic or high plasticity silt based on its Atterberg lim-
its, the mechanical response of the soil is similar to that of a clay, i.e. the engineering
parameter used for characterization of its strength during static loading is undrained
shear strength.
2.3 Engineering Interpretation of the CPT

This particular location was for a large (850,000 barrel) settlement-sensitive tank, and
the key foundation considerations were bearing capacity and settlement. Therefore,
the primary engineering parameters of interest from the CPT were undrained shear
strength and compressibility. While overconsolidation is a meaningless concept in a
residual soil, oedometer tests were carried out on the material to determine the com-
pressibility. These showed that the yield stress in compression (analogous to precon-
solidation pressure) was greater than double the design tank loading. Settlement was
therefore characterized by elastic parameters for the soil.
Undrained shear strength, cu (kPa) Shear modulus, G (MPa)

0 200 400 600 0 50 100 150
0
UU Triaxial Gmax from Vs
CPT Nkt=16 G/Gmax = 0.2
PMT BH427 Gur (BH 427)
PMT BH494 Gur (BH 494)
CPT 434
10
15
a) Undrained shear strength b) Shear modulus

Figure 4. Interpretation of CPT in residual soils; b) shows Gmax from Vs, 0.2 × Gmax, Gur from pres-
suremeter tests, and G from CPT tip resistance.
As illustrated in Figure 4a, a relatively simple technique was used to estimate Nkt,
the empirical parameter required to calculate undrained shear strength (cu) in triaxial
compression from CPT. Nkt is defined in Equation 1, where qt is the cone tip resis-
tance and σv is the total geostatic vertical stress. We varied Nkt until the CPT graph of
undrained shear strength with depth on Figure 4a provided the best visual match to
the UU tests on samples from nearby borings. A value of Nkt = 16 provided the best
match. It is apparent that the small number of pressuremeter measured undrained
shear strengths (abbreviated as PMT in Figure 4a) are higher than UU test results, and
this approach of using UU tests as a reference might be somewhat conservative in this
soil, i.e. Nkt could be less than 16.
qt  v
cu 
Nkt (1)
Given that the bearing capacity exceeded the design load even using conservative
values, settlement of the tank, and hence modulus of the residual soils, became the
focus of the interpretation. In situ test information on modulus was available from
shear wave velocity (Vs), pressuremeter tests, and CPT-based empirical relationships,
in addition to laboratory tests. As a first step, we noted that laboratory tests suggested
moduli that were significantly lower than those suggested from in situ tests. When we
identified this distinct difference between the results, we did not consider the labora-
tory values further when selecting design parameters, as they were likely affected by
sampling bias (i.e. selective sampling and sub-sampling of laboratory specimens) and
sample disturbance. Figure 4b shows the range of shear modulus estimates from:
• Gmax determined from downhole shear wave velocity measurements, Gmax =
ρVs2, where ρ is the mass density of the residual soil determined from tube sam-
ples.
• G from shear wave velocity with a degradation factor of G/Gmax = 0.2 to approx-
imate 1% shear strain.
• Gur determined from the slope of an unload/reload cycle during a pressuremeter
test.
• G from the CPT considering the constrained modulus, M, from Equations 2 and
3 with αM = Q when Q ≤ 14 and αM = 14 when Q > 14. (Robertson, 2009)
M (1  2 )
G (2)
21  
M   M qt   v  (3)
The final CPT-based method results in much lower modulus than the degraded
shear wave velocity method and the pressuremeter, while the pressuremeter and shear
wave velocity methods give similar results.
The above derivation of G from the CPT passes through the intermediate step of
constrained modulus M. However, there are direct correlations between Gmax and tip
resistance for cohesionless soils, for example Equation 4 (Robertson 2009)
Gmax   G qt   v   0.0188  10 0.55 I c 1.68  qt   v  (4)
While Equation 4 would not strictly be applicable for undrained, cohesive soils, Fig-
ure 5 shows a comparison between Gmax from Equation 4 and Gmax derived from the
shear wave velocity. As before, the CPT gives significantly lower values than from
shear wave velocity measurements, but not as low as when using Equations 2 and 3
for cohesive soils. Schnaid (2005) has suggested that such differences might be the
result of aging or cementation in cohesionless soils, which is also plausible for resi-
dual soils but caution is needed in its implementation. There may be factors related
to pore pressures and drainage that affect the interpretation in cohesive versus cohe-
sionless soils.
We decided to use the pressuremeter and shear wave-based values, since they re-
sult from direct measurement of shear modulus. In contrast, the CPT measures a
complex mix of shear strength and stiffness response to the cone penetration and the
CPT-based method is indirect. This choice to use the pressuremeter and shear wave
values was justified during subsequent monitoring of tank settlements during hydro-
testing and operation of the facility, as settlement predictions compared favorably
with measurements.
Shear modulus, Gmax (MPa)

0 50 100 150
0
10
Gmax from Vs
Gmax from CPT (equation 4)

15
Figure 5. Comparison of Gmax from for residual soils from a) shear wave velocity and b) CPT correla-
tion for cohesionless soils (Equation 4)
2.4 Summary
The residual soil case history is interesting in that CPT correlations are typically de-
veloped for sedimentary soils, not for residual soils. The soil behavior index, Ic, is
appropriate in this case; undrained clay-like behavior was observed in laboratory tests
and was correctly identified by the CPT, but the CPT was incapable of showing that
the material was a highly plastic silt, rather than a clay. The estimated Nkt value of 16
or less to determine undrained shear strength is well within the band of expectation
for CPTs in stiff clays and presents no surprise. However, the problem of using the
CPT to estimate modulus in an atypical soil is also highlighted. The CPT primarily
results in shearing of the soil, and its use to estimate modulus is limited, to some de-
gree, to how well modulus and undrained shear strength are related. Even for sedi-
mentary materials, there is a wide range of modulus to undrained shear strength ra-
tios, and there is no reason to believe that a residual soil should fall within the same
range. We do not recommend estimating soil stiffness using empirical correlations

from the CPT in a new area or in a soil for which there is no prior experience, even at
a screening level. However, in this case deploying a seismic CPT (SCPT) to obtain
measurements of the shear wave velocity would have been helpful to identify the
high stiffness relative to the measured tip resistance (e.g. Schnaid, 2005).
3 ESTUARINE CLAY
In this section, we present an example of a nearshore organic clay layer that was in-
vestigated for a large gas processing site. The engineering considerations for this
project included undrained shear strength for slope stability and compressibility.
CPTs and soil borings were advanced through the organic clay layer and samples re-
covered for testing. Index tests were carried out on most samples, however, it was
impossible to acquire sufficient undisturbed samples for laboratory compressibility
and strength testing to characterize the material fully. Given the paucity of data, we
relied on calculations of preconsolidation pressure and shear strength using CPT tip
resistance and empirical relationships, but verified that the choice of Nkt, k, and
SHANSEP parameters were compatible.
The clay was deposited in what was likely a mangrove area of a bay within the
estuary of a large river. Based on the condition of the soil, it is unlikely that the clay
was subjected to mechanical preconsolidation in this quiescent depositional environ-
ment. The clay was lightly overconsolidated due to aging and creep, but not due to
prior overburden stresses.
3.1 Index Tests

The clay was characterized by carrying out moisture content, Atterberg limits, and
particle size distribution tests. The “organic clay” (OH) designation was based on
visual observations of the presence of organic matter, measured organic content in la-
boratory tests, slight odor, and the comparison of the liquid limit value of the material
at its natural state with the liquid limit measured after oven-drying the specimen. The
moisture content of the material was highly variable (Figure 6), and the liquid limit
was typically greater than 50, reaching values as large as 160. The plasticity index
typically ranged between 60 and 101, and the material plotted slightly above or below
the A-line on the Casagrande plasticity chart (Figure 7). The liquidity index (LI)
ranged between 0.3 and 2.4 (Figure 6). In this instance the groundwater was above
the surface and the average unit weight of the clay was approximately 14 kN/m3, giv-
ing a relatively low submerged unit weight of 4.2 kN/m3 for CPT interpretation.
3.2 CPT Data

The plots on Figure 8 present typical values of the tip resistance qt, friction ratio F,
Bq, and Ic for the organic clay layer (which occurs in the upper 11.5 m of the profile).
The material exhibited increasing penetration resistance with depth (at an average
rate of about 50 kPa/m) as expected in this geological environment. The friction ratio
is between 2% and 3% and positive excess pore water pressure occurred during test-
ing such that typically 0.2 < Bq < 0.6. The soil behavior index varies between 2.8 < Ic
< 3); therefore, the material classifies as “silty clay” and “clay” according to both the
Jefferies and Davis (1993) and the Robertson and Wride (1998) systems. Figure 9
shows the organic clay on the Robertson (1990) soil behavior type chart, indicating
an accurate behavior type classification. (The material is not a peat and the organic
content is relatively small.)
Moisture content (%) Unit weight (kN/m 3)

Liquidity index (LI)
0 100 200 10 15 20
0 0 1 2 3
0
0
2 2
2
Depth below seabed (m)
4 4
4
6 6 6
8 8 8
10 10 10
12 12 12
14 14 14
Figure 6. Plot of moisture content, unit weight and liquidity index with depth, estuarine organic clay.
120
100
Plasticity index
CH
80
60
CL
40
20 MH
ML
0
0 50 100 150 200
Liquid limit
Figure 7. Plasticity chart showing location of organic clay on Casagrande plasticity chart.

The strength and compressibility of the organic clay were sought for engineering
evaluation. Given the dearth of laboratory test data, a value for Nkt (Equation 1) be-
tween 12 and 16 was initially selected, with a preferred value of 14. A value of k =
0.2 was then selected to estimate the preconsolidation pressure (σ´p) using Equation
5.
 ' p  k  qt  v  (5)
However, in clays, cu and σ´p, or the overconsolidation ratio (OCR = σ´p/σ´v), are
generally also related. Ladd and Foote (1974) empirically developed the relationship
in Equation 6, where (cu/σ´v )OCR =1 may also be designated as S. The values of the
SHANSEP parameters S and m should be calculated through curve-fitting laboratory
test data using the procedure recommended by Ladd and Foote (1974); however, in
this case, these parameters also had to be estimated given the shortage of strength test
data for the material.
 ' p 
m
cu  cu 
    (6)
 ' v   ' v  OCR 1   ' v 
qt (MPa) F (%) Bq Ic (J&D) Ic (R&W)

0 1 2 3 4 5 0 1 2 3 4 5 -0.2 0.2 0.6 1.0 1 2 3 4 1 2 3 4
0
5
silty sand to sandy

Depth (m)
10
Bottom of organic clay

clays
clays
15
organic soil
organic soil
20
Figure 8. Plots of tip resistance, friction ratio, normalized pore water pressure, and soil behavior index
(according to both the Jefferies and Davies and the Robertson and Wride systems).
1000
8 1. Sensitive fine grained

7
9
100 4. Silt mixtures – clayey silt to silty

6 clay
silt
5
10 4 7. Gravelly sand to sand
1 8. Very stiff sand to clayey sand
3 9. Very stiff fine grained
2
1
0.1 1 10
Figure 9. Organic clay CPT results on the Robertson (1990) classification of soil behavior types.
The term S varies slightly as a function of the failure mode (testing method, strain
rate), but may be considered to be constant in this case. Ladd and De Groot (2003)
recommend S = 0.25 with a standard deviation of 0.05 (for simple shear loading) and
m = 0.8 for sedimentary deposits of silts and organic soils (Atterberg limits plot be-
low the A-line), excluding peats and clays with shells. S may take a value of approx-
imately 0.30 to 0.35 for triaxial compression loading (e.g. Ladd and DeGroot, 2003).
Only certain combinations of the empirical parameters Nkt (tip resistance to cu), k
(tip resistance to OCR), and S and m are compatible with all of Equations 1, 4, and 5.
Substituting Equation 1 and 5 in Equation 6 to determine the appropriate combination
of parameters gives:
1 m
 qt  v 
Nkt  S  k   
m
(7)
  'v 
The left hand side of Equation 7 is a constant for a given soil layer, therefore, the
magnitude of this constant may be calculated by first selecting an appropriate value
for m and then by plotting the right hand side of the equation, as shown on Figure 10
for eight CPT soundings, where m = 0.85 was chosen. Significant variability in the
data is noticeable on Figure 10 due to variability of the tip resistance around the
project site. The median value of the right hand side of Equation 7 is 1.45, estimated
from the measured data which are plotted on Figure 10 and shown as a bold vertical
line. Therefore, the constants Nkt, S, and k on the left hand side of Equation 7 should
be chosen accordingly.
Since there is no single solution to Equation 7, different scenarios or combinations
of parameters were considered for engineering analyses. Table 1 shows combinations
of compatible parameters, where m = 0.85 in all cases, that were used to calculate cu
and σ´p for subsequent stability and settlement analyses. Nkt and S were selected and
the value of k calculated corresponding to the median of the data, Figure 10. Scenario
A has a typical value of S for strength in simple shear, but although k has a reasona-
ble value, the magnitude of Nkt appears to be too large for a soft, lightly overconsoli-
dated clay. The combination of Scenario B in Table 1 has a reasonable value of Nkt
and S for triaxial compression, but k appears to be high. Scenario C has high Nkt, rea-
sonable S for triaxial compression, but low k. Scenario D has reasonable Nkt and k,
but S appears to be too high. Consideration of these four scenarios provided bounds
on the settlement and stability calculations that would have been much broader if this
approach had not been adopted.
(qt-σv/σ'v)1-m and Nkt x S x km

1.0 1.2 1.4 1.6 1.8 2.0
0
2
Depth below Seabed (m)
10
12
14
Figure 10. Plot of CPT tip resistance term on the right hand side of Equation 7, with the median value
shown as a bold vertical line.
Table 1. Combinations of compatible empirical parameters for CPT interpretation.

Parameter Scenario A Scenario B Scenario C Scenario D
Nkt 20 14 20 14
S 0.25 0.30 0.30 0.36
m 0.85 0.85 0.85 0.85
k (median) 0.246 0.302 0.198 0.244
3.4 Summary
The standard of practice to develop empirical factors to correlate undrained shear
strength and preconsolidation pressure to CPT tip resistance is to use site-specific la-
boratory test data. However, with only a handful of unconsolidated undrained triaxial
tests and consolidation tests available, there was an insufficient amount of data.
Therefore, we found ourselves in the undesirable situation to have to rely heavily on
index test data, experience, and judgment to select appropriate values for Nkt, k, and
SHANSEP parameters. Obviously, interpreting the CPT without laboratory test data
is a process that introduces significant uncertainty in the choice of engineering para-
meters.
A procedure was developed to verify the compatibility of the choice for Nkt, k, and
SHANSEP parameters through the manipulation of Equations 1, 5, and 6. We rec-
ommend the application of this verification process to analyze CPT data, as it is use-
ful even when a good quantity and quality laboratory test data are available.
The value of the parameter Nkt typically depends on the value of OCR and clay
plasticity (e.g. Karlsrud et al., 2005). Given that the SHANSEP parameters depend
mostly on the index properties of the clay layer (and it is reasonable to assume that
they do not vary within a clay layer), then the value of k should also vary with OCR
within the layer, a somewhat circular proposition. Thus, it appears that neither k nor
Nkt should be constant in a clay with varying OCR (the changes in k and Nkt could,
however, be inversely proportional). The normalized tip resistance, Q, could be used
as a clue to changes in Nkt and k, but we have not explored this in any detail.
A comprehensive laboratory test program should always be completed to deter-
mine the values of Nkt, k, S, and m. A comprehensive laboratory program should in-
clude a sufficient number of oedometer tests to determine the overconsolidation pro-
file as well as undrained shear strength tests to determine the SHANSEP parameters
S and m for each clay soil. In addition, undrained shear strength measurements at in
situ conditions or on undisturbed samples consolidated to the estimated in situ
stresses are required to provide another (direct) check on Nkt and as a check on the
SHANSEP tests and procedures. Despite the experience on this particular project, the
CPT should not be used as the sole measurement.
4 OFFSHORE SILT
This example considers a thick offshore silt deposit in about 140 m of water where
we had information to a depth of about 100 m below the seabed. The full depth pene-
trated consisted of recent marine sediments, which typically presented as grayish
green silt interbedded with fine sand and shell fragments. The soil properties and CPT
results are presented in Figure 11 and Figure 12. The CPT was a downhole wireline
device with a 1.5-m stroke, and the small stroke unfortunately results in some spikes
and some discontinuities in the profile. Looking at both the CPT and soil classifica-
tion tests, there are essentially three substrata, all silts; an upper silt, a high plasticity
silt and a lower silt.
4.1 Index Tests

From seabed to a depth of about 64 m, low plasticity silt is the prevalent material.
The liquid limit is less than 38 with a plasticity index less than 9 and this material is
therefore classified as ML, according to the USCS. The natural moisture content is
typically from 35% to 40% and is in general close to or above the liquid limit as indi-
cated by the liquidity index on Figure 11. Sand layers are present as inferred primari-
ly from the CPT records, but each of these layers is relatively thin and interbedded
with the silt. The sand layers are more prevalent and seemingly denser between
depths of 28 m and 38 m below seabed. Clay layers were not observed in the samples.
The second stratum between 64 m and 80 m is mostly high plasticity silt (MH in
USCS) with a natural moisture content range from 37% to 61%, and the respective
liquid limits between 43 and 84, leading to liquidity indices from a low of -0.4 to a
high of 1.42. The upper part of this subunit is more plastic and is uniform on the CPT
trace, which also shows the likely presence of sandy or silty interbeds in the lower
part the stratum.
Water content (%) Liquid limit Plasticity index Liquidity index

0 20 40 60 80 0 20 40 60 80 100 0 10 20 30 40 -1 0 1 2 3
0
10
20
30
Upper ML Silt
40
50
60
70
MH Silt
80
90 Lower ML Silt
100
Figure 11. Soil index properties at marine silt site.
Below 80 m, the plasticity drops again to that of low plasticity (ML) silt. The li-
quidity index is very close to 1, with a liquid limit between 30 and 38, and natural
moisture content between 29% and 38%. The plasticity index is between 27 and 29.
The CPT records show the presence of abundant sand lenses.
4.2 CPT Data

In the CPT on Figure 12, a highly stratified deposit ranging between clays and sands
is apparent from the tip resistance, friction ratio, pore pressure parameter, and materi-
al behavior index. However, some caution is needed in the interpretation, as the mea-
surements in one layer may be heavily influenced by the layers above and below. The
same variability was not apparent in the index tests on the soil samples, which were
predominantly silts with very little sand or clay present in any of the samples. Visual
examination of the samples (good quality, thin walled push samples) also did not in-
dicate the stratification noted in the CPT. Plotting the CPT data on a soil behavior
chart shows a confusing picture of material types ranging from clays through silts to
clean sands, that looks like a shotgun blast on Figure 13. The pore pressure parameter
Bq (Figure 12) shows both drained penetration (Bq ~ 0) and undrained or partially
drained penetration, with Bq values as high as 0.4. In addition, the liquidity indices
are generally greater than one (Figure 11).
The initial interpretation of this rather confusing picture was that there might be
something fundamentally different about the silt. Diatomaceous materials are found
qt (MPa) F (%) Bq Ic (J&D)

0 10 20 30 40 50 0 1 2 3 4 5 6 7 -0.2 0.0 0.2 0.4 0.6 1 2 3 4
0
10
organic soil
20
clays
30
Upper ML Silt
40
50

60

70 MH Silt
80
Lower ML Silt
90
100
Figure 12. CPT profile at marine silt site.
1000
8
7
1. Sensitive fine grained
9
100
4. Silt mixtures – clayey silt to silty
6 clay
silt
5
7. Gravelly sand to sand
10 4 8. Very stiff sand to clayey sand
9. Very stiff fine grained
3
1
2
1
0.1 1 10
Figure 13. Plot showing where data for marine silt falls on the Robertson (1990) classification dia-
gram.
in deeper offshore waters, and may have been transported to these shallower depths
by deep seabed currents. However, microscopic examination of the silt grains indi-
cated nothing unusual; this was a textbook siliceous silty material. Assuming then
that both the CPT tests and the laboratory testing and soil descriptions were accurate,
we concluded that these offshore silts were indeed strongly bedded, but that the pri-
mary difference between each layer was its strength and stiffness, rather than its visu-
al appearance or soil classification. This variability in strength and stiffness was rea-
dily apparent in the CPT, although it may be muted by the influence of layering on
CPT parameters. The denser or stiff silt layers exhibit drained “sand-like” behavior,
while the less dense, less stiff silts exhibits “clay-like” behavior.

A first step in the interpretation was to separate sand-like behavior from clay-like be-
havior, and the obvious way to do this with the CPT (assuming it includes pore pres-
sure measurement) is based on the material behavior index. In this case, the Jefferies
and Davies (1993) index was used, and the data were separated at an Ic value of 2.4,
corresponding approximately to the boundary between zones 4 and 5 on the soil be-
havior type chart. On the sandy side of the divide (soil types 5, 6, and 7), the state pa-
rameter ψ was determined, while on the clayey side (soil types 3, and 4) undrained
shear strength was calculated, as shown on Figure 13.
The CPT tip resistance is highly variable. Undrained shear strength (cu) from tri-
axial compression test results also show a wide range from about 100 kPa to 900 kPa.
Both tip resistance and undrained shear strength seem high and inconsistent with a li-
quidity index greater than 1. The highs and lows in undrained strength may reflect
differences in the proportions of sand and clay within the silt, as high undrained
strength values would be typically associated with the more granular (sand-like) silts,
while low values would be typically associated with a more cohesive (clay-like) silt.
The variability in undrained strength reduces between depths of 40 m to 80 m, which
coincides with an increase in plasticity. Here triaxial compression undrained shear
strength values have a range from about 150 kPa to 350 kPa.
From the above description of soils at the site, it is reasonable to consider that the
low points in the CPT tip resistance are most representative of the undrained shear
strength of the silts. In many cases, “lows” in the tip resistance might not be repre-
sentative of the undrained shear strength in a thin layer and will tend to overpredict
the shear strength, so it is appropriate to consider something close to the lower bound
of the tip resistance profile. On Figure 14, the laboratory undrained shear strength
measurements and the CPT undrained shear strength for Ic values greater than 2.4 are
shown, using a value of Nkt = 14, on the same axis. There is a large variability in both
the CPT and the laboratory shear strengths, as expected, but considering the variabili-
ty, the laboratory data line up reasonably well with the low points of the CPT data.
This alignment confirms that the selected value of Nkt = 14 is applicable. There is al-
so an indication in the higher plasticity material between 60 m and 70 m that a
slightly lower Nkt can be used, since the CPT plots below the two laboratory strength
measurements in this depth interval. A rational design undrained strength profile in
the depth intervals that are predominantly clay-like is indicated as a dashed line on
Figure 14.
The state parameter values on Figure 14 for material with Ic values less than 2.4
have been calculated using the approach in Jefferies and Been (2006). The Ic value
was used to determine values for the constants m and k, and then the state parameter
was calculated directly from Q. Figure 14 shows that the silt has a state parameter
generally smaller than ψ = -0.1 in the upper 40 m of the profile, and at greater depths,
the state parameter lies between about 0 and -0.1. Under static loading, the silt is ex-
pected to be dilatant at these states, but under the high seismic loading possible at this
site, it is predicted that high cyclic pore pressures could be generated. A design pro-
file for state parameter in the depth intervals that are predominantly sand-like is
shown on Figure 14, noting that a less negative state parameter indicates looser ma-
terial and that the line is therefore at about a 10 percentile exceedance level.
cu (kPa) ψ
0 250 500 -0.3 -0.1 0.1
0
10
20
30
40
50
60
70
80
90
CPT
UU Triaxial
100
Figure 14. Plots of undrained shear strength and state parameter for marine silt (Nkt = 14 for undrained
shear strength when Ic(J&D) >2.4)
4.4 Summary
This case study presents an interesting quandary for pile design. It is possible either
to treat the whole soil profile as a nearly uniform silt (based on the soil properties)
and select design parameters accordingly, or to try to determine what percentage of
the profile is clay-like and what is sand-like, and then select appropriate sand or clay
parameters for each soil type as was done above. It may not be important which ap-
proach the designer chooses, but it is certainly important to use a combination of the
information from the examination of the samples, laboratory testing, and the CPT to
present a wider picture of the soil composition.
5 CONCLUSIONS
The experiences described in this paper confirm the much repeated advice that the
CPT cannot be used in isolation and requires an appropriate program of sampling and
laboratory or related testing to confirm the selected soil parameters.
The residual soil example yielded an Nkt = 16 for undrained shear strength, which
is well within the expected range for soils of the same plasticity. However, the mod-
ulus would not have been predicted well from the CPT and the example highlights
that the CPT should not in general be used to determine parameters through indirect
correlations. Application of a seismic CPT can be helpful in this context.
Classical soil behavior could be expected for the lightly overconsolidated estuarine
clay site, and thus OCR and undrained shear strength are related through the relation-
ships suggested by Ladd and Foote (1974). As a result, only certain combinations of
parameters Nkt and k are compatible and were used in the absence of extensive labora-
tory strength testing during the early stages of this project.
The marine silt case provides a particular challenge for interpretation, both be-
cause of the layering and because it is a silt (i.e. not a clearly identified sand or clay).
The material behavior type index does, however, provide a rational basis to separate
sand-like from clay-like behavior in the CPT profile, and undrained shear strength
and state parameter profiles were then developed corresponding to clay-like and
sand-like behaviors.
It is also apparent from these conclusions that there are many soils that do not easi-
ly fit the idealized “sand” or “clay” behavior, on which much of soil mechanics de-
pends. Therefore, the geotechnical engineer needs to have a flexible approach. In ad-
dition, without laboratory testing, visual examination of the samples and
consideration of the geological history of the soil, the CPT can easily mislead in
terms of soil type, strength and particularly modulus.
6 ACKNOWLEDGEMENTS
The authors wish to thank their clients for allowing us to use the data from their re-
spective projects for this paper, in particular for Mr. Jim Collins of Marathon Oil
Company.
7 REFERENCES
Jefferies, M.G. & Davies, M.P. (1993) Use of CPTu to estimate equivalent SPT N60, Geotechnical
Testing Journal, ASTM, 16, 4, 458-468.
Jefferies, M.G. & Been, K. (2006) Soil liquefaction: A critical state approach. Taylor and Francis, New
York.
Karlsrud, K., Lunne, T., Kort, D.A. & Strandvik, S. (2005) CPTU correlations for clays. Proceedings
of the 16th International Conference on Soil Mechanics and Geotechnical Engineering, Vol 2, 693-
702.
Ladd, C.C. & Foote, R. (1974) New design procedure for stability of soft clays. Journal of the Geo-
technical Engineering Division, Vol 100, No. GT7, pp. 763-786.
Ladd, C.C. & De Groot, D.J. (2003) Recommended practice for soft ground site characterization: Ar-
thur Casagrande Lecture. Proceedings of the 12th Pan-American Conference on Soil Mechanics
and Foundation Engineering, pp. 3-57.
Robertson, P.K. (1990) Soil classification by the cone penetration test. Canadian Geotechnical Jour-
nal, 27, 1, 151-158.
Robertson, P.K. (2009) Interpretation of cone penetration tests – a unified approach. Canadian Geo-
technical Journal, 46, 11, 1337-1355.
Robertson, P.K. & Wride, C.E. (1998) Evaluating cyclic liquefaction potential using the Cone Penetra-
tion Test. Canadian Geotechnical Journal, 35, 3, 442-459.
Schnaid, F. (2005) Geocharacterization and engineering properties of natural soils by in situ tests. Pro-
ceedings of the 16th International Conference on Soil Mechanics and Geotechnical Engineering,
Vol 1, 3-45.

Been 2010 CPT in Practice CPT10

Uploaded by

Copyright:

Available Formats

Been 2010 CPT in Practice CPT10

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Been 2010 CPT in Practice CPT10

Uploaded by

Copyright:

Available Formats

2nd International Symposium on Cone Penetration Testing, Huntington Beach, CA, USA, May 2010

Interpretation of the CPT in engineering practice

Soil behavior is in general complex. In addition to physical interactions at the particu-

2 RESIDUAL SOIL SITE

2.1 Index Tests

2.2 CPT Data

qt (MPa) F (%) Bq Q Ic (J&D) Ic (R&W)

silty sand to sandy silt

silty sand to sandy

clayey silt to silty clay

100 4. Silt mixtures – clayey silt to silty

2.3 Engineering Interpretation of the CPT

Undrained shear strength, cu (kPa) Shear modulus, G (MPa)

a) Undrained shear strength b) Shear modulus

Shear modulus, Gmax (MPa)

Gmax from CPT (equation 4)

range. We do not recommend estimating soil stiffness using empirical correlations

3.1 Index Tests

3.2 CPT Data

Moisture content (%) Unit weight (kN/m 3)

3.3 Engineering Interpretation of the CPT

qt (MPa) F (%) Bq Ic (J&D) Ic (R&W)

silty sand to sandy silt

sands: clean to silty

silty sand to sandy

Bottom of organic clay

8 1. Sensitive fine grained

100 4. Silt mixtures – clayey silt to silty

(qt-σv/σ'v)1-m and Nkt x S x km

Table 1. Combinations of compatible empirical parameters for CPT interpretation.

4.1 Index Tests

Water content (%) Liquid limit Plasticity index Liquidity index

Figure 11. Soil index properties at marine silt site.

4.2 CPT Data

qt (MPa) F (%) Bq Ic (J&D)

sands: clean to silty

clayey silt to silty clay

4.3 Engineering Interpretation of the CPT

You might also like