Accepted Manuscript

Cone Penetration Testing in Unsaturated Soils

Gerald A. Miller, Norman K. Tan, Rodney W. Collins, Kanthasamy K.

Muraleetharan

PII: S2214-3912(18)30139-9
DOI: https://doi.org/10.1016/j.trgeo.2018.09.008
Reference: TRGEO 193

To appear in: Transportation Geotechnics

Received Date: 23 June 2018

Revised Date: 25 September 2018
Accepted Date: 25 September 2018

Please cite this article as: G.A. Miller, N.K. Tan, R.W. Collins, K.K. Muraleetharan, Cone Penetration Testing in
Unsaturated Soils, Transportation Geotechnics (2018), doi: https://doi.org/10.1016/j.trgeo.2018.09.008

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Cone Penetration Testing in Unsaturated Soils
Gerald A. Millera, Norman K. Tanb, Rodney W. Collinsc, Kanthasamy K. Muraleetharand

a
(corresponding author) Professor, School of Civil Engineering and Environmental Science,
University of Oklahoma, 202 W. Boyd St., Rm. 334, Norman, OK, 73019, USA.
[email protected].
b
Manager Geotechnical Department, Terracon, 4701 North Stiles Ave., Oklahoma City, OK,
73105, USA. [email protected].
c
Geotechnical Engineer, Midwest Engineering and Testing Corporation, 2025 S. Nicklas Ave,
Oklahoma City, OK 73128, USA. [email protected].
d
Professor, School of Civil Engineering and Environmental Science, University of Oklahoma,
202 W. Boyd St., Rm. 334, Norman, OK, 73019, USA. [email protected].

Abstract: Results of Cone Penetration Tests (CPTs) on compacted clayey silt under different
moisture contents and matric suctions conducted in a calibration chamber are presented along
with results of CPTs conducted in unsaturated clayey soil at two test sites over a period of
approximately two years. CPT data obtained from published literature are presented for
comparison. The accumulated data demonstrate the dependence of penetration resistance on
moisture content and suction and reveal empirical relationships that can be exploited for
preliminary assessment of tip resistance under changing moisture conditions. A preliminary
treatment of tip resistance in unsaturated soil using bearing capacity theory is presented and
shows promise as a semi-empirical method for assessing changes in tip resistance as a function
of matric suction.

Key Words: Cone Penetration Testing; Unsaturated Soil; Calibration Chamber; Field Testing;
Matric Suction; Bearing Capacity Theory
 Gerald A. Miller, Ph.D., P.E.
Prof. Miller is a Presidential Professor and Associate Director of the
School of Civil Engineering and Environmental Science at the University
of Oklahoma. Over his 24-year academic career, his research has focused
primarily on laboratory investigations of unsaturated soil behavior, in situ
testing and foundation engineering in unsaturated soils, soil-structure
interaction and chemical improvement of soils. More recently, he has also
been focusing on desiccation cracking and slope stability in unsaturated soils. He is a licensed
professional engineer in Oklahoma and active in geotechnical engineering consulting.

Norman K. Tan, Ph.D., P.E.

Dr. Tan joined Terracon in 2005, where he is the geotechnical department
manager for Terracon’s Oklahoma City, Oklahoma, office. His duties
include project coordination, engineering analyses, proposal and report
preparation, and supervision of field exploration and laboratory testing.
Prior to joining Terracon, he received his M.S. and Ph.D. degrees at the
University of Oklahoma. While at the University of Oklahoma, his
research, funded by the National Science Foundation, was focused on pressuremeter and cone
penetrometer testing in unsaturated soil in a calibration chamber.

Rodney W. Collins, Ph.D., P.E.

Dr. Collins is a geotechnical engineer at Midwest Engineering and Testing
Corporation (METCO). He joined METCO after receiving his Ph.D. at the
University of Oklahoma in 2016. Prior to his Ph.D. studies, he received his
M.S. at the University of Alaska, Fairbanks. His Ph.D. research focused on
interpretation of cone penetration and pressuremeter tests in unsaturated
soils with application to shallow foundation analysis. The research
involved in situ testing at two instrumented test sites over a period of two years and model
foundation and pressuremeter testing in a laboratory soil chamber.

K.K. "Muralee" Muraleetharan, Ph.D., P.E., G.E., F. ASCE

Prof. Muraleetharan is the Kimmell-Bernard Chair in Engineering and a
David Ross Boyd and a Presidential Professor of Civil Engineering and
Environmental Science at the University of Oklahoma (OU). He is also an
Associate Director of OU's National Institute for Risk and Resilience. Prof.
Muraleetharan is interested in large-scale computer simulations of
infrastructure (bridges, roads, levees, etc.) subjected to extreme events
(earthquakes, hurricanes, etc.), validations of these simulations using small-scale (e.g. centrifuge
models) and full-scale testing, and resilience of infrastructure systems following extreme events.
He is also interested in computer simulations of fluid flows and solid deformations in multiphase
porous media.
1. Introduction

1.1 Overview

Cone penetration testing is an excellent soil profiling method for numerous geotechnical
applications in roadway construction and for other civil infrastructure. Compared to traditional
drilling and discrete sampling it provides significant advantages in that it is faster and results in a
near continuous profile of soil information. Primary disadvantages of standard Cone Penetration
Tests (CPTs) are the inability to obtain samples, and interpretation of soil types and properties
relies on empiricism. A comprehensive site investigation should therefore include both
traditional drilling and sampling and CPTs. Use of CPTs can provide much greater quantities of
soil data to characterize spatial variability while simultaneously reducing reliance on drilling and
sampling. However, some drilling and sampling remains crucial to provide site specific empirical
parameters and validation data for interpretation of CPT results.

There are many scenarios where CPTs can play an important role in transportation projects
including: subsurface exploration for new construction, rehabilitation projects, and forensic
investigations. In the latter two cases, CPTs are often conducted through existing embankments
to collect information about the compacted fill and underlying foundation soil. Generally,
embankment soils that are not submerged in water will be in an unsaturated state, as are the
upper portions of natural soil profiles above the water table. Thus, it is important to consider
current soil moisture conditions in interpreting CPT data, since variations in moisture contents
will affect matric suctions and corresponding stress states. This in turn will affect CPT tip
resistance and sleeve friction. Some significant research recent decades has improved our
understanding of the influence of variable saturation on CPT results; however, as of yet there are
only limited methods of interpretation of soil properties from the CPT that explicitly consider the
influence of matric suction.

There are many challenges to developing an interpretation framework for cone penetration in
unsaturated soils. Cone penetration is a complex boundary value problem, even for saturated
cohesionless sands and clays, but solutions to these problems have been developed with benefit
of simplifying assumptions regarding drainage conditions. For saturated sands, drained
conditions with zero excess pore water pressures are assumed to prevail, while for saturated clay,
constant volume shearing with excess pore water pressure development is assumed. The addition
of pore water pressure transducers in the case of the Piezocone Penetration Test (PCPT or
CPTU) has allowed for direct measurement of pore water pressures and greatly enhanced
interpretation of results. For unsaturated soils, these drainage assumptions are generally
inadequate and it is not yet possible to reliably measure pore air and pore water pressures during
penetration. Because soil volume changes can occur during penetration in unsaturated soil, it is
likely that matric suction will also change in response to changes in volumetric water content and
distortion of the air-water interface. However, in many cases it may be reasonable to assume that
penetration occurs under constant water content conditions, for soils not near saturation, and
some researchers have treated this as another drainage condition (e.g. Yang and Russell 2015a).
Nevertheless, interpretation of cone penetration in unsaturated soils is highly complicated due to
complex drainage conditions with respect to pore air and pore water, and volume changes that
cannot be addressed with simplifying assumptions. In addition, any reasonable interpretation
framework will require information about the initial saturation or suction state of the soil before
penetration, which further emphasizes the need for companion sampling activities at parallel
locations. In spite of these challenges, the CPT in unsaturated soil can provide considerable
insight into soil spatial variability and information about soil properties when conducted in
parallel with traditional drilling and sampling. However, the interpretation of data must consider
the potential for changing moisture conditions.

1.2 Objectives

The purpose of this paper is to present results of research elucidating the influence of soil suction
on results of CPTs. In doing so, it also provides a preliminary empirical basis for interpreting
results of CPTs in unsaturated soils. Specifically, work presented in this paper provides the
means for a preliminary assessment of the influence of matric suction on cone tip resistance and
to a lesser extent the friction ratio. Research conducted by the authors includes results from
calibration chamber testing using a low plasticity, compacted clayey silt and results of testing at
two instrumented test sites containing clayey soils of varying plasticity. This paper also draws on
published work of other researchers to provide a broader view of the importance of soil suction
on CPT results. Further, a simple theoretical treatment of cone penetration in unsaturated soil
using bearing capacity theory is presented as a preliminary means to assess the dependence of tip
resistance on matric suction.

1.3 Background

1.3.1 General

The cone penetration test of today is a successor of the first deep mechanical cone penetration
test developed at the Delft Laboratory of Soil Mechanics beginning in 1935, as described by
Massarsch (2014). Since that time the CPT has undergone electrification, standardization, and
modifications to enable a variety of measurements during penetration. Furthermore, extensive
research has been performed to enhance interpretation and application of CPT results.

A standard friction cone is a stainless steel probe, 35.7 mm in diameter and tipped with a conical
point with an apex angle of 60 degrees (ASTM D5778-12, Standard Test Method for Electronic
Friction Cone and Piezocone Penetration Testing of Soils). The probe is instrumented such that
during vertical penetration of a soil profile the forces generated on the tip and on a friction sleeve
immediately behind the tip can be recorded. Tip resistance, qc, and sleeve friction, fs, are
computed, respectively, by dividing the tip force by the cross-sectional area (10 cm2) and sleeve
force by the sleeve area (150 cm2). Due to rate dependent behavior of soils it is important to
maintain a standard rate of penetration of 2 cm/s. Friction Ratio (FR) is defined as the sleeve
friction divided by the tip resistance, expressed in percentage, and is very useful in estimating
soil type.

A piezocone has an additional sensor to detect pore water pressures during penetration and is
typically located on the face of the conical point or just behind the tip, below the friction sleeve.
Measurement of pore water pressures is extremely beneficial in saturated soils for interpreting
soil types and properties. However, in unsaturated soils the pore water pressure sensor will not
function properly because of the need to maintain a completely saturated measurement system to
ensure rapid sensor response. Penetration through an unsaturated soil profile can rapidly
desaturate a pore water pressure measuring system, leading to significant hydrodynamic time lag
due to compressibility of air. When this occurs pore pressure measurements cannot be interpreted
in a meaningful way since the response of the sensor may be slower than the penetration rate and
air within the sensor influences the response in a random manner.

Data obtained from CPT and PCPT tests can be interpreted to provide information about the soil
type, relative density and friction angle of sands, and undrained shear strength and stress history
of clays (e.g. Robertson and Campanella 1983a and 1983b, Mayne and Kemper 1988, Robertson
2009, Saye et al. 2013, Robertson 2016). Comprehensive guides for application of cone data in
geotechnical engineering are also available (e.g. Schmertmann 1978, Lunne et al. 1997,
Robertson and Cabal 2015). In spite of the vast number of publications on the topic, some of
which are cited here, there are relatively few that address interpretation of CPT results in
unsaturated soil profiles.

1.3.2 CPT in Unsaturated Soils

A summary of articles involving CPT and unsaturated soils is presented in Table 1. One of the
earliest studies to systematically investigate the influence of moisture condition on penetration
resistance was conducted at the Army Engineer Waterways Experiment Station (now Engineer
Research and Development Center) in relation to performance evaluation of lunar vehicle wheels
(Freitag et al. 1970). It was found that in addition to being strongly dependent on relative
density, the penetration resistance of sand changed significantly for gravimetric moisture
contents between 0% and 2%. Durgunoglu and Mitchell (1973) developed a bearing capacity
theory for interpreting penetration resistance in soils accounting for both friction and cohesion
(“cohesion-friction soil”). Validation was accomplished by conducting penetration tests in the
laboratory on a silty sand (crushed basalt lunar simulant) where cohesion was achieved by
adding a small amount of water. Hryciw and Dowding (1987) studied resistance of a 1.27-cm
diameter penetrometer in sand at varying degrees of saturation in small laboratory samples to
provide insight into behavior of blast densified sands containing gas bubbles. At low degrees of
saturation, tip resistance was higher than the dry or saturated condition but at higher degrees of
saturation between 70 to 90%, the tip resistance was slightly lower than the saturated condition.
Others have since reported results of CPTs in the field under varying moisture conditions, some
with companion determinations of soil suction at the time of testing (Houston et al. 1995, Lehane
et al. 2004, Nevels 2006, Collins and Miller 2014, Rocha et al. 2016, Tang et al. 2017, Lo Presti
et al. 2018). In all of these cases, increases in effective stress associated with increases in suction
generally resulted in increasing tip resistance. Some of these results obtained from other studies
are presented in this paper along with the authors’ data.

Some studies in calibration chambers have been conducted to examine the influence of variable
saturation on tip resistance under controlled stress conditions in sands (Pournaghiazar et al. 2013,
Jarast and Ghayoomi 2016), silty sand (Yang and Russell 2013 and 2015b) and clayey silt (Tan
et al. 2003). Results of these studies were used in the development and validation of analytical
and numerical methods of analyzing CPT results in unsaturated soils. Some authors have
presented methods for estimating soil properties from cone penetration in unsaturated soils
(Yang and Russel 2015b, Tang et al. 2017) while some have presented empirical methods for
estimating bearing capacity from cone penetration tests in unsaturated sand (Mohamed and
Vanapalli 2015). While these methods are a significant step forward in development of a
framework for interpreting CPT results in unsaturated soil, they are largely empirical and
validated for a limited number of soils.

This paper presents results of two studies conducted at the University of Oklahoma involving
CPTs in a calibration chamber containing compacted clayey silt and at two test sites underlain by
clayey soils. These results are used to demonstrate relationships between cone penetration
parameters, water content and matric suction, and are compared to results of other studies where
CPT and suction data were obtained. Compiled results provide insight into the importance of
considering suction in the interpretation of CPT results in unsaturated soils. Further, the trends
observed can be used to provide a preliminary means of estimating changes in CPT penetration
resistance due to changes in moisture content and matric suction.

2. CPTs in a Calibration Chamber Containing Unsaturated Compacted Clayey Silt

2.1 Soil Bed Preparation and Testing

Cone penetration tests were conducted in a flexible wall calibration chamber filled with
compacted clayey silt. The test soil, known as Minco Silt, classifies as CL-ML soil according to
the Unified Soil Classification System (USCS). A summary of soil properties for Minco Silt is
provided in Table 2. Soil beds were created using static compaction on each of nine lifts to
produce a soil bed of 1.4 m height and 0.61 m diameter. Six different soil beds were created at
different moisture contents to produce different matric suction conditions prior to penetration. In
each soil bed, four CPTs were conducted: one in the center and three located half way between
the center and edge of the soil bed and spaced 120 degrees apart. The penetrometer had a
diameter of 1.78 cm and was advanced at a rate of about 0.5 cm/sec. This diameter is smaller
than the standard diameter of 3.57 cm, and a lower rate of penetration was used to provide better
control for the short push duration in the chamber. Because penetration probably occurred under
constant water content conditions, it is possible that both the lower penetration rate and standard
penetration rate would have produced similar results. Skin friction was not measured.

A BC1 boundary condition (e.g. Ghionna and Jamiolkowski 1991) was utilized for all calibration
chamber tests, whereby average vertical and horizontal stresses applied to external boundaries
were maintained equal and constant. Horizontal stresses were applied to vertical boundaries via
pressurized air against a rubber membrane enclosing the sample. Vertical stresses were applied
via a hydraulic jack acting against the rigid bottom platen. The top platen was fixed against the
chamber cover. For all chamber tests, average external stress of 100 kPa was applied to
horizontal and vertical soil boundaries and time was permitted to achieve stress equilibrium prior
to CPTs.

Soil samples were obtained after the CPTs to determine moisture content and unit weight of the
soil. Matric suction was estimated using moisture content measurements and a Soil Water
Characteristic Curve (SWCC) developed for the test soil. A wetting path SWCC for Minco Silt
used in this study is shown in Figure 1. The wetting path SWCC seemed appropriate since
compression of the soil beds was expected to reduce the void ratio and increase degrees of
saturation during the consolidation stage of soil bed preparation. Experimental data in Figure 1
were obtained using a custom made cell whereby water was precisely injected in carefully
measured increments into a compacted sample and suction was measured with an embedded
tensiometer following equilibrium. The device is similar to that used by Hamilton et al. (1981)
for doing instantaneous profile measurements except the sample length is much reduced to
achieve equilibrium in a reasonable amount of time. Tests were conducted on samples having
four different dry densities encompassing those of soil beds used for calibration chamber testing.
As shown in Figure 1, the gravimetric water content-matric suction SWCC for Minco Silt is
relatively insensitive to changes in dry unit weight. To corroborate suctions estimated using the
SWCC, for some of the calibration chamber tests, tensiometers were installed in the soil beds
prior to CPT testing or tensiometers were used to measure suction in samples collected after
CPTs. Generally, tenisometer measurements were in agreement with those estimated using the
SWCC. Additional details of the CPT testing can be found in Tan et al. (2003) and Tan (2005).

2.2 Results of CPTs in Calibration Chamber

Results of CPTs in two of the test beds are shown in Figure 2. CPT-1 and CPT-4 represent the
driest condition and wettest conditions with average gravimetric moisture contents of 5.8 and
11%, respectively. Plots of tip resistance (qc) with depth show that the four soundings in each
soil bed are similar and exhibit a repeatable pattern with depth. This indicates good repeatability
and uniform soil bed conditions were achieved. Determinations of water content (w) and dry unit
weight (d) corroborate this observation. Tip resistance was sensitive to the slight variations in
unit weight within each compacted layer as indicated by the sinusoidal nature of the soundings.
Unit weight was greater at the top of compacted layers compared to the center, as was tip
resistance. Matric suctions predicted based on measured gravimetric water contents and SWCCs
are shown to be relatively uniform with depth. As shown in Figure 2, tensiometer measurements
of suction near the top of the soil bed for CPT-4 were reasonably close to predicted matric
suction, with the exception of one measurement. Average matric suction (ua-uw) for CPT-1 and
CPT-4 was 0.059 and 0.018 MPa, corresponding to average tip resistance of 4.1 and 2.7 MPa,
respectively. Average tip resistance was computed using values between depths of 0.15 and 1.1
m, to exclude results that were possibly influenced by the top and bottom boundaries of the
chamber.

A summary of all 24 CPT soundings in each of six soil beds in the calibration chamber is
presented in Figure 3. This figure depicts average tip resistance plotted against average
gravimetric water content, matric suction and dry unit weight for each sounding in each of the
six soil beds. Results indicate a strong dependence of tip resistance on variations in moisture
content and suction. Interestingly, while average dry unit weight decreases, tip resistance
increases with increasing suction in Figure 3. This is because suction was higher for lower
density soil beds, and suction variations had more influence on strength than observed density
variations. While efforts were made to control the dry unit weight during compaction, soil beds
at higher moisture contents experienced greater volumetric compression following application of
external stress. It is expected that if dry unit weight was similar in all soil beds, the influence of
suction on tip resistance would be more pronounced. The observed increase in tip resistance with
increasing suction is consistent with observations of other researchers (e.g. Yang and Russell
2015b, Pournaghiazar et al. 2013) using calibration chambers where relative void ratios and
confining stresses were similar.

3. CPTs at Two Instrumented Field Sites

Over a period of approximately two years, CPT soundings were conducted at two test sites
known as North Base and Goldsby, near Norman, Oklahoma. Standard friction cone penetration
tests were conducted using a cone having a diameter of 3.57 cm and penetration rate of 2 cm/sec.
Each time CPTs soundings were conducted, companion test borings were performed to obtain
samples for moisture content and suction determinations. Soil profiles for North Base and
Goldsby are shown in Figures 4 and 5, respectively. As shown in Figure 4, the North Base
profile consists of three clayey soil layers of varying plasticity. The Goldsby profile shown in
Figure 5, is composed of two soil layers that have different sand content and slightly different
plasticity. Also shown in Figures 4 and 5 are the minimum and maximum natural water contents
at each test depth determined during the monitoring period. Dry unit weight and degree of
saturation data are also presented in Figures 4 and 5. At both sites, water content variations were
greatest near the ground surface, and relatively small toward the bottom of the soil profile, which
is approaching the bottom of the active zone. Additional details about this study can be found in
Collins and Miller (2014) and Collins (2016).
Results of CPTs on different dates are shown in Figures 6 and 7 for North Base and Goldsby
Test Sites, respectively. On each date, three CPT soundings were conducted and values of tip
resistance and friction ratio were averaged at each depth to produce the plots of qc and FR
against depth. While results of CPT soundings are quite variable at each site, results show that
the higher tip resistances correspond to drier soil conditions. At both sites, tip resistance
variations between testing dates were much lower at depths below 2 m, which is consistent with
the lower variation in moisture contents at greater depths. Variations in friction ratio are erratic
with respect to variations in moisture contents at the North Base Site. However, at the Goldsby
Site, the higher friction ratios seem to correspond to wetter conditions, and the variation in
friction ratio is less at depths below 2 m where moisture variations are lower.

Matric suction profiles shown in Figures 6 and 7 were determined using gravimetric moisture
content measurements and SWCCs determined for corresponding soil layers shown in Figure 1.
Similar to moisture content variations, suction variations are considerable near the ground
surface, with maximum differences of nearly 3 MPa between driest and wettest conditions. The
SWCCs for Soils 2 thru 5, shown in Figure 1, were estimated using the empirical method of
Zapata et al. (2000). This method incorporates the PI and percent passing the #200 sieve to
estimate fitting parameters for the Fredlund and Xing (1994) SWCC equation. Shown for
comparison are total suction and water content measurements obtained for field samples
collected on various testing dates and some results of pressure plate tests. Total suction
measurements were obtained with a chilled mirror device, the WP4 Potentiameter from Decagon
Devices, Inc. It was noticed through extensive testing of saturated specimens with this device
that the soils at the two test sites have a significant osmotic suction component, which helps to
explain the difference between the predicted SWCC and the total suction measurements at high
water contents. Pressure plate results compare favorably to the predicted SWCCs; however, they
had higher initial water contents due to unrestrained swelling that occurs from lack of
confinement in pressure plate testing during saturation of field samples. As shown in Figure 1,
for the North Base Site, a separate SWCC was estimated for Soils 2 and 3 because the
corresponding soil layers were considerably different in terms of the PI and %passing the #200
sieve. Also, on average the WP4 total suction measurements for Soil 3 were lower than for Soil
2, which is consistent with the predicted SWCCs shown in Figure 1. For the Goldsby Site, the
difference between Soils 4 and 5 were not large and so a single SWCC was estimated using the
average soil properties for both layers, as shown in Figure 1.

In Figure 8, CPT results are summarized for the North Base and Goldsby Sites along with the
results from the calibration chamber study. For the North Base Site, Layer 1 corresponds to the
layer with the highest average PI in Figure 4; while Layer 2 includes data from top and bottom
layers shown in Figure 4. Note, most of the Layer 2 data came from the bottom layer at North
Base. Data for Goldsby Layers 1 and 2 correspond to the top and bottom layers shown in Figure
5, respectively. A summary of average layer properties is provided in Table 2. In Figure 8,
average tip resistance and friction ratio are plotted against moisture content and matric suction
(log scale). To calculate the average tip resistance and friction ratio corresponding to a given
moisture content and suction, qc or FR was averaged over tributary depths corresponding to the
depth at which the moisture content or suction was determined. Since moisture content
determinations were made on 0.15 or 0.3 m intervals, qc and FR values were averaged over depth
intervals of 0.15 or 0.3 m, respectively. While scatter is significant, there are some obvious
trends in the plots of tip resistance against water content and suction. In Figure 9, the tip
resistance normalized by total vertical overburden stress (qc/vo) is plotted against water content
and suction. Normalizing by overburden pressure to some degree accounts for dependency of tip
resistance on total stress state. While a normalization scheme using effective stress may be more
appropriate based on theoretical considerations, normalization by total vertical stress is more
practical. This is because to estimate effective stress, using a single stress state variable
approach, requires an estimate of a parameter related to the volume fraction of water present.
Thus, to simplify this analysis for empirical purposes, in this paper tip resistance was normalized
by total vertical stress. While scatter is significant, general trends in the plots of Figure 9 can be
useful for assessing potential variations in tip resistance as a function of water content or matric
suction while accounting for overburden pressure. Results of linear regression corresponding to
trend lines in Figures 8 and 9 are presented in Table 3. These relationships can be useful for
assessing expected changes in tip resistance for a given change in moisture content or suction for
similar soils. However, the relationships are strictly empirical and in theory may not be linear
with regard to the x and y variables in Table 3. Nevertheless, since data such as this are very
limited, these empirical relationships can serve as a preliminary means of assessing potential
variations in tip resistance due to changes in water content or suction for similar soils.

4. Data Obtained from Published Literature

Data from four additional sources were analyzed alongside the authors’ data. The nature of the
soil involved and the sources of data are presented in Table 2. Additional soils are numbered 6
through 11. Soils 6, 7, 10 and 11 involved CPTs conducted in the field with companion estimates
of suction. Soils 8 and 9 were investigated with a small CPT in a calibration chamber, and
represent the same sand prepared to different void ratios. For Soils 8, 9, 10, and 11, SWCC data
can be found in the source papers. For Soils 6 and 7, no SWCC information was reported in the
source papers.

Results of all 11 soils represented in Table 2 are summarized in Figure 10, where tip resistance
and tip resistance normalized by total vertical stress are plotted against matric suction. Graphs on
the left side of Figure 10 represent lower PI soils while graphs on the right represent higher PI
soils. Trends in the data depicted in the plots of log(qc) against log(ua-uw) exhibit positive slopes
in most cases, except for Soil 10 where the slope of the trend line is negative. When qc is
normalized by vo and plotted against ua-uw in a log-log space, slopes of trend lines appear to be
nearly horizontal in the case of lower PI or sandy soils, or positive for higher PI soils. Flatter
trend lines for sandy soils suggests that matric suction is not as critical relative to overburden
pressure, which is consistent with lower values of suction in sand and its lower impact on shear
strength relative to clayey soils. In Figure 11 the slopes of trend lines from the graphs of suction
versus tip resistance and normalized tip resistance (Figure 10) are plotted against the average PI
of the soil. In the upper graph of Figure 11, the slopes from log(qc) versus log(ua-uw) trends are
shown plotted against PI, and no obvious trend is exhibited. In the lower graph of Figure 11, the
slopes of the log(qc/vo) versus log(ua-uw) plotted against PI are shown; with the exception of one
outlier (Soil 11), there is a trend of increasing slope with increasing PI. These observations
emphasize the increasing importance of changes in matric suction to changes in shear strength,
and hence normalized tip resistance, as soils become more plastic.

5. Discussion of CPT Data and Trends

Results of tip resistance and companion determinations of suction are presented from calibration
chamber tests at two different laboratories and field tests obtained by four different research
groups at five different field sites. The data were subdivided into 11 categories according to
variations in soil type, sand content, plasticity index and void ratio for one of the sands. While
there is much scatter in graphical relationships between tip resistance, normalized tip resistance
and matric suction, there are nevertheless consistent trends in the relationships that, with few
exceptions, generally agree with expectations based on unsaturated soil mechanics theory. It is
not surprising that there is significant scatter in the data presented considering the many factors
that influence tip resistance and suction that were not accounted for in analyzing the data. Some
of these are discussed as follows.

1) Suction and tip resistance data obtained from the literature for the field sites were not
available in tabular form and so the present authors extracted tip resistance and corresponding
suction values from graphs of these values plotted against depth. Thus, some degree of
interpretation and averaging of profile data was necessary in some cases. In the case of the data
for Soil No. 7 (Perth Sand), suction was estimated using an equation presented by Lehane et al.
(2004) relating very small strain shear stiffness (Gvho) to mean effective stress computed as the
sum of matric suction and total stress. Suction at various depths corresponding to tip resistance
measurements was thereby estimated using the profiles of Gvho. Thus, some errors relative to
“true” in situ suction values and tip resistance are quite possible due to assumptions involved and
the present authors’ interpretation of the data.

2) Only two parameters, vertical total stress and matric suction, were considered in relationship
to tip resistance. Tip resistance in soils depends on many interrelated factors other than matric
suction and vertical total stress, including: void ratio (or dry unit weight), stress history,
horizontal stresses, macro- and micro-fabric, mineralogy, and soil chemistry. Under field
conditions, variations in these parameters can effect penetration resistance and can contribute to
the scatter observed since they were not explicitly considered in the relationships presented in
Figures 8, 9 and 10. The exceptions were the data obtained for Lyell Silty Sand that were
separated into two categories based on void ratio ranges as represented by Soils Nos. 8 and 9. It
is clear from Figure 10, as expected, that the void ratio has a strong influence on the resulting tip
resistance. As noted previously, the influence of dry unit weight, and hence void ratio, is also
seen in Figure 2, particularly for CPT-4, where subtle variations in dry unit weight with depth
correspond to the sinusoidal variations in tip resistance. While limited, these observations
suggest consideration of void ratio in developing CPT methods of interpretation for unsaturated
soils is important.

3) Different methods were used in each study to determine matric suction corresponding to
CPTs. For Soil 1, Tan (2005) used a combination of techniques including determination of
matric suction using moisture contents and a SWCC obtained by a direct measurement
technique, tensiometer measurements in the soil test bed, and tensiometer measurements on
samples obtained after testing. For Soils 2, 3, 4 & 5, Collins (2016) used field moisture content
measurements and SWCCs to predict soil matric suction. SWCCs were estimated using an
empirical method supplemented with pressure plate measurements and total suction
measurements on field samples. Nevels (2006) for Soil 6, also used field moisture contents and a
SWCC determined using the filter paper method to establish matric suction profiles. As
mentioned, for Soil 7, a method based on small strain shear stiffness was used to determine
matric suction. For Soils 8 and 9, Yang and Russell (2015b) either controlled matric suction
directly in their calibration chamber via high air entry porous discs and axis translation or
measured it directly with tensiometers. Tang et al. (2017), for Soils 10 and 11, measured suction
in companion Shelby Tube samples using tensiometers. All of these methods are subject to
varying degrees of accuracy and so experimental error is expected to contribute to the scatter
observed in relationships between tip resistance and matric suction. Furthermore, suction
determinations for field studies do not correspond to exact points where tip resistances are being
measured and so some error may be introduced by spatial variations in field suction between
sampling and CPT sounding locations. In addition, field moisture content measurements, when
used with a single SWCC, do not account for drying and wetting patterns at field sites that may
introduce hysteretic behavior in the SWCC and associated mechanical soil behavior.

4) The analysis of CPT results in this paper does not account for possible differences in CPT
testing procedures used in each study such as differences in cone diameter, in the case of
calibration chamber tests, or variations in penetration rate.

In spite of all of the potential factors that may have influenced the tip resistance and matric
suction data presented in this paper, trends in Figures 8, 9 and 10 emphasize the importance of
matric suction on tip resistance, particularly for soils with high plasticity. Furthermore, the
scatter in the data suggests more research is needed that involves more controlled CPT testing
and detailed characterization of soil parameters, in addition to moisture content and matric
suction, to better understand the importance of other parameters on penetration resistance.

6. Prediction of Tip Resistance as a Function of Matric Suction Using Bearing Capacity

Theory

6.1 Bearing Capacity Theory Applied to Cone Penetration in Unsaturated Soil

As discussed in Yu and Mitchell (1998), different theoretical approaches can be used to model
penetration resistance in soil. Among the simplest are those that use bearing capacity theory.
However, simplicity comes at a cost in that bearing capacity represents an ultimate state and the
dependency of bearing capacity on soil compressibility is largely ignored in most methods. Some
methods of calculating bearing capacity factors that depend on soil compressibility have evolved
from cavity expansion theory (e.g., Vesic 1972). However, for the present analysis such
formulations were avoided since accounting for compressibility requires determining soil
parameters governing compressibility, which for unsaturated soils is quite challenging.
Nevertheless, using bearing capacity theory is a logical place to start with the analysis of
penetration resistance in unsaturated soil considering the added complexity introduced by partial
saturation. To demonstrate the potential of bearing capacity theory for the analysis of tip
resistance in unsaturated soil, the method of Durgunoglu and Mitchell (1973) was employed.
This method incorporates a Mohr-Coulomb failure law and applies to soils that have both stress
independent (cohesion, c) and stress dependent (friction angle, ) strength parameters. The
authors adopted this method due to its simplicity and demonstrated viability as a means of
modeling penetration resistance in -c soils (Durgunoglu and Mitchell 1973). The method was
extended to unsaturated CPT problems by incorporating an unsaturated strength model proposed
by Vanapalli et al. (1996).

According to the method of Durgunoglu and Mitchell (1973), tip resistance (qc) is calculated
using bearing capacity Equation 1.

qc = cNcc + sBNqq (1)

where: c = cohesion,
s = unit weight of soil,
B = width of the cone base,
Nc, Nq = bearing capacity factors, and
c, q = shape factors.

To account for the influence of matric suction, the shear strength model (Equation 2) of
Vanapalli et al. (1996) was incorporated.
 = c’ + (n-ua)(tan’) + (ua-uw)(tan’)(-r)/(s-r) (2)

where: c’ = effective stress cohesion,

n-ua = net normal stress on failure plane,
n = total normal stress on failure plane,
ua = pore air pressure,
’ = effective stress friction angle,
ua-uw = matric suction,
uw = pore water pressure,
 = volumetric water content,
r = residual volumetric water content, and
s = saturated volumetric water content.

In using Equation 2 it is typically assumed that effective stress-strength parameters c’ and ’ are
constant and equal to the values for saturated soils. However, the component of strength
associated with matric suction varies as a function of matric suction and the volume fraction of
water relative to the saturated and residual conditions. An expression for the cohesion,
incorporating the stress independent strength component (c’), and the component attributed to
suction in Equation 2, is given by Equation 3.

c = c’ + (ua-uw)(tan’)(-r)/(s-r) (3)

The elegance in this formulation comes from the fact that the entire strength envelope can be
estimated using a SWCC and saturated effective stress-strength parameters c’ and ’. Thus, to
capture the influence of matric suction on predicted tip resistance for unsaturated soil, Equation 3
is substituted into Equation 1, which gives Equation 4.

qc = [c’ + (ua-uw)(tan’)(-r)/(s-r)]Ncc + sBNqq (4)

By having a SWCC, c’ and ’, for a soil of interest, the tip resistance, qc, over the entire range of
suction represented in the SWCC can be estimated.

Bearing capacity factors depend on the soil friction angle (), relative roughness of the cone
given by the ratio of the cone-soil interface friction angle to the soil friction angle ( ), relative
depth of penetration, and apex angle of the cone tip (2). The equations for bearing capacity
factors Nc and Nq, as given by Durgunoglu and Mitchell (1973) are shown as Equations 5 and 6,
respectively. Alternatively, the design charts provided by Durgunoglu and Mitchell (1973) can
be used.
 (5)

(6)

where: Nc = bearing capacity factor for the cohesion term,

Nq = bearing capacity factor for friction-surcharge,
 = soil friction angle,
 = the topmost angle of the plane shear zone,
 = 90o - 
 = ½ the cone apex angle,
o = 180o – ( + ) + 
 = cone base to soil friction angle,
K = lateral earth pressure coefficient,
m = relative depth (D/B), D = cone tip depth, B = cone diameter,

The angles , , o and , and parameters m’ and I relate to geometry of the assumed rupture
surface around the cone tip, as shown in Durgunoglu and Mitchell (1973). The angle  depends
on the soil friction angle (), relative depth of penetration (m), roughness ratio (/), and cone
apex angle (2). For large relative depths the angle  will be equal to the soil friction angle .
For most situations, for a standard cone ( = 30o) assuming a perfectly smooth cone surface ( /
= 0), relative depths of penetration greater than ten (D/B > 10) can be considered large relative to
assuming  = , for  values up to about 45o. For a standard cone with a diameter of 0.0357 m, a
D/B of ten would correspond to a depth of penetration of 0.357 m. Angle  depends on the soil 
and  /; for a perfectly rough condition ( / = 1)  is equal to zero and for perfectly smooth
condition ( / = 0) it varies nearly linearly from 50o to 70o for  between 10o and 50o.

Shape factorsc and q depend on the penetrometer aspect ratio and soil friction angle, and q
also depends on the relative depth of penetration. Durgunoglu and Mitchell (1973) used the
semi-empirical shape factors suggested Hansen (1961), as these were found to provide the best
match to their experimentally determined values. These values were adopted here; for deep
penetration, c and q are the same and calculated using Equation 7.
c, q = 1.0 + (0.2 + tan6B/L) (7)

where: B/L was taken as 1.

6.2 Examination of Bearing Capacity Theory Relative to CPT Results in Unsaturated Soil

Bearing capacity theory, as represented by Equation 4, was investigated as a way to model CPT
tip resistance in unsaturated soil. This was accomplished by examining how theoretically
generated curves of qc versus suction compared to experimental values of qc and corresponding
suction. To accomplish this, estimated SWCCs presented in Figure 1 were used to generate
predicted qc curves, which required an estimate of c’ and ’ for each soil. For this exercise, c’
and ’ were estimated by varying their values within Equation 4 to achieve a reasonable match
between the predicted qc and measured qc. Then the values of c’ and ’ used to achieve a
reasonable match were compared to available, but limited, laboratory determined values of c’
and ’ in Table 5.

Shown in Figure 12 are predicted values of qc generated using Equation 4 with SWCCs in Figure
1 and model soil properties shown in Table 5. The SWCCs are included in the lower portion of
Figure 12 for comparison. For determining bearing capacity factors, it was assumed that D/B =
10 and  / =0, which implies deep penetration with a perfectly smooth penetrometer.

Predicted trends in qc compare favorably to trends in the measured qc values in Figure 12.
However, values of c’ and ’ used in the model in some cases are significantly different than
laboratory determined values shown in Table 5. For Soil 1 (Minco Silt), the best fit was achieved
with a ’ value of 46o and c’ of zero, as compared to laboratory values of 30o and 0.012 MPa,
respectively. Laboratory values were obtained through a series of suction-controlled triaxial
tests. Interestingly, borehole shear tests on Minco Silt in a laboratory test chamber produced a
range of friction angles from 33 to 46 degrees and cohesion intercepts near zero, for a matric
suction range of zero to 0.056 MPa, respectively (Miller et al. 1998). For Soil 2, ’ of 13o and c’
equal to 0.04 MPa was used to achieve a reasonable match between predicted and measured qc.
Soil 2 is overconsolidated residual clay that has the highest PI of the test soils and on average is
borderline classified as fat clay. Although strength tests data are not available for this layer, it
would be expected that this soil would have the lowest friction angle; however, 13o, seems too
low for an overconsolidated soil with average PI of 29 (e.g. Sorensen and Okkels 2013). For Soil
3, results from the three triaxial tests on back-pressure saturated specimens shown in Table 5
represent three different test depths and vary considerably with respect to c’ and ’. However, c’
and ’ used in the model predictions compare reasonably well to the laboratory strength
parameters having the lowest friction angle. For Soil 4 and 5, the best predictions were achieved
with model c’ and ’ values about 1.5 times as large as the laboratory determined values from
triaxial tests.
There are numerous factors that may contribute to the lack of agreement between the c’ and ’
parameters in the “best fit” model and values obtained through laboratory testing. These factors
include possible experimental errors in testing, variability in and quality of laboratory tests
samples, assumptions implicit in the bearing capacity theory and assumptions made in applying
the theory. As discussed by Yu and Mitchell (1993) use of bearing analysis does not consider the
dependence of tip resistance on soil stiffness and compressibility, or the influence of the
penetration process on the stresses, which generally includes an increase in horizontal stresses
around the cone. In determining the bearing capacity factors, it was assumed that the cone was
perfectly smooth ( / = 0), but higher  / values may be more realistic. For example, in the
case of Soil 1, a  / of 0.5 reduced the best fit ’ value from 46o to 42o and for Soils 4 and 5 the
reduction was from 30o to 25o and 35o to 30o, respectively. Another factor is the assumption of a
lateral stress ratio (K) equal to 1-sin’, which generally applies to normally consolidated soils.
Further, implicit in the use of the strength model for cohesion (Eq. 3) is the assumption that the
suction is constant during penetration, which may not be reasonable for many soils.

In spite of all the factors that can affect the predicted qc values, the bearing capacity model
provided predicted curves of qc versus ua-uw that match observed trends in the field CPT results.
Thus, if an SWCC is available for a given soil in which CPTs are conducted, the bearing
capacity model could be used to generate a family of curves, corresponding to a range of strength
parameters (estimated or measured) and assumed conditions, for comparison to field results. By
superimposing the field results on the family curves one can get an idea of how the tip resistance
may change as a function of changing field moisture conditions. This is demonstrated in Figure
13, where for Soil 3, qc predictions are presented corresponding to all of the laboratory ’ and c’
pairs in Table 5, along with the best fit curve. It is important to note that unlike typical projects
in practice, field CPT measurements for North Base and Goldsby were obtained on several
different dates over a period of two years. For practical applications, CPT measurements from
only a single visit may be available and trends in the actual CPT data (qc versus ua-uw) may be
difficult to see depending on how much the suction varies within a given soil layer. It is possible
that a single CPT sounding will provide data showing variations in suction with tip resistance
since moisture profiles are usually not uniform near the ground surface, particularly during
drying periods. This can be observed in the profiles corresponding to individual dates in Figure 6
and 7. In situations where CPT data are limited with respect to qc versus ua-uw variations, there
will be more reliance on and less validation of predicted qc versus ua-uw curves. This is
demonstrated in Figure 14 for Soil 3, where only CPT field data from two dates, representing one
of the wettest and the driest profiles, are presented along with the qc predictions based on actual
laboratory estimates of c’ and ’. While the trends in field data are less apparent for a given date,
the predicted trend lines still provide useful insight regarding potential changes in qc due to
changes in suction. Thus, if CPT soundings are conducted during a dry period, which is the worst
case scenario, some assessment of how they may change during a wet period can be made.
7. Summary and Conclusions

Results of CPTs in unsaturated soil obtained in the laboratory and field are summarized in this
paper. CPT data were obtained by the authors during a laboratory study using a calibration
chamber filled with compacted low plastic clayey silt and from field testing at two sites
underlain by clayey soils of various plasticity. Additional data obtained by other researchers was
extracted from the literature for comparison. Analysis of the data demonstrates the dependency
of tip resistance on moisture content and matric suction. Some empirical expressions are
presented to permit a preliminary assessment of expected changes in tip resistance with moisture
content. Some conclusions follow:

1) Tip resistance and normalized tip resistance generally increases with increasing matric suction
and decreasing moisture content.

2) The dependency of tip resistance on suction is greater for soils with significant plasticity
compared to non-plastic soils. This is expected given that the range of matric suction is much
greater, and hence its impact is much greater on the shear strength of clayey soils.

3) Data from calibration chamber testing revealed that void ratio (or dry unit weight) has a
significant effect on tip resistance. For the silty sand tested (Soils 8 and 9), tip resistance was on
average much higher for the lower void ratios at all values of suction tested. For the clayey silt,
while tip resistance was sensitive to variations in dry unit weight, the matric suction had a more
dominant effect such that average tip resistance in the soil bed with the highest dry unit weight
and lowest matric suction was less than that for the soil bed with the lowest dry unit weight and
highest matric suction.

4) Bearing capacity theory, incorporating matric suction, holds promise as a semi-empirical

method for predicting changes in tip resistance due to changes in water content and matric
suction.

8. Acknowledgements

The authors are grateful to the Oklahoma Department of Transportation (ODOT) and Oklahoma
Transportation Center for their financial support of field research. Additional thanks to ODOT
for conducting CPTs and test borings during field research. Funding for the calibration chamber
research funding was provided by the National Science Foundation under Grant No. 9813137,
and is gratefully acknowledged.
9. References

Abedin, M.Z. and Hettiaratchi, D. R. P. (2002). State Parameter Interpretation of Cone

Penetration Tests in Agricultural Soils. Biosystems Engineering, Vol. 83, No. 4, 469–479.

ASTM (2012). D5778-12 Standard test method for electronic friction cone and piezocone
penetration testing of soils. Annual Book of ASTM Standards, 4, 1587-1605.

Hansen, B. (1961). A general formula for bearing capacity. Danish Geotechnical Institute,
Bulletin, 11, 38-46.

Collins, R.W. (2016). In situ testing and its application to foundation analysis for fine-grained
unsaturated soils. Ph.D. Dissertation, University of Oklahoma, 217 p.

Collins, R., & Miller, G. A. (2014). Cone penetration testing in unsaturated soils at two
Instrumented test sites. In Proceedings of the 6th International Conference on Unsaturated Soils,
UNSAT, Sydney, Australia, 2-4.

Durgunoglu, H. T., & Mitchell, J. K. (1973). Static penetration resistance of soils. For NASA
Grant NGR 05-003-406: Lunar Soil Properties and Soil Mechanics, Space Sciences Laboratory
Series 14, Issue 24, UC Berkeley, 223 p.

Fredlund, D. G., & Xing, A. (1994). Equations for the soil-water characteristic curve. Canadian
Geotechnical Journal, 31(4), 521-532.

Freitag, D. R., Green, A. J., & Melzer, K. J. (1970). Performance evaluation of wheels for lunar
vehicles (No. AEWES-TR-M-70-2). Army Engineer Waterways Experiment Station, Vicksburg,
MS.

Ghionna, V. N., & Jamiolkowski, M. (1991). A critical appraisal of calibration chamber testing
of sands. In Proceedings of the 1st International Symposium on Calibration Chamber Testing
(ISOCCTI), Potsdam, NY Elsevier, 13-39.

Houston, S. L., Houston, W. N., & Anderson, T. W. (1995). Moisture and strength variability in
some Arizona subgrades. Transportation Research Record, (1481), 35-43.

Hryciw, R. D., & Dowding, C. H. (1987). Cone penetration of partially saturated sands.
Geotechnical Testing Journal, 10(3), 135-141.

Jarast, S. P., & Ghayoomi, M. (2016). Simple numerical model to simulate penetration testing in
unsaturated soils. In E3S Web of Conferences (Vol. 9). EDP Sciences.

Jarast, S. P., & Ghayoomi, M. (2017). Numerical Evaluation of the Effect of Unsaturated Soil
above the Water Table on Cone Resistance. In Geotechnical Frontiers, 623-632.
Lehane, B. M., Ismail, M. A., & Fahey, M. (2004). Seasonal dependence of in situ test
parameters in sand above the water table. Géotechnique, 54(3), 215-218.

Lo Presti, D., Stacul, S., Meisina, C., Bordoni, M., & Bittelli, M. (2018). Preliminary Validation
of a Novel Method for the Assessment of Effective Stress State in Partially Saturated Soils by
Cone Penetration Tests. Geosciences, 8(1), 30.

Lunne, T., Robertson, P. K., & Powell, J. J. M. (1997). Cone penetration testing in geotechnical
engineering. Spon Press an Imprint of Taylor and Francis Group, 312 p.

Mayne, P. W., & Kemper, J. B. (1988). Profiling OCR in stiff clays by CPT and SPT.
Geotechnical testing journal, 11(2), 139-147.

Massarsch, K. R. (2014, May). Cone penetration testing—a historic perspective. In Proceedings

3rd international symposium on cone penetration testing. Las Vegas, Nevada, USA (pp. 13-14).

Miller, G. A., Azad, S., & Hassell, C. E. (1998). Iowa borehole shear testing in unsaturated soil.
Geotechnical Site Characterization, Proc. of the First International Conference on Site
Characterization, Eds. P.K. Robertson and P.W. Mayne, Balkema, 2, 1321-1326.

Mohamed, F. M., & Vanapalli, S. K. (2015). Bearing capacity of shallow foundations in

saturated and unsaturated sands from SPT–CPT correlations. International Journal of
Geotechnical Engineering, 9(1), 2-12.

Nevels, Jr, J. B. (2006). Spatial distribution of electric cone tip resistance measurements with
depth for a Burleson clay site. In GeoCongress 2006: Geotechnical Engineering in the
Information Technology Age.

Pournaghiazar, M., Russell, A. R., & Khalili, N. (2013). The cone penetration test in unsaturated
sands. Geotechnique, 63(14), 1209.

Robertson, P. K. (2009). Interpretation of cone penetration tests—a unified approach. Canadian

geotechnical journal, 46(11), 1337-1355.

Robertson, P. K. (2016). Cone penetration test (CPT)-based soil behaviour type (SBT)
classification system—an update. Canadian Geotechnical Journal, 53(12), 1910-1927.

Robertson, P.K. and Cabal, K.L. (2015), Guide to Cone Penetration Testing for Geotechnical
Engineering, 6th Edition, Gregg Drilling & Testing, Inc., 133 p.

Robertson, P. K., & Campanella, R. G. (1983a). Interpretation of cone penetration tests. Part I:
Sand. Canadian geotechnical journal, 20(4), 718-733.

Robertson, P. K., & Campanella, R. G. (1983b). Interpretation of cone penetration tests. Part II:
Clay. Canadian Geotechnical Journal, 20(4), 734-745.
Rocha, B. P., dos Santos, R. A., Bezerra, R. C., Rodrigues, R. A., & Giacheti, H. L. (2016).
Characterization of unsaturated tropical soil site by in situ tests. Geotechnical and Geophysical
Site Characterisation 5 – Lehane, Acosta-Martínez & Kelly (Eds), © 2016 Australian
Geomechanics Society, Sydney, Australia, ISBN 978-0-9946261-2-7 ISC-5, 1129-1136.

Russell, A. R., & Khalili, N. (2006). On the problem of cavity expansion in unsaturated soils.
Computational mechanics, 37(4), 311.

Saye, S. R., Lutenegger, A. J., Santos, J., & Kumm, B. P. (2013). Assessing overconsolidation
ratios in soil with piezocone: Referencing soil index properties. Journal of Geotechnical and
Geoenvironmental engineering, 139(7), 1075-1085.

Schmertmann, J. H. (1978). Guidelines for cone penetration test: performance and design (No.
FHWA-TS-78-209). United States. Federal Highway Administration.

Sorensen, K. K., & Okkels, N. (2013). Correlation between drained shear strength and plasticity
index of undisturbed overconsolidated clays. In Proceedings of the 18th International Conference
on Soil Mechanics and Geotechnical Engineering, ISSMGE, Paris, 1-6.

Tan, N.K. (2005). Pressuremeter and cone penetrometer testing in a calibration chamber with
unsaturated Minco Silt. Ph.D. Dissertation, University of Oklahoma, 198 p.

Tan, N. K., Miller, G. A., & Muraleetharan, K. K. (2003). Preliminary laboratory calibration of
cone penetration in unsaturated silt. In Proceedings of the 12th Pan American Conference on Soil
Mechanics and Geotechnical Engineering, Cambridge, Mass, Vol. 1, 391-396.

Tang, C. T., Borden, R. H., & Gabr, M. A. (2017). Approach for Estimating Effective Friction
Angle from Cone Penetration Test in Unsaturated Residual Soils. Journal of Geotechnical and
Geoenvironmental Engineering, 143(11), 04017087.

Vesic, A. S. (1972). Expansion of cavities in infinite soil mass. Journal of Soil Mechanics &
Foundations Div, ASCE, 98(sm3), 265-290.

Vanapalli, S. K., Fredlund, D. G., Pufahl, D. E., & Clifton, A. W. (1996). Model for the
prediction of shear strength with respect to soil suction. Canadian Geotechnical Journal, 33(3),
379-392.

Yang, H., & Russell, A. R. (2013). Preparing unsaturated samples of decomposed granite for
laboratory controlled CPTs. In Proc. 1st Pan-Am Conf. on Unsaturated Soils, Cartagena, 191-
197.

Yang, H., & Russell, A. R. (2015a). Cavity expansion in unsaturated soils exhibiting hydraulic
hysteresis considering three drainage conditions. International Journal for Numerical and
Analytical Methods in Geomechanics, 39(18), 1975-2016.
Yang, H., & Russell, A. R. (2015b). Cone penetration tests in unsaturated silty sands. Canadian
Geotechnical Journal, 53(3), 431-444.

Yu, H. S., & Mitchell, J. K. (1998). Analysis of cone resistance: review of methods. Journal of
geotechnical and geoenvironmental engineering, 124(2), 140-149.

Zapata, C. E., Houston, W. N., Houston, S. L., & Walsh, K. D. (2000). Soil–water characteristic
curve variability. ASCE, Advances in unsaturated geotechnics, GSP 99, 84-124.

Zawadzki, Ł., & Bajda, M. (2016). Impact of the moisture content in medium sands on CPTU
test results. Annals of Warsaw University of Life Sciences–SGGW. Land Reclamation, 48(3),
221-231.
Table 1. Summary of articles addressing CPT in unsaturated soil.
Source Soil Types Description
Investigated penetration resistance with varying relative
Freitag et al. 1970 sand
density and water content, in small laboratory samples.
Studied penetration resistance in the laboratory in dry and
Durgunoglu and Mitchell
silty sand moist soil to validate a bearing capacity based theory for
1973
interpreting tip resistance in "cohesion-friction" soils.
Laboratory study of penetration resistance at varying
Hryciw and Dowding 1987 sand
degree of saturation in small laboratory samples.
sands, silts, and clays of Examined influence of water content on penetration
Houston et al. 1995
various plasticity resistance at 18 field sites, primarily sandy and silty soils.
Abedin and Hettiaratchi Developed a state parameter-penetrometer function based
sandy loam
2002 on miniature penetrometer tests in a calibration chamber.
Studied influence of matric suction on tip resistance in a
Tan et al. 2003, Tan 2005 clayey silt
calibration chamber.
Studied influence of water content and suction variations
Lehane et al. 2004 poorly graded sand
on tip resistance and skin friction at a field site.
Compared tip resistance, water content and suction from
Nevels 2006 lean to fat clay
CPTs in the field, obtained during wet and dry seasons.
Used a constitutive model for unsaturated soils to study
Russell and Khalili 2006 sand and clay cylindrical and spherical cavity expansion relative to
interpretation of pressuremeter and CPT results.
Studied influence of suction on penetration resistance of
Pournaghiazar et al. 2013 sand
sand in a calibration chamber.
Developed a method for preparing calibration chamber
Yang and Russell 2013 silty sand
test beds of unsaturated soil.
Studied tip resistance and skin friction from CPTs at two
Collins and Miller 2014,
lean to fat clay instrumented test sites in relationship to seasonal moisture
Collins 2016
content and suction changes.
Developed a relationship between CPT penetration
Mohamed and Vanapalli
poorly graded sand resistance and bearing capacity from plate load tests in
2015
saturated and unsaturated soil.
Developed an unsaturated cavity expansion analysis for a
Yang and Russell 2015a silty sand
soil exhibiting hydraulic hysteresis.
Studied cone penetration tip resistance under suction
Yang and Russell 2015b silty sand
variations in a calibration chamber.
Reported tip resistance and skin friction in comparison to
Rocha et al. 2016 clayey fine sand water contents obtained a test site during wet and dry
seasons.
Studied influence of saturation on CPTU results in sand
Zawadzki and Bajda 2016 sand
model constructed in a circular cavity in the ground.
Jarast and Ghayoomi 2016 Developed a simple numerical model for modeling cone
sand
and 2017 penetration in unsaturated soil.
Developed an approach for estimating effective friction
highly plastic silt and
Tang et al. 2017 angle from cone penetration test at a field site in
clayey sand
unsaturated residual soils.
Validated a method they developed for assessing effective
Lo Presti et al. 2018 high plasticity clay
stress in unsaturated soils via the CPT
Table 2: Soil identification, average properties and source.
Soil Test Soil UCSC
I.D. Site Name Classification LL PI %F* %S# Source
CL-ML, Silty
1 Minco Silt 23 6 73 27 Tan 2005
clay with sand
North Base Collins
2 CL, Lean clay 47 29 91 9
Soil 1 2016
North Base Collins
3 CL, Lean clay 35 19 89 11
Soil 2 2016
Goldsby Collins
4 CL, Lean clay 32 11 92 8
Soil 1 2016
Goldsby CL, Lean clay Collins
5 30 9 80 20
Soil 2 with sand 2016
MH, Elastic Nevels
6 Burleson Clay 55 20 > 85+ < 15
silt 2006
SP, Poorly Lehane et
7 Perth Sand NA 0 <5 > 95
graded sand al. 2004
Yang and
Lyell Sand,
8 SM, Silty sand NA 0 27 73 Russell
e>0.63
2015b
Yang and
Lyell Sand,
9 SM, Silty sand NA 0 27 73 Russell
e<0.63
2015b
Piedmont MH, Elastic Tang et al.
10 61 22 88 12
Silty Soil silt 2017
Piedmont SC, Clayey Tang et al.
11 34 11 32 68
Clayey Soil sand 2017
Notes: * % Fines, # % Sand, + based on classification.
Table 3: Summary of linear regression results for equation y=yo+mx for lines in Figures
7 thru 9. Data obtained by authors.
Soil
I.D. x y yo m r2 x y yo m r2
1 w qc 5.11 -0.22 0.75 w log(qc/vo) 1.70 -0.03 0.77
2 w qc 6.31 -0.23 0.64 w log(qc/vo) 3.34 -0.06 0.65
3 w qc 7.01 -0.25 0.44 w log(qc/vo) 2.41 -0.03 0.07
4 w qc 10.22 -0.42 0.58 w log(qc/vo) 3.03 -0.04 0.20
5 w qc 12.18 -0.44 0.25 w log(qc/vo) 3.28 -0.07 0.41
1 log(ua-uw) qc 6.24 2.01 0.70 log(ua-uw) log(qc/vo) 1.85 0.26 0.72
2 log(ua-uw) qc 1.99 2.13 0.64 log(ua-uw) log(qc/vo) 2.23 0.55 0.65
3 log(ua-uw) qc 2.80 2.12 0.52 log(ua-uw) log(qc/vo) 1.99 0.37 0.21
4 log(ua-uw) qc 6.05 2.50 0.34 log(ua-uw) log(qc/vo) 2.64 0.31 0.20
5 log(ua-uw) qc 8.04 3.85 0.25 log(ua-uw) log(qc/vo) 2.57 0.61 0.44
Note: w in %, qc and ua-uw in MPa.

Table 4: Summary of linear regression results for equation y=yo+mx for lines in Figures
11 and 12. Data obtained from literature.
Soil
I.D. x y yo m r2 x y yo m r2
6 log(ua-uw) qc 9.26 5.10 0.52 log(ua-uw) log(qc/vo) 3.35 0.96 0.50
7 log(ua-uw) qc 12.53 2.68 0.36 log(ua-uw) log(qc/vo) 2.22 0.11 0.07
8 log(ua-uw) qc 5.28 0.95 0.02 log(ua-uw) log(qc/vo) 2.17 0.33 0.17
9 log(ua-uw) qc 11.35 2.10 0.08 log(ua-uw) log(qc/vo) 1.75 -0.01 0.00
10 log(ua-uw) qc -1.03 -2.38 0.26 log(ua-uw) log(qc/vo) 2.54 0.66 0.29
11 log(ua-uw) qc 6.01 0.40 0.00 log(ua-uw) log(qc/vo) 3.89 1.42 0.63
Note: w in %, qc and ua-uw in MPa.
Table 5: Parameters used for bearing capacity analysis via Equation 4.
From Lab
Testing Used with Bearing Capacity Model
' c' ' c' B
Soil Test
I.D. Soil
USCS Class. (degs.) (Mpa) (degs.) (Mpa) (m) Nc Nq c, q
CL-ML,
Minco
1 Silty clay 30 0.012 46 0.0 0.0178 118 491 1.32
Silt
with sand
North
CL, Lean
2 Base NA NA 13 0.04 0.0357 6.4 19 1.20
clay
Soil 1
North 23.3, 0.027,
CL, Lean
3 Base 26.4, 0.0, 20 0.04 0.0357 9.5 33 1.20
clay
Soil 2 32.3 0.006
Goldsby CL, Lean
4 20 0.019 30 0.04 0.0357 20 81 1.21
Soil 1 clay
CL, Lean
Goldsby 19.2, 0.026,
5 clay with 35 0.04 0.0357 31 135 1.24
Soil 2 24.7 0.0
sand
 0.1 Soil 1
Minco Silt
best fit
0.01
measured, gd=15.7 kN/m
3

measured, gd=14.1 kN/m3

0.001 measured, gd=13.4 kN/m
3

measured, gd=12.6 kN/m

3
ua- uw (MPa)

0.0001
Soils 2 & 3
100
North Base Test Site
10
1 pressure plate
Soil 2 field total suction
0.1 Soil 3 field total suction
Soil 2 predicted ua-uw SWCC
0.01
Soil 3 predicted ua-uw SWCC
0.001
Soils 4 & 5
100
Goldsby Test Site
10
1
0.1 pressure plate
Soils 4 & 5 field total suction
0.01 Soil 4 & 5 predicted ua-uw SWCC
0.001
0 5 10 15 20 25 30 35
w (%)

Figure 1: Soil Water Characteristic Curves for Soil 1 (Minco Silt), Soils 2 and 3 from the North
Base Test Site and Soils 4 and 5 from the Goldsby Test Site.
 0.0

0.5
Depth (m)

1.0

CPT-1 CPT-1 CPT-4

CPT-4 CPT-4 Tens.
bottom of soil bed
1.5
1 2 3 4 5 6 0.00 0.04 0.08 0 2 4 6 8 10 12 11 12 13 14 15
qc (MPa) ua- uw (MPa) w (%) gd (kN/m3)

Figure 2: Results of CPT-1 and CPT-4 in the OU Calibration Chamber (under isotropic confining
stress of 100 kPa): tip resistance (qc), matric suction (ua-uw), water content (w) and dry unit
weight (d) versus depth from top of soil bed.
 CPT-1 CPT-2 CPT-3 CPT-4 CPT-5 CPT-6
4.5
qc=5.11-0.22w
r² =0.75 qc=2.31+26.3(ua- uw)
4.0 r² =0.75
qc (MPa)

3.5

3.0

2.5
4 6 8 10 12 0.00 0.02 0.04 0.06 13.0 13.5 14.0 14.5 15.0
w (%) ua- uw (MPa) d (kN/m3)

Figure 3: Summary of CPT results in the OU Calibration Chamber (under isotropic confining
stress of 100 kPa): average tip resistance versus water content, matric suction, and dry unit
weight.
 0.0
CL
PIavg=10

0.5

CL, CH
PIavg=29
Depth (m)

1.0

1.5

CL
PIavg=19
2.0
PL
LL % fines
wn % sand
2.5
USCS 10 20 30 40 50 10 20 30 0 30 60 90 12 15 18 0 30 60 90
Group
PL, LL, w (%) PI % fines gd (kN/m )3 S (%)
Symbol & sand

Figure 4: Summary of soil properties determined for the North Base Test Site.
 0.0

CL
0.5 PIavg=11
% sand=8

1.0
Depth (m)

1.5

2.0

CL
2.5 PIavg=9
% sand=20

3.0
PL
% fines
LL
wn % sand
3.5
USCS 0 10 20 30 40 50 5 10 15 0 30 60 90 12 15 18 0 30 60 90
Group
PL, LL, w (%) PI % fines gd (kN/m )3
S (%)
Symbol & sand

Figure 5: Summary of soil properties determined for the Goldsby Test Site.
 0.0

0.5

1.0
Depth (m)

1.5
02/01/2013
05/06/2013
2.0 09/03/2013
11/21/2013
02/18/2014
2.5 09/10/2014
12/30/2014

3.0

3.5
0 2 4 6 8 0 2 4 6 8 10 15 20 25 30 0 1 2 3 4
qc (MPa) FR (%) w (%) ua-uw (MPa)

Figure 6: Summary of CPT tip resistance, friction ratio (FR), water content and matric suction
determined at various times over a period of approximately two years at the North Base Test
Site.
 0.0

0.5

1.0
Depth (m)

1.5

2.0

02/01/2013
2.5 05/06/2013
07/29/2013
11/21/2013
02/18/2014
3.0
09/10/2014
02/25/2015

3.5
0 2 4 6 8 0 2 4 6 8 0 5 10 15 20 0 1 2 3 4
qc (MPa) FR (%) w (%) ua-uw (MPa)

Figure 7: Summary of CPT tip resistance, friction ratio, water content and matric suction
determined at various times over a period of approximately two years at the Goldsby Test Site.
 12
10
8
FR (%)

6
4
2
1: CL-ML, Minco Silt
0 2: CL, N.B. Layer 1
3: CL, N.B. Layer 2
8 4: CL, Goldsby Layer 1
5: CL, Goldsby Layer 2
qc (MPa)

6 Regression Line

0
0 5 10 15 20 25 0.01 0.1 1 10
w (%) ua- uw (MPa)

Figure 8: Water content and matric suction plotted against tip resistance and friction ratio
obtained from OU calibration chamber tests and from North Base and Goldsby Test Sites.
 1: CL-ML, Minco Silt 3: CL, N.B. Layer 2 5: CL, Goldsby Layer 2
2: CL, N.B. Layer 1 4: CL, Goldsby Layer 1 Regression Line

1000
qc / svo (MPa)

100

10
5 10 15 20 25 0.01 0.1 1 10
w (%) ua- uw (MPa)

Figure 9: Water content and matric suction plotted against tip resistance normalized by vertical
total stress (qc /vo). Data obtained from OU calibration chamber tests and from North Base and
Goldsby Test Sites.
 1: CL-ML, PI=6, S=27% 5: CL, PI=9, S=20% 9: SM, PI=0, S=27%, e<0.63
2: CL, PI=29, S=9% 6: MH, PI=20, S=<15% 10: MH, PI=22, S=12%
3: CL, PI=19, S=11% 7: SP, PI=0, S>95% 11:SC, PI=11, S=68%
4: CL, PI=11, S=8% 8: SM, PI=0, S=27%, e>0.63 Regression Line
10000

1000
vo
qc /

100

10
15
qc (MPa)

10

PI=0 to 6 PI=9 to 33
0
0.001 0.01 0.1 0.01 0.1 1 10
ua- uw (MPa) ua- uw (MPa)

Figure 10: Matric suction plotted against tip resistance and tip resistance normalized by vertical
total stress. Data obtained from OU calibration chamber tests, from North Base and Goldsby Test
Sites, and from the literature (see Table 2 for source information). Data are separated into lower
(on left) and higher (on right) PI ranges.
 6
5
4
m(qc , ua- uw)

3
2 1: CL-ML, PI=6, S=27%
1 2: CL, PI=29, S=9%
3: CL, PI=19, S=11%
0
4: CL, PI=11, S=8%
-1
5: CL, PI=9, S=20%
-2
6: MH, PI=20, S=<15%
1.5 outlier
7: SP, PI=0, S>95%
m(qc /svo , ua- uw)

8: SM, PI=0, S=27%, e>0.63

9: SM, PI=0, S=27%, e<0.63
1.0 10: MH, PI=22, S=12%
11:SC, PI=11, S=68%
Regression Line
0.5

m = 0.192 + 0.019 PI
0.0 r ²=0.50

0 10 20 30
PI

Figure 11: Plots of PI versus slope, m, obtained from linear regression lines in the Figure 10 and
summarized in Tables 3 and 4. Upper graph shows m from log(ua-uw) versus log(qc) and lower
graph shows m from log(ua-uw) versus log(qc /vo).
 1: CL-ML 2: CL, CH 3: CL 4: CL 5: CL
Predicted Tip Resistance
f ' = 46o, c' = 0.00 MPa f ' = 13o, c' = 0.04 MPa
f ' = 20o, c' = 0.04 MPa f ' = 30o, c' = 0.04 MPa
f ' = 35o, c' = 0.04 MPa
10

8
qc (MPa)

0
Soil 1
25 Soil 2
Soil 3
Soils 4&5
20
w (%)

15

10

0
0.01 0.1 1 10
ua- uw (MPa)

Figure 12: Upper graph shows field values of qc and ua-uw plotted alongside lines representing
“best fit” theoretical calculations of tip resistance. Theoretical calculations utilize the bearing
capacity equations of Durgunoglu and Mitchell (1973), shear strength equation of Vanapalli et
al. (1996) and Soil Water Characteristic Curves (SWCC) shown in the lower graph. Model
parameters are given in Table 5.
 Field CPT data:: North Base Soil 2
Predicted Tip Resistance (D/B = 10, d /f = 0)
Best fit: f' = 20 , c' = 0.04 MPa
o

10
f' = 23.3o, c' = 0.027 MPa
f' = 26.4o, c' = 0.0 MPa
f' = 32.3o, c' = 0.006 MPa
} from laboratory tests

8
qc (MPa)

0
0.01 0.1 1 10
ua- uw (MPa)

Figure 13: Actual and predicted values of qc for Test Soil 3: field values from seven different
testing dates over two years.
 10
Field qc Predicted qc
8 2/1/13 f' = 32.3o, c' = 0.006 MPa
2/18/14
f' = 26.4o, c' = 0.0 MPa
f' = 23.3o, c' = 0.027 MPa
qc (MPa)

0
0.01 0.1 1 10
ua- uw (MPa)

Figure 14: Actual and predicted values of qc for Test Soil 3: field values are show for two dates
representing dry and wet conditions.