Development and Testing of a Portable In-Situ Near-Surface Soil Characterization System A Dissertation Presented by Ehsan Kianirad to The Department of Civil and Environmental Engineering in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the field of Geotechnical and Geo-Environmental Engineering Northeastern University Boston, Massachusetts April, 2011
362
Embed
Development and testing of a portable in-situ near …803/fulltext.pdf · Development and Testing of a Portable In-Situ Near-Surface Soil Characterization System A Dissertation ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Development and Testing of a Portable In-Situ Near-Surface Soil Characterization System
A Dissertation Presented
by
Ehsan Kianirad
to
The Department of Civil and Environmental Engineering
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
in the field of
Geotechnical and Geo-Environmental Engineering
Northeastern University Boston, Massachusetts
April, 2011
Abstract Rapid and accurate in-situ measurement of shallow soil properties is still a challenge
and there is a need for new and advanced devices and methods. The RapSochs (Rapid Soil
Characterization System) is newly developed as a man-portable instrument for rapid
comprehensive field characterization of near surface soil properties. Potential applications
include construction quality control, contingency site selection, and quick determination of
load carrying capacity of unfamiliar unpaved airfields and terrains. Sensing technologies
similar to Electronic Cone Penetrometer and a moisture sensor are combined in a small
impact driven system similar to DCP (Dynamic Cone Penetrometer) configuration. The main
objective of this research is to develop and assess methods to interpret geotechnical
properties from the RapSochs measurements.
Several real-size tests are conducted in different soil samples prepared in large soil
cells. The DCP is used as a benchmark for soil strength profiles and RapSochs performance
is compared with that of the DCP. An analytical physics-based energy model to predict soil-
instrument interaction in dynamic penetration is developed. The model is calibrated for
RapSochs and DCP and is used to explain the penetration process. It is shown that this model
is more accurate than the widely-used Dutch formula. The model is used to correlate the
RapSochs penetration rate to DCP and CBR (California Bearing Ratio). The RapSochs-CBR
correlation is proved to predict CBR with acceptable accuracy, higher resolution, and near-
to-ground surface measurements.
The Maximum Likelihood Estimation method is adopted for the average dynamic
cone and friction forces estimation. Cone and friction strength, and friction ratio profiles
similar to those measured by the CPT are developed based on estimated forces. The effect of
variable applied energy on the soil strength estimation is found to be insignificant. This
method is proven to provide acceptable estimation of the soil resistance in different soil
types. Soil classification to cohesive and cohesion-less materials is accomplished using a
chart developed based on cone and friction resistance. Effects of overburden, sample size,
boundary conditions, variable hammer drop height, and penetration rate on RapSochs
measurements are also assessed.
It is concluded that the RapSochs instrument provides consistent, repeatable and
reliable results in laboratory prepared homogenous soil samples. The algorithms and methods
to obtain strength profiles of in-place soil or compacted layers are developed. The estimation
of soil strength and friction resistance, soil classification, and correlation to CBR is achieved.
While the approach is developed specifically for RapSochs, it is applicable to a wide class of
dynamic penetrometers.
DeNevelopNear-S
ment aSurfac
Bosto
and Tee Soil C
by Eh
Ap
Copyright
on, Massach
esting oChara
hsan Kian
pril, 201
2011, Ehsan
husetts
of a Poacteriza
nirad
11
n Kianirad
ortableation S
P a g
e In-SitSystem
g e | ii
tu m
P a g e | i
، مهدي و زري، و مادربزرگ مهربانم، م، كيانا، پدر و مادرعزيزماين پايان نامه را به همسر .كنم حاجي مامان تقديم مي
This dissertation is dedicated to my loving wife, Kiana, my father and mother, Mehdi and Zari, and my dear grandmother, Hajimaman.
P a g e | ii
Acknowledgments
Countless people have supported and inspired me during my doctorial study. Without
them this work would not have succeeded. First of all, I wish to thank my adviser Professor
Akram N. Alshawabkeh for his outstanding support, helpful comments, and editing this
dissertation. I am very grateful for his help during the period of this research. Special thanks
to Mr. Ronald W. Gamache, the principal investigator on the project, without his cooperation
and assistance this research would not have been possible. I really enjoyed working with him
and I learned a lot from him. During last few years, I had the opportunity to work with
Professor Mishac Yegian on numerous projects. I truly thank him for all the things I learned
from him. I am sure these would help me in my future professional carrier. I would like to
thank Professor Arvin Farid, at Boise State University, for his contributions and comments. I
also wish to recognize with sincere appreciation help, collaboration, comments and
suggestions of Professor Luca Caracoglia, David Brady, Thomas Sheahan, Ferdi Hellweger,
and Dr. David Whelpley, at Northeastern University. I really appreciate all the students
assisted me with their hard-work during experiments with RapSochs: graduate students
Ibrahim El-Shawabkeh and Payam Bakhshi, undergraduate students Eilish Corey, David
Diaz, and Christopher Wiley, and high school students Patt Hongsmatip and Feng Wu. I also
appreciate helps of Dr. Eduard Kleyn, from South Africa, and Mr. Kaveh Mahdaviani, from
University of Alberta, Canada. I would like to thank Dr. Mina Hassan-Zahrae and Amir
Afkhami for hosting me in my first year in Boston and all their advices.
This work is partially supported by U.S. Army Corps of Engineers, Engineer
Research and Development Center (ERDC) under contract W912HZ-06-C-0063. The author
conveys his appreciation to TransTech Systems, Inc. and Applied Research Associates, Inc.
for their collaboration. The author would also like to thank the Bernard M. Gordon Center for
Subsurface Sensing and Imaging Systems (Gordon-CenSSIS) at Northeastern University for
their support.
Last but not least, I take this opportunity to express my deepest gratitude to my
beloved wife, Kiana, my father and mother, Mehdi and Zari, and my dear grandmother,
Hajimaman for their wholehearted support and patience during this study. I dedicate this
dissertation to them.
P a g e | iii
Table of Contents
Table of Contents ................................................................................................................... iii
List of Figures .......................................................................................................................... x
List of Tables ........................................................................................................................ xiv
RapSochs is a combination of CPT sensing technologies and electrical impedance
spectroscopy in a small impact driven system similar to portable DCP configuration for in-
situ characterization of near surface soil properties. In technologies that involve dynamic
sounding, recorded or measured data are indices as either, penetration per blow (e.g., DCP),
or number of blows needed for a specific penetration (e.g., SPT). The indices are usually
correlated to material properties (e.g., SPT N-values to undrained shear strength) or used
directly in design or evaluation (e.g., SPT N-values for calculation of net bearing capacity of
spread footings) after comprehensive experimental studies.
Due to the large forces required for CPT or dilatometer operation, the units are
vehicle mounted to provide the requisite reaction force. For this reason, the constant push
method of advancing the cone is not appropriate for a portable field device. However, most
of the research and practice related to sensing soil strength, type, and moisture in a
penetrometer configuration has been accomplished on CPT style devices. While a rigorous
theoretical analysis of quasi-static cone penetration is still difficult, the same approach to
dynamic penetration is more challenging, considering cost, instrumentation, and processing
limitations.
The only approach practical for portable use in the field is a variation of the standard
DCP. In its most basic form, the DCP consists of a rod fitted with a conical tip that is driven
into the soil by energy provided by a slide hammer. The hammer is dropped a fixed distance
onto an anvil attached to the rod thereby transferring the kinetic energy to the conic tip. If
the energy is high enough the soil fails in shear and the tip advances. The penetration of the
rod into the soil, as a result of the imparted energy, is related to the strength of the soil. To
provide the wide dynamic range, a dual mass hammer is utilized to better cover the range.
Other devices, such as the Sol Solution PANDA (Langton, 2001), use a fixed mass
hammer. However, the impact is controlled manually and measured electronically to better
cover the full range of soil strength encountered in the field. Alternatively, the USACE, U.S.
Army, and U.S. Air Force use the trafficability cone penetrometer and the airfield cone
penetrometer, which are pushed by a person into the soil. Those instruments, although still in
use, in most projects, are replaced by DmDCP (Dual Mass DCP) to evaluate soils
trafficability, predict ground strength for vehicle operations, or field testing for pavement and
airstrips design and construction.
P a g e | 9
Although several studies on CPT, SPT, and DCP are conducted in the past, they are
not directly applicable to RapSochs type applications. One critical issue associated with
interpretation of measured data is the dynamic impact compared to the quasi-static nature of
CPT type devices. Understanding of the recorded strain and acceleration signals is another
challenge in this project. On the other hand, while the mechanical behavior of soil under
static loading is well studied, the dynamic modeling of loading and failure of soil during a
penetration test is not sophistically formulated. This is one of the main reasons for the
experimental and statistical approach to correlate RapSochs properties and measurements to
soil characteristics.
This chapter starts with the introduction of DCP, the historical developments in
related researches and applications, followed by a discussion about advantages and
drawbacks of the instrument. An introduction of the CPT and incorporated sensing
technologies is also presented. Different types of CPT are compared and the measurements
and applications are explained. Later, other technologies, instruments, and test methods used
for near surface in-situ soil characterization, important for trafficability, are reviewed. The
measurements are explained and the advantages and disadvantages of instruments and
methods are discussed. Since the moisture content development and calibration was not
within the scope of this study, in-situ moisture content measurement technologies are not
discussed.
RapSochs is configured as a miniature pile driver that employs an impact system to
advance the cone into the soil. Because of the mechanism similarity, a short discussion on
pile driving equations, pile capacity evaluation and dynamic energy measurement is
presented in the last section.
2-2- DCP
The Dynamic Cone Penetrometer (DCP) is an instrument designed to measure the in-
situ strength of fine-grained and granular subgrades, granular base and subbase materials, and
weakly cemented materials. It is a handheld device designed to penetrate soils to depths of 1
m with a 20 mm (0.79 in.) diameter cone. The 60-degree cone is forced into the ground by
raising and dropping an 8 kg (17.6 lb) hammer. Figure 2-1 shows the schematic of DCP.
P a g e | 10
Two people are usually required to run
a regular DCP test. However, in instrumented
types where data are logged by an electronic
device, the manpower is reduced to one
person. Application of DCP may be either
direct use of the DCPi for design and
evaluation or indirectly by correlations to other
parameters. The average penetration per blow
for a certain penetration depth is usually used
as an index for in-situ shear strength and it is
correlated to CBR, resilient modulus or other
soil parameters. However, there is no
consistent method to obtain the DCP index. In
this chapter, this index is referred to as DCPi
regardless of the method it is calculated. In
Chapter 4, the procedure to conduct a DCP test
is explained. In Chapter 5, DCPi calculation,
data presentation, and analysis of DCP test
results are discussed. In Chapter 6, existing
correlations to CBR are summarized. The
corresponding standard test method is ASTM
D6951, which was introduced in 2003.
DCP Historical Developments
Although early versions of the DCP
had a 30-degree cone, 60-degree cone become
more popular in latest years due to its
durability in high-strength materials. The cone
angle in the current ASTM method is 60
degrees.
Scala (1956) introduced the dynamic
cone penetrometer based on the previous
designs in Switzerland. The hammer drop
Figure 2-1: Schematic of DCP Device.
Anvil
Hammer, 8 Kg
Hammer Guide
Safety Handle
Steel Rod, 16 mm diameter
The Cone
20 mm
Cone angle 60°
Dro
p H
eig
ht, 5
75 m
m (
22.
6 in
)
Reference Point
P a g e | 11
height was 20 inches and the hammer weight was 20 lb. The cone angle was 30 degrees with
0.5 in2 surface area (0.8 inch = 20.3 mm diameter). Scala (1956) penetrometer was used with
an extension to investigate to a depth of 1.8 m below ground. He developed the theoretical
relationship between the applied energy, soil resistance and penetration rate, and he was the
first to develop the DCP-CBR correlation and use DCP for pavement design. Gawith and
Perrin (1962) reported the use of the same DCP in Australia and using a DCP-CBR
correlation curve. In South Africa, Van Vuuren (1969) introduced the modern DCP by
modifying the penetrometer, which has been in use in Australia. It was made of a 10-kg
hammer sliding on a 16-mm rod dropping from 460 mm height. The cone was 20 mm in
diameter. Van Vuuren (1969) presented the first DCP to in-situ CBR calibration for
moderately fine-grained soils.
Sowers and Hedges (1966) introduced a DCP device with a 15-lb (≈ 6.8-kg) hammer,
falling 20 in (508 mm) on the driving rod. The cone point was enlarged to minimize the
circumferential resistance. It was used for field exploration and verification of soil conditions
at individual footings. Their tests were performed in augured holes.
Since 1973, the DCP has been used in South Africa (Kleyn, 1975). The version used
in South Africa consisted of an 8-kg hammer dropping from 575 mm height with a 30-degree
cone, which was 20 mm in diameter. Kleyn (1975) is one of the pioneers who discovered the
linear relationship between DCPi and CBR on a log-log scale. After running tests on samples
with different moisture contents, compacted with the same effort, Kleyn (1975) concluded
that the DCP and CBR react in a similar manner to varying moisture content and dry density.
Trasvaal Provincial Administration (1978) in South Africa was the first organization to
suggest the use of a minimum DCPi values at different depth for pavement materials design
subjected to heavy, medium, and light traffic.
Kleyn et al. (1982) listed various applications of the DCP in pavement design, road
construction, and pavement evaluation and monitoring. They reported that the DCP measures
in-situ CBR rather than laboratory soaked CBR, and that the DCP correlates better with
pavement’s field performance than the laboratory soaked CBR. In a particular test, it was
demonstrated that DCP can detect the deterioration of pavement materials very well.
However, Kleyn and Savage (1982) excluded the cemented materials since they carry loads
in bending and are subject to fatigue damage. The DCP does not evaluate these materials in a
manner that relate to their behavior in the field. They presented a design and evaluation
method for thin surface unbound gravel pavements using DCP.
P a g e | 12
Smith and Pratt (1983) provided a correlation between DCPi (30-degree cone,
hammer weighted 9.08 kg, and dropping 508 mm) and in-situ CBR tests in clayey materials.
They concluded that the DCP results are as acceptable as the in-situ CBR while the
coefficient of variation (CV) of DCP tests is smaller than that of the in-situ CBR tests at the
same location. They compared the CBR values for materials molded at field moisture content
and density and in-situ CBR and recommended in-situ CBR and DCP measurements over
laboratory CBR tests for pavement evaluation.
Sampson (1984) reported that the DCP is used to estimate the bearing capacity of
subbase and base layers composed of coarse, granular, and stabilized materials. 2.54 mm and
5.08 mm penetration CBRs were used to obtain a DCP-CBR correlation using 60-degree
cone. Based on plasticity of the materials, different DCP-CBR correlations were proposed.
To improve the correlation, other soil parameters including grading modulus, plastic limit,
and dry density were incorporated into the correlation equation. It was concluded that, in
every case, the correlation of DCPi to the CBR of 5.08 mm penetration was better than that
of the 2.54 mm penetration.
In 1986, the Council for Scientific and Industrial Research (CSIR of South Africa)
developed the first software package for evaluation and analysis of DCP data, which has
been updated several times since then (CSIR Transportek, 2000).
Harison (1986, 1987), provided theoretical explanation for the linear log-log relation
of DCP and CBR. He performed 72 pairs of DCP and CBR tests on clay-like, well-graded
sand, and well-graded gravel samples prepared in standard CBR molds and presented
correlation equations. The regression analysis showed that the log-log model relates DCP and
CBR better than the inverse model. It was concluded that moisture content and dry density
have similar effects on CBR and DCP, and therefore, the DCP-CBR correlation may not be
affected by these variables. It was also concluded that the soaking process does not have a
significant effect on the calibration.
Livneh and Ishai (1987) used a dynamic cone penetrometer with a 30-degree cone for
pavement evaluation in Israel. Based on laboratory and field tests on a wide range of natural
and compacted materials, they suggested a correlation between DCPi and CBR. However,
they did not provide the soil classification and other material parameters. Based on the CBR
P a g e | 13
correlation, they developed methods for evaluation of airport and highway pavement as well
as evaluation of the dynamic stiffness modulus and load classification number.
Livneh (1987) concluded that the coefficient of variation of the CBR results for any
particular material is considerably higher than that of the DCP.
Chua (1988) developed a one-dimensional model for DCP penetration to back
calculate the elastic modulus of the soil. The model assumes a horizontal disc on which the
cone penetration causes a plastic deformation due to the plastic shock wave propagation. He
presented the results as series of graphs for different soils that correlate DCPi to elastic
modulus.
Chua and Lytton (1988) used a DCP with an accelerometer mounted on top of the
handle to analyze the dynamics of the system. A simple model of springs and dashpots
representing hammer-rod-soil interactions was developed. Capability of determination of the
damping ratio of the soil was demonstrated.
Harrison (1989) presented a new correlation between DCPi and CBR, which is
corrected to account for the confinement effects of laboratory CBR tests. He also reported
that the DCP test results in lower coefficient of variation than the CBR, and therefore, it is
more repeatable than CBR test.
Livneh (1989) showed that the CBR values derived from a DCP, with a 30-degree
cone, is different than that from a DCP with a 60-degree cone. The DCP and in-situ CBR
tests was performed on clay mixed with fine gravel and heavy clay (of subgrade of airfield
runways) and silty soil (of urban roads). It was concluded that in-situ CBR values obtained
from DCP tests can be used with plausible reliability. The effect of overburden pressure on
the results of tests in the above-mentioned materials was negligible. However, it was pointed
out that the difference in geographic areas may lead to changes in the empirical correlation
equations.
Ayers et al. (1989) examined DCPi to shear strength correlations for a range of
granular materials. The equations correlate the DCPi to deviator stress under different
confining pressures. The tested materials included sand, sandy gravel, and crushed dolomitic
ballast with different percentage of fines. They emphasized the role of the confining pressure
P a g e | 14
under field loading conditions. Ayers (1990) reported that overburden pressure, geometry of
the cone (diameter and angle) have a significant effect on DCP penetration.
Buncher and Christiansen (1991), after comparing Electric Cone Penetrometer results
with DCP and in-situ CBR, concluded that the DCP is very susceptible to skin friction in
cohesive soils. They reported that, in all cases, the DCP values increased as the depth
increased thru the cohesive soils.
De Beer (1991) presented a method to use DCP for flexible road design. He also
presented and empirical relationship between the elastic modulus and DCPi based on
calibration of the one-dimensional linear model with depth deflection measurements by
heavy vehicle simulator.
Livneh (1991) reported that, in-situ and in-the-CBR-mold DCP tests with 60° and 30° cone
showed significant different results. The DCP values for 30° cone were approximately 10%
greater than those obtained with the 60° cone. He also developed a correlation between CBR
and elastic modulus.
Webster et al. (1992) from USACE developed a correlation between DCP and in-situ
CBR based on tests in various materials. They also presented a procedure to use DCP for
evaluating unsurfaced soil or aggregate surfaced roads and airfields for military vehicles and
aircrafts. A Dual mass DCP (DmDCP), where the 17.6-lb (8-kg) hammer was convertible to
a 10.1-lb (5-kg) hammer, was used. A lighter hammer was used in materials with CBR values
of less than 10, and the cone penetration rate was multiplied by two to obtain an equivalent
penetration rate by a 17.6-lb hammer. This provision allows having a good resolution and
therefore more accurate measure of soil strength in weak materials. Use of disposable cone
was also introduced. In soils where the standard cone is difficult to remove, it eliminates the
need for an extraction jack or tremendously reduces the manpower needed to run several
tests. It was suggested in their work to stop the test, if the penetration of more than 25 mm
was not achieved after 10 blows since the hard material will damage the instrument. Their
work was a part of a project named “soil strength determination for non-paved operating
surfaces”.
Livneh et al. (1992) described a pneumatic automated DCP, which needs a
compressor for operation. That system was able to run up to 24 blows per minute. In their
paper, there is not a mention of any data collection system, so it may be assumed that it was
P a g e | 15
remained manual. They compared the DCP results from the automated and manual system.
Although the regression analysis showed that the automated DCP resulted lower values than
the manual DCP, but the statistical analysis indicated that they are identical. However, CBRs
from the automated system, calculated using Livneh’s equation (Livneh, 1987), were on
average 14% smaller than the ones obtained from the manual DCP. They examined the effect
of blow rate in sandy clay, but it was found not influential. They did not inspect the effect of
the penetrometer and guides weight as its effect is formulated in Dutch equation. The weight
of the automated system is not reported in their paper. The conclusion that DCP results are
identical but the calculated CBR is smaller is questionable.
Weintraub (1993) developed an automated DCP to be used for measurement of
unsurfaced airfield bearing strength. The mechanical design process and relevant details are
explained in his work. He also found that the results of DCP and the automated DCP are not
similar and have developed a correlation. After numerous tests, he validated the reliability
and consistency results of DCP.
Burnham and Johnson (1993) studied the use of the DCP for in-situ characterization
of soil profiles. The presented examples of its application in preliminary soil surveys,
embankment and back-fill construction control, structural evaluation of existing pavements,
and supplementing foundation testing for design purposes.
Webster et al. (1994) examined the minimum penetration depth required in DCP to
measure the strength of surface layers. The required depth is reported between 1 to 11 inches
(2.5 to 28 cm) for materials ranging from highly-plastic clay to poorly-graded sand. It was
also shown that the DCP can determine the thickness and location of a weak soil layer in a
pavement. They verified the DCP-CBR correlation proposed by Webster et al. (1992) and
proposed to use new correlations for CH (High-plasticity clay) and CL (Low-plasticity clay)
clays.
Ese et al. (1994) showed that a DCPi of less than 2.6 mm/blow in the well-graded
gravel base layer is critical to have a good serviceability in a road. It was concluded that the
DCP tests during thawing give the best correlation to the serviceability of a road in Norway.
They reported variation of DCP values due to variations of the moisture content. In DCP-
CBR correlation, for well-graded gravel with 9% to 19% fines, they showed that this
correlation is independent of moisture content and dry density. An automated DCP with
computerized data collection was developed to reduce the number of operators, overcome the
P a g e | 16
difficulties of running more DCP tests, reduce the instrument’s dependency to operator, and
to reduce the test time by automatic data recording.
Bratt et al. (1995) developed a DCPi to dry density correlation. They showed that
DCP could substitute moisture-density tests for compaction construction control of
embankment and subgrade inspection.
Truebe et al. (1995) used DCP to evaluate the strength of a low volume road of Forest
Service. They presented a DCP to in-situ CBR correlation for the aggregate surface and
subgrade.
Livneh et al. (1995) verified the vertical confinement effect of 1) granular layers, 2)
cohesive layers, and 3) rigid structural layers on clayey materials, as well as 4) the effect of
upper asphalt layers on the DCPi of granular materials. They concluded that the vertical
confinement does not have any effect on DCPi of a cohesive subgrade layer. However, DCP
measurement in granular soil depends on the vertical confinement. They recommended that
for pavement evaluation purposes any DCP measurement should be conducted through a
narrow boring in asphalt layer and not after removal of a wide strip of asphalt. Equations to
obtain unconfined DCPi based on confined DCPi for different top layers were presented.
They also described the theoretical background of vertical confinement effects and different
failure mechanisms. The DCP used in their research was a 30-degree angled cone.
Hassan (1996) examined existing correlations between DCPi and resilient modulus
for sand and fine-grained soils. The specimens were Oklahoma soils molded and compacted
in a 6-in-diameter and 12-in-height special mold. The DCPi was calculated by dividing the
sample depth (equal to the length of penetration, which is 12 in.) to the number of blows. The
experiment results showed that, in fine-grained soils, the increase in moisture content above
the optimum significantly increase DCPi, increase in soil dry density decreases DCPi, and an
increase in confining pressure does not significantly affect DCPi. In granular soils, it was
showed that the confining pressure is a significant factor affecting DCPi. However, this
effect is less for materials with higher coefficient of uniformity. They also found that the
maximum aggregate size has an insignificant effect on DCPi. A correlation between DCPi
and resilient modulus was developed in fine-grained soils at optimum moisture content. It
was stated that specific relationship may be developed for particular moisture state and
should not be generalized for all conditions.
P a g e | 17
MnDOT (1996) introduced a user guide to determine the fabrication details, test
procedures, applications, available methods, and correlations of DCP to be used in the
Minnesota department of transportation (MnDOT). This user guide is one of the few
instructions that covers the instruments specification and data recording and processing
procedures to details. It determines the combined weight of the upper shaft, anvil, lower
shaft, and cone to be 3.1 kg, approximately. In addition, dimensions and material properties
of hammer and anvil are determined. To record the instrument’s reading when one person
operates the DCP the use of video camera is proposed. To extract the DCP, use of a farm-
purpose jack is recommended, and striking the hammer against the handle is prohibited. The
penetration after each blow is used as DCPi. In a road subgrade, DCPi of greater than 25
mm/blow was considered as criteria for the need of remedial measures. In the pavement edge
drain backfill, DCPi of less than 75 mm/blow indicates satisfactory compaction. For various
base materials, limiting values of DCPi, assuming adequate confinement near the test
surface, are also proposed. Other applications of DCP are listed and using of automated DCP
with computerized data recording is reported.
Al-Refeai and Al-Suhaibani (1997) compared the DCP and CBR tests in laboratory
prepared samples of poorly-graded sand, silty sand, and clay and showed that both tests
respond in a similar manner to changing moisture content and dry density. They also
developed several DCPi-CBR correlations for the tested materials. They concluded that
variability in DCPi versus CBR data increases as the soil changes from fine-grained to
granular.
Burnham (1997) reported application of DCP as a quality control device during the
backfill compaction of pavement edge drain trenches and granular base layer compaction in
MnDOT. A limiting DCPi value for each particular subgrade soil and base type was
proposed as incorporated in MnDOT specifications. A relationship between DCPi and the
required remedial thickness of granular backfill/lime modification was presented.
Chai and Roslie (1998) correlated the number of blows required by DCP to penetrate
30 cm to subgrade modulus back-calculated from FWD.
Parker et al. (1998) developed and automated a version of DCP where the instrument
was mounted on a trailer. The system was designed to lift the hammer, record data, and
extract the rod after penetration.
P a g e | 18
Coonse (1999) performed DCP and CBR tests on remolded residual clayey soils in
laboratory. He demonstrated that the CBR and DCP show the same strength response
(although numerically inverse) to change in moisture content while the compaction effort is
constant and to change in compaction effort while the moisture is constant around optimum.
By comparing the results of CBR and DCP tests in soaked and unsoaked samples, he showed
that soaked specimen loose strength and both tests identified that. It was also showed that the
change of moisture content can greatly change the strength of the cohesive soil. He also
verified the effect of the mold size on DCP and CBR test results. A new DCP-CBR
correlation for CH and CL materials was also derived.
Chen et al. (1999) showed a strong correlation between a 30-degree DCP cone results
and the back-calculated elastic modulus from FWD (Falling Weight Deflectometer) in
mostly clayey and silty soils in Kansas. They presented the correlation equation along with
95 percent confidence lines in a plot based on test results.
Siekmeier et al. (1999) compared the elastic modulus obtained from DCP, SSG (Soil
Stiffness Gauge), FWD, and Portable FWD. The tests were performed on granular base and
subgrade of several road projects and mixture of clayey and silty sand of an embankment in
Minnesota. They used Webster et al. (1992, 1994) correlation of DCPi to CBR and then
Powell et al. (1984) correlation of CBR to elastic modulus. The DCPi was the average of
DCP penetration rate of the upper 75-mm and 150-mm, excluding the first drop.
Nazarian et al. (2000) instrumented a DCP with a load cell and an accelerometer to
measure the transferred energy to the anvil. They showed a typical force and velocity-time
histories up to one millisecond and plotted them versus normalized distance (i.e., equivalent
to the time that stress waves travel from anvil to tip). The force and velocity were then used
to calculate transferred energy using FV (Force-Velocity) method. They calculated the
energy transferred to soil by comparing the energy needed to penetrate a bucket filled with
30 cm of very loosely packed foam and the energy needed to penetrate a soil sample and
related it to the soil resistance. They have presented a plot of energy loss as a function of
penetration rate, where the energy loss is 75 to 120 KN-m for 50 to 2 mm/blow penetration
rates, respectively. While the total delivered energy by the 8-kg DCP hammer is about 45 N-
m, it is not clear how the energy loss could be more than 1000 times. They have mentioned
that further analysis is needed for reliably extracting information from their measurement.
P a g e | 19
Livneh et al. (2000) showed that the DCP-CBR correlation suggested by Livneh and
Ishai (1987) was valid for light to heavy clayey soils and there was no need for a separate
correlation as suggested by Webster et al. (1994). The CBR test they performed was on
undisturbed samples extracted from pits. They also compared the DCP test results of 1, 1.3,
1.5, and 2 meter long DCP rods. They showed that the results were different and a correction,
if DCP with longer rods were used, was needed.
Livneh (2000) developed a method and calibrated it to correct the effect of skin
friction forces on the DCP rod during testing in cohesive materials. He incorporated the
measured torque moments at different depths in the CBR correlation equation. His reason
was that occasionally the CBR values derived from DCP tastings are higher than the CBR
values obtained by other in-situ tests. However, the CBR-DCPi correlation that he used
results higher CBRs for DCPi’s between 2 to 80 mm/blow compared to Webster et al.
(1992)’s equation. But he mentioned that “improper penetration”, which mobilizes the skin
friction may be the reason. To minimize the effect of skin friction he suggested attaching a
base frame to the DCP apparatus in order to ensure vertical penetration. The length extension
of the standard penetrating rod to a total of about 2.0 m (by screwing on an additional 1.0-m
rod after the original rod) is also used and completely inserted into the pavement or subgrade.
He mentioned that such a procedure found to be necessary when the total thickness of the
pavement layers, including the relevant subgrade layers, exceeds the length of a standard
penetrating rod
George and Uddin (2000) used manual and trailer-mounted automated DCPs in their
research to determine the subgrade resilient modulus of subgrade soils in the state of
Mississippi. Test results showed no differences between the manual and automated DCP
measurements. Subgrade moduli were determined by laboratory triaxial tests and also by
analyzing the deflection profiles obtained from the FWD. The DCPi was correlated with
laboratory and back-calculated resilient modulus using two different models for fine-grained
and coarse-grained soils. For further improvement of the models, soil physical properties
were found necessary.
Chen et al. (2001) conducted more than 60 DCP tests on two test pavements on US-
281 near Jacsboro, TX. DCP test were conducted in 3 different procedures through asphalt
concrete, a narrow borehole in asphalt concrete and directly on the base of highway by
removing a wide strip of pavement. They presented equations to convert DCP results from
one type of tests to the other. For comparison of these tests they plotted the cumulative
P a g e | 20
number of blows versus depth with some adjustments. Average DCPi is used for correlations
to CBR and then to elastic modulus using the correlation proposed by Powell et al. (1984).
The elastic modulus obtained from DCP is then compared with those obtained by FWD-
MDD (Falling Weight Deflectometer-Multi Depth Deflectometer) tests and by resilient
modulus laboratory tests. They concluded that the elastic modulus of the base and subgrade
layers determined by DCP and FWD-MDD tests are very close and laboratory determined
subgrade modulus were slightly higher than those.
Gabr et al. (2001) developed a correlation between DCPi and liquid index and
Saturation ratio based on laboratory tests on piedmont residual soils of Davidson County, NC
with more than 60% fines. They used these correlations to predict the dry unit weight and
water content of the soil.
Konrad and Lachance (2001) used a 51-mm diameter cone in dynamic penetration
test in base and subbase materials, due to concerns regarding the effect of grain size on
penetration resistance. They also correlated the penetration index to the back-calculated
elastic modulus from plate load test in unbound base and subbase materials.
There is a DCP Data Acquisition System (DCP-DAS) developed by Applied
Research Associates Inc., which became commercially available before 2002. It uses a string
potentiometer to automatically measure the depth of penetration and number of drops.
However, the effect of the pulling force applied by the string potentiometer on DCP
measurements is not clear. DCP data is recorded in an electronic format that can be easily
accessed by a computer for further analysis. Automatic data acquisition allows one-person
operation, reducing testing cost, while improving the accuracy of the data collected (Vertek,
2010).
Rahim and George (2002) ran DCP and automated DCP (which is mounted on a
trailer) tests atop of subgrade through drilled holes at 12 sites in Mississippi. Shelby tube
samples were obtained and tested to calculate Resilient Modulus (MR) following AASHTO
TP46 Protocol. DCPi and other soil properties were then correlated to resilient modulus by
two different equations for coarse-grained and fine-grained soils. They found that other soil
state variables are significant in MR prediction.
Karunaprema & Edirisinghe (2002) tested clayey and silty gravel soils mixed in the
lab in Sri Lanka. They reported that a clear relationship between the soaked CBR and the
P a g e | 21
DCP value could not be defined as for unsoaked CBR. However, they found a significant
correlation between the DCP and the difference between the unsoaked and soaked CBR
value. In addition to developing DCP-CBR correlations, they also correlated moisture
content and DCP to dry density, compaction ratio, and optimum moisture content..
Herrick and Jones (2002) developed a dynamic penetrometer with a 2-kg hammer for
measuring soil compaction in agricultural and rangelands. They used an adjustable hammer
drop height to have the flexibility, which allowed them to use a single instrument on a broad
range of soils without any loss in sensitivity.
ASTM introduced the DCP standard test method in 2003 for shallow pavement
application (ASTM D6951-03). Later in 2009, the standard was revised (ASTM
D6951/D6951M-09).
Amini (2003) reviewed the application of DCP in pavement design and construction.
He warned the use of DCP for materials with an aggregate size larger than 50 mm (2 in.).
Abu-Farsakh et al. (2004) evaluated the use of DCP for quality control/assurance
evaluation of pavement layers and embankments during construction. The DCP and plate
load tests (PLT) were carried out in the laboratory on silty clay and clayey silt. The DCP,
PLT, and FWD tests were conducted on subgrade and base sections of several projects in the
state of Louisiana. Laboratory CBR tests were also performed on samples from the lab and
field tests. Correlations between DCPi and elastic modulus (from plate tests), unsoaked CBR,
and resilient modulus (from FWD) were developed. The DCPi was the average penetration
in the top 30 cm of soil. They suggested 5.5 mm/blow as the acceptance limit for the crushed
limestone base. They concluded that the DCP device can be used for evaluation of uniformity
of compaction, thickness, and stiffness of pavement layers and subgrade.
Rahim et al. (2004), developed a model based on the pore collapse theory and
cylindrical cavity expansion to predict DCP penetration resistance based on cohesion, angle
of internal friction, and initial porosity. The parametric study of the model reveled that for
small initial porosity, the penetration resistance is strongly depends on internal angle of
friction but it is not as sensitive to cohesion. They compared their prediction with DCP test
data, obtained by Mississippi department of transportation. In general, the reported angle of
internal frictions versus DCPis were scattered. Although the developed model can predict the
general trend of the average data, but it did not predict the internal friction with a reasonable
P a g e | 22
confidence level. The correlation between cohesion intercept and DCPi was also poor. The
initial porosity showed a better correlation to DCPi.
Amini (2004) reviewed the previous researches in application of DCP in the quality
control of compaction in backfill, base layer, and backfill around utilities.
Abu-Farsakh et al. (2005) conducted a series of DCP, PLT (plate load tests), FWD,
and CBR laboratory tests. They presented correlations between DCPi and the Elastic Moduli
of soil (obtained from Plate Load Test), unsoaked CBR values, and resilient modulus
(obtained from FWD tests). They concluded that DCP can be used reliably to evaluate the
stiffness and strength of pavement materials, in addition to verify the uniformity of
compaction. They also recommended the criteria of 5.5 mm/blow for crushed limestone
highway base and concluded that it results satisfactory stiffness and compaction level.
Chen et al. (2005) developed a correlation between DCPi and elastic modulus back-
calculated from FWD tests. They compared their relationship with the elastic modulus
obtained by using the DCPi-CBR correlation of Webster et al. (1992) and CBR to elastic
modulus correlation proposed by Powell et al. (1984). The DCPi was corrected to take into
account the effect of overburden pressure in case of conducting the test through a drilled hole
in the asphalt layer.
Edil and Benson (2005) conducted several tests on the exposed subgrade and subbase
to the maximum depth of 38 cm across the state of Wisconsin. The DCPi was obtained from
the weighted average of penetration rates without excluding any datapoints for further
analysis and correlations. A linear relationship was observed between SSG stiffness in
regular scale and DCPi averaged from depth 0 to 152 mm in logarithmic scale. They also
showed that in plots of DCPi versus unit weight or water content, a general pattern can be
observed but datapoints were so dispersed that a unique correlation could not be developed.
For the quality control of earthworks they proposed a procedure consisting of normalizing
the DCPi by the deviation of compaction moisture content from the optimum moisture
content and showed that, for compacted natural earthen materials, a value around -8.4 is an
indication of high relative compaction.
Dai and Kremer (2006) summarized specifications and implementation of the DCP
testing in Minnesota and other states. They performed tests with DCP (equipped with DCP-
DAS) and other tests on several construction projects in the State of Minnesota. A modified
P a g e | 23
DCP specification for road construction projects was recommended and a testing procedure
was suggested. The use of DCPi as the criteria in place of relative densities for the granular
fills was also discussed.
Ampadu and Arthur (2006) developed a correlation between DCPi and the level of
compaction based on tests on compacted gravel in a road construction site in Ghana. They
concluded that this correlation depends on the material and the water content, and the
proposed equation is not unique.
Swenson et al. (2006) studied moisture effects on the measurements of several
laboratory and field devices and their interpreted modulus values. They reported a significant
scatter of resilient modulus from DCP. Overall, the results showed that both moisture and
density had a measurable effect on the modulus of fine-grained soils. However, they found
that, DCP is effective in quantifying the uniformity of compacted soil volumes, and
estimating the apparent Young’s modulus through empirical formulas.
Wu and Sargand (2007) showed that DCP is a viable device for evaluation of base
and subgrade during construction. They concluded that DCP can greatly improve the quality
monitoring of pavement unbound materials, and therefore, enhance its performance. An
automated DCP was used in their research, and they reported that it reduces the required time
to run one test to one-fifth. However, very small or negative penetration rates were observed
in some of the tests, which they related to non-homogenous nature of subgrade soil and
presence of small rocks. It was not investigated if the automatic DCP is also participating in
the errors. They conducted two corrections to adjust the profile graphs to enhance
interpretation. To identify uniform layers from penetration rate profiles, a method proposed
in AASHTO 1986 was used. The average DCPi from tests on the base of 10 road projects in
Ohio is reported between 5 to 14 mm/blow. The average DCPi in asphalt concrete was
reported between 2 to 7 mm/blow. They suggested accepting of DCP into pavement design
methods, since the validity of DCP to measure the soil strength was proved, and in any
correlation, estimation errors are unavoidable. A set of DCP acceptance criteria and standard
was proposed.
Booth et al. (2008) raised concerns about legitimacy of DCP to CBR correlations
after comparing laboratory CBR values with those obtained from correlation equations from
tests in sandy slightly gravelly silt and silty very gravelly sand.
P a g e | 24
Puppala (2008) reviewed DCPi to resilient modulus correlations. He reported that the
DCP has been used by different transportation agencies for years to estimate the moduli of
compacted subgrades and granular soils. However, he warned that the majority of the
correlations are site specific and empirical in nature and their use for other soils requires a
careful examination and engineering judgment.
Siekmeier et al. (2009) proposed the minimum required DCPi values to be used for
construction quality assurance based on tests conducted on granular and fine-grained soil
samples prepared in the lab for a range of moisture contents and densities. They obtained
graphs which by knowing the grading number1 and the moisture content in granular soils and
plastic limit and the moisture content in fine-grained soil the minimum required DCPi can be
estimated.
DCP Advantages The advantages of DCP test are summarized by Livneh (1987), Ayers et al. (1989),
Webster et al. (1994), MnDOT (1996), Karunaprema and Edirisinghe (2002), and Wu and
Sargand (2007) and listed as follows.
- It characterizes the in-situ strength of soil,
- It characterizes the strength with depth,
- It could be used to determine the thickness and depth of underlying soil layers,
- It could be used to verify uniformity of compaction,
- It is repeatable and reliable,
- It can be used in soils with a wide range of particle sizes and strengths,
- It is man-portable,
- It is relatively inexpensive,
- It is sturdy and the maintenance is simple and inexpensive,
- It could be used in developing countries for evaluation and design purposes,
- It is simple enough to be used by an inexperienced person,
- It could be used to verify whether if a stabilized soil has achieved its potential
stiffness,
- It requires less penetration depth than the CPT to measure the surface layer strength,
- It is relatively fast and usually does not take more than 10 min (time varies depending
on the strength of the material and maximum depth of penetration). 1 The grading number is the sum of the percentages of particles passing specific sieves.
P a g e | 25
DCP Disadvantages
While the operation of DCP is physically arduous (Amini, 2003), the manual raise
and drop of the hammer could be a source of error in a DCP test. The user has to ensure that
the hammer is touching the bottom of the handle but not lifting the cone before it is allowed
to drop (Webster et al., 1992). The operator should be careful not to exert any downward or
upward force on the handle and not to influence the free fall of the hammer by hand
movement (Webster et al., 1992).
Utilization of skin friction can cause erroneous results in cohesive materials. After
comparing the Electric Cone Penetrometer results with DCP and in-situ CBR, Buncher and
Christiansen (1991) mentioned that the DCP is very susceptible to skin friction in cohesive
soils. For the same reason, Webster et al. (1992) suggested to limit the depth of penetration to
12 in. (about 30 cm) in highly plastic clays and clean and lubricate the rod after each test to
minimize sticking of clay to the rod. However, Webster et al. (1994) stated that oiling the rod
does not improve test results in CH soils significantly. Livneh et al. (1995) showed that the
inclined (15 degrees tilted) penetration in clayey materials yield 28% lower DCPi on average.
They reported that angled penetration is not unusual in deep DCP penetrations. The problem
of keeping the penetrometer perfectly vertical in manual operation of the DCP is also
mentioned by others (e.g., Wu and Sargand, 2007). Livneh et al. (1995) suggested using a
vertical supporting frame to overcome this problem. Livneh (2000) reported that occasionally
DCP predict higher CBR values due to inclined probe penetration, which utilizes the skin
friction forces. This increase in the resistance increases the DCPi value which will result in
higher CBR values consequently. To correct this effect, he incorporated the measured
torsional moments (i.e., the moment required to rotate the DCP at the place) at different
depth in the CBR correlation equation.
The other problem with DCP is the extraction of the instrument after deep tests in
some cases (Weintraub, 1993, Wu and Sargand, 2007). Using disposable cone tips as
suggested by Webster et al. (1992) may be one solution. However, ASTM 6951 suggests
using an extraction jack, if disposable cone tips are not used.
Ayers et al, (1989) has shown that the maximum aggregate size has a notable effect
on the test results. They found the maximum aggregate size of around 38 mm to be where the
DCP is no longer a viable test. However, Webster et al. (1992) reported that DCP is not
suitable for soils having significant amount of aggregates greater than 51 mm. Konrad and
P a g e | 26
Lachance, (2001) reported that use of DCP in pavements with aggregates larger than 20 mm
was questionable.
For tests in loose materials, the 8-kg hammer causes excess penetration (Webster et
al. 1992). In this case, no DCP measurement is possible close to surface in dry sand or gravel
(Webster et al., 1992) due to no confining pressure in the procedure. Using a different
hammer mass is suggested by ASTM 6951 after Webster et al. (1992). However, far less
experimental data for using a 4.6-kg hammer is reported in the literatures.
The manual reading and recording the number of blows and depth of the cone could
also cause some errors (Webster et al., 1992). Since the regular DCP test needs one operator
to lift and drop the hammer while keeping the instrument vertical, another operator should
keep track of the penetration after each blow. To address this problem, some innovation kits
are added to DCP. For instance, Kessler Inc. (Kessler Soils Engineering Products, Inc., 2007)
introduced a kit to write the number of blows for each set of blows on a removable tape along
the ruler, or use a magnetic ruler data collection device. Applied Research Associates Inc.
developed a DCP Data Acquisition System (DCP-DAS), which uses a string potentiometer to
automatically measure the depth of penetration and number of drops (Vertek, 2010).
Otherwise, data collection and analysis is time consuming, if no automated measurement
system and software are used.
MnDOT (1996) reported that for in situ foundation testing in addition to DCP tests,
soil samples should be obtained on a periodic basis to be used for soil classification and
moisture content determination. They mentioned that sand cone or nuclear density tests may
be taken for further investigation and compaction correlation.
2-3- CPT
The cone penetration testing (CPT) consists of pushing a cone penetrometer into the
soil and continuous measurements with depth. The test is also called static or quasi-static
penetration test. Cone penetrometers are available in different diameters and types. The
earlier version, known as mechanical cone penetrometer (originally called Dutch cone)
consisted of cylindrical rods with a conical point used to measure the end-bearing resistance
of the cone. With addition of a telescoping mechanical friction sleeve, measuring local soil
friction became possible in mechanical friction-cone penetrometer (also known as Begemann
P a g e | 27
friction-cone). In these instruments, the force is measured with a hydraulic or electric load
cell or proving ring. The measurements are in less than 20 cm increments. The corresponding
standard test method is ASTM D3441-05.
The more commonly used version is electric (or electronic) cone penetrometer. In
electric cone penetrometer, the tip is instrumented with transducers to measure the cone
resistance and side resistance continuously with depth. The configuration of the tip may vary
depending on design, but the friction sleeve is located right behind the cone, and its outside
diameter is always equal to the diameter of the base of the cone. The pushing system can
consist of a standard drill rig or a system mounted on a truck, track, trailer, all-terrain vehicle,
skid arrangement, or portable unit. The penetrometer is pushed into the ground at a rate of
200 mm/sec and measurements are collected in at least 50 mm intervals. Although electric
cone penetrometers provide finer resolution compared to mechanical cone penetrometers,
they are more likely to be damaged in hard materials (Mayne, 2007). ASTM D5778-07 is the
standard test method for electric cone penetrometers.
There are other types of penetrometers. Some are equipped to provide or measure
shear wave velocities, electric conductivities, acoustic emissions, temperature, images of soil,
and water samples. Some of the versions, which are more used in practice, are listed in Table
2-1. Lunne et al. (1997) and Mayne (2007) reviewed and discussed the history of the tests,
equipments and procedures in details. In this dissertation, the term CPT is interchangeably
used to refer to the electric cone penetrometer or the test in general regardless of the
penetrometer type.
The applications of CPT include site investigations for exploring soils for support of
embankments, retaining walls, pavement subgrades, and bridge foundations (Mayne, 2007).
The results are used to classify the materials in a soil profile and to obtain estimates of soil
properties (Newcomb and Birgisson, 1999). Various correlations are available between CPT
measurements and undrained shear strength, internal friction angle, the overconsolidation
ratio, soil behavior classification, mean-grain size, relative density, and SPT number (Lunne
et al., 1997). CPT is also a valuable test for quality control during ground modification,
because it allows a direct comparison of before and after modification (Mayne, 2007).
P a g e | 28
Table 2-1: Different Types of Cone Penetration Tests in Site Characterization (After Mayne, 2007).
Table 4-4 presents the compaction effort used to prepare each samples. Figure 4-3
shows the compaction curves of soil materials obtained following ASTM D698-00. The
compaction curves for sample C6S, CSB, and CSR are not obtained. Compaction properties
of bentonite and bentonite-sand materials are available in literatures (e.g., Ito, 2006; Tay et
al., 2001; Akgun et al., 2006; Filippo et al. 1998; and Ito and Komine, 2008). Based on
similar cases, the optimum moisture contents for CSB and CSR are estimated to be 10 to 20
percent and 25 to 35 percent for C6S. However for C6S, the variations in dry density are very
small for different water-content values. It should be mentioned that based on the moisture
content of samples presented in the next section, the CSB and CSR are prepared on the wet
side, and C6S is prepared on the dry side of compaction curve. As mentioned earlier, sticking
of the clay to the hammer and jamming of the hammer made it impossible to measure the
compaction effort for C6S, CSB, and CSR accurately.
Table 4-4: Compaction Effort Used to Prepare Samples.
Sample IDCompaction Effort
(kN-m/m3)
BSC ~600
BSL Pluviated
C6S Not Measured
CSB Not Measured
CSR Not Measured
GRV 600
RSC 300
RSD 600
RSL Pluviated
SIL 600
SIS 300
MS0 600
MS1 600
MS2 600
MS3 600
MS4 600
MS5 600
MS6 600
P a g e | 78
A) B)
C) D)
Figure 4-3: Compaction Curves of Soil Materials for Samples A) SIL and SIS, B) RSD, RSC, and RSL,
C) MS0, MS1, MS2, MS3, MS4, MS5, MS6, and GRV, and D) BSC and BSL.
Geotechnical Properties of Soil Samples
Table 4-5 presents a summary of the geotechnical properties of the prepared soil
samples. Density, void ratio, porosity and saturation ratio are calculated using phase relation
formulas. Specific Gravity is measured based on ASTM D854-05. For soil samples which
1.60
1.80
2.00
2.20
2.40
0% 5% 10% 15% 20%
Dry
Den
sity
(M
g/m
3 )
Water Content (%)
100% degree of Saturation
Standard Effort
1.60
1.80
2.00
2.20
2.40
0% 5% 10% 15% 20%
Dry
Den
sity
(M
g/m
3 )
Water Content (%)
100% degree of Saturation
Standard Effort
1.60
1.80
2.00
2.20
2.40
0% 5% 10% 15% 20%
Dry
Den
sity
(M
g/m
3 )
Water Content (%)
100% degree of Saturation
Standard Effort
1.60
1.80
2.00
2.20
2.40
0% 5% 10% 15% 20%
Dry
Den
sity
(M
g/m
3 )
Water Content (%)
100% degree of Saturation
Standard Effort
P a g e | 79
were prepared by mixing other materials, the specific gravity of the original materials is used
to calculate the specific gravity of mixture by weighted averaging. Measured specific
gravities include: Gravel = 3.036, Fine sand = 2.671, Silt = 2.779, Coarse Sand = 2.706, and
Bentonite = 2.901. For wet soil samples, moisture content and moisture profile were
measured following ASTM D2216-98.
The only soil with non-uniform moisture profile was GRV. Although each layer of
the sample was prepared by mixing the same amounts of water and soil, the moisture profile
was not uniform, and the sample showed increasing moisture content with depth. The sample
was also allowed to drain after the first test. The moisture profile obtained by sampling
immediately after the final test is presented in Figure 4-4.
Table 4-5: Soil Samples Phase Properties.
Sample
ID
Specific
Gravity
Density (kg/m3) Water
Content (%)
Void
Ratio
Porosity
(%)
Saturation
Ratio (%) Total Dry
BSC 2.67 1582 1542 2.6 0.73 42.2 9.4
BSL 2.671 1194 1159 3.0 1.30 56.5 6.2
C6S 2.901 1916 1585 20.9 0.77 43.5 76.3
CSB 2.76 ~1840 ~1334 ~38 ~1.07 ~51.6 ~98.2
CSR 2.76 1840 1334 38 1.07 51.6 98.2
GRV* 2.94 2380
(2318)
2232 6.6
(3.8)
0.31 23.9 61.7
(36.0)
RSC 2.71 1773 1773 0 0.52 34.4 0
RSD 2.71 1787 1787 0 0.51 33.8 0
RSL 2.71 1656 1656 0 0.63 38.7 0
SIL 2.78 2245 2074 8.2 0.34 25.2 67.6
SIS 2.78 2092 1937 8 0.43 30.2 51.3
MS0 2.94 2427 2427 0 0.21 17.2 0
MS1 2.94 2401 2377 1 0.23 18.9 12.6
MS2 2.94 2480 2431 2 0.21 17.1 28.8
MS3 2.94 2475 2403 3 0.22 18.0 39.5
MS4 2.94 2550 2455 3.9 0.19 16.3 58.2
MS5 2.94 2592 2467 5.1 0.19 15.9 79.2
MS6 2.94 2611 2445 6.8 0.20 16.6 99.6
* The sample drained for different tests. Numbers in parentheses represents those values after 48 hr of draining. Moisture related parameters are averaged.
P a g e | 80
Figure 4-4: Moisture Content Profile of GRV after Draining.
4-3- Test Setup
RapSochs and DCP penetration tests are conducted up to a depth between 305 mm
and 610 mm (12 to 24 inches), with most tests conducted up to 460 mm (18 inches). The total
possible penetration of the instrument is 910 mm (36 inches). A guide fixed on a frame was
used for all tests to support the RapSochs and DCP to keep it perpendicular to the soil surface
during testing. Figure 4-5 shows the assembled RapSochs set up ready for a test. The samples
were large enough so that multiple non-interfering penetrations are made for repeatability
and spatial averaging. The minimum spacing between penetrations was 5.9 times of cone
diameters. The effects of the boundary on test results are verified in next chapters.
-700
-600
-500
-400
-300
-200
-100
0
0% 2% 4% 6% 8% 10%
Dep
th (m
m)
Moisture Content
P a g e | 81
Figure 4-5: Running a test with RapSochs at SoilBED facilities of NU.
4-4- RapSoChs Test Procedure
The RapSochs tests are conducted following the instructions provided in Appendix B.
This is the latest revision of “RapSochs Rev. 0 Penetrometer Operating Instructions Revision
2 Phase 1” that was revised on 08/20/2007. In the instruction, the test sequence, operation
procedure, instrument assembly, inspection and maintenance are explained. The instruction is
revised three times to include changes in the software and/or instrument itself due to
improvement of the system. Summary of the instruction and explanation of other procedures
that are not included in the instruction is presented in this section. A one-page checklist
developed for testing at NU is also attached in Appendix E.
General Test Sequence 1. The system is assembled, checked, and initialized as described in the Operating
Instructions.
2. The RapSochs is placed on a 125 mm by 125 mm by 6 mm rigid metal plate on the
surface of the soil sample. The moisture sensor is activated to capture the initial
P a g e | 82
position of the rod. Then, the plate is removed, and the RapSochs is allowed to
penetrate by its own weight.
3. The operator raises the hammer to a predetermined height. The hammer is released
and dropped under its own weight. The drop height is determined based on the
procedure described in next subsection.
4. For each hammer blow, a dynamic data sequence is triggered by the impact of the
hammer and data are recorded.
5. After completion of the dynamic sequence, a moisture sensor reading is taken by the
operator after the penetrometer has stopped moving.
6. Steps 3 to 5 are repeated until the desired penetration depth is achieved. A MATLAB
script is used to check the penetration depth at any time to show the progress.
7. After the test reaches the desired depth, the instrument is disconnected from the
laptop and the main processor. Then, all connections are opened and removed,
including the string potentiometer. In soils with high adhesion, a system of pulleys
and cables are used to extract the penetrometer from the soil.
Hammer Drop Height Procedure
In general, the hammer drop height was adjusted to produce a penetration between 13 and 25
mm (0.5 and 1 inch) per blow. The procedure for determination of drop height was modified
once during a test series. In the first procedure followed for tests in GRV, BSC, RSC, RSD,
RSL, and the first test in SIL (for tests before 8/15/2007), the drop height started from 51 mm
(2 inches), and for any penetration less than 13 mm (0.5 inch), the drop height was increased
by 25 mm (1 inch). For tests conducted later, the drop height started from 127 mm (5 inches),
and any time that the penetration was less than 13 mm (0.5 inch) the drop height was
increased to 254, 381, 508 and 559 mm respectively (10, 15, 20 and 22 inches). In addition,
in order to assess the effect of variable hammer drop height, after the drop height was
established in 559 mm (22 inches), the drop height was decreased to 254 and 381 mm (10
and 15 inches) for few blows, and then back to 559 mm (22 inches) for the remaining blows.
The manual release of the hammer from a certain height is not very accurate. Experiences
showed that the precision is limited to ± 13 mm (0.5 in.).
P a g e | 83
Data File Format
A data block, 204.8 milliseconds long acquired at 10K samples/sec., is collected. The
total block is then 2048 samples long with 100 of the samples taken prior to the trigger event.
The data for each blow, including tip strain, sleeve strain, acceleration, string potentiometer
displacement, and tip temperature are stored in an ASCII file under the filename:
<JOB_ID>_tip_nn.dat
where nn is the file index which starts with 1 and is automatically incremented after each
blow. The converted data into engineering units are stored in:
<JOB_ID>_tip_nn_converted.dat
In a converted data file, data is saved in columns separated by at least one space in
signed ASCII format. The file contains five columns of data as follows.
NOTE: First three letters are sample name. Second number shows the order of the test in the sample. R represents RapSochs tests. D represents DCP tests. The last number is the spot number.
5-2- Test Measurements
The RapSochs and DCP testing procedures were described in Chapter 4. This section
summarizes the original measurements, recorded data and parameters. All tests were
conducted at the SoilBed facilities. The laboratory temperature was continuously monitored
and reported values were between 20 to 32 °C (68 to 90 °F).
RapSochs
Recorded data for each test includes: Sample-ID, location in the sample (Spot Number),
order of the test in the sample, hammer drop height for each blow and data-files (including
<JOB_ID>_moisture_nn.dat, <JOB_ID>_tip_nn.dat, and <JOB_ID>_tip_nn_converted.dat
as explained in previous chapter).
Sample-ID, spot number, and the order of the test define the Test-ID. Hammer drop height of
all the RapSochs tests are presented in Appendix F. Although the original measurements are
in inches, they are converted to equivalent units in SI systems for analysis. In the “Data File
P a g e | 89
Format” subsection of “RapSochs Test Procedure” of Chapter 4, the file format of raw data
recorded in a RapSochs test was explained. The raw data for all RapSochs tests are presented
on a CD-ROM along with this document where test results are organized by sample and test
IDs.
DCP
The data recorded during a DCP test includes the drop number and corresponding cumulative
penetration. The initial sinking of DCP due to its weight before starting a test was also
recorded. The distance between the bottom of the anvil and soil surface is measured after any
drop using a vertical scale. Subtracting the tip-to-anvil distance results in the depth of the
cone after each drop. Depth of the cone is the depth of the top of the shoulder of the cone
shown as the “reference point” in Figure 2-1. DCP measurements for all the DCP tests are
presented in Appendix G.
ASTM Standard D 6951-03 suggests to record DCP test data in a form where test data
includes “Number of Blows” between test readings and “Cumulative Penetration” after each
set of hammer blows. In all tests conducted, “Number of Blows” between test readings is
one. It means that DCP data are recorded for any individual blow. Other required information
according to ASTM D 6951-03 can be obtained from provided data in Appendix G.
5-3- Database
A database in MAT-files format was developed to facilitate data access, processing, and
mining. The format for the database was designed to be consistent with MATLAB®, the
programming language used for data analysis. MAT-files provide a flexible system that can
facilitate data entering, editing, mining, checking, processing, and analysis in MATLAB®.
MAT-files are binary files with double-precision and may contain different types of data,
including strings, matrices, multidimensional arrays, structures, and cell arrays. They are
saved MATLAB® workspace in files with “.mat” extension. This format makes it very easy
to enter, revise, and save data in MATLAB environment and can also be read and written
from external C or Fortran programs.
P a g e | 90
Database Structure
Each test data are stored in a separate MAT-file and are named based on Test-ID. There are
two different types of MAT-files; RapSochs MAT-files and DCP MAT-files. The variables
stored in each type are different since the test and measured parameters are different. Each
type can be recognized based on the file name which corresponds to the Test-ID.
To establish the database, all data-files of each test are stored together in a single MAT-file.
A MATLAB code developed for this task. All the codes which are developed for database
establishment and later in data analysis are listed in Appendix C and are available in the
folder “\m-files” on the CD-ROM along this document.
The resulting MAT-files of all DCP and RapSochs tests are stored on the CD-ROM in the
folder “\Database”. These MAT-files include raw data as well as the results of data analysis.
All new variables that are added to the database are explained in the related section. Table
5-1 also presents the list of MAT-files available at “\Database”.
DCP MAT-Files
In a DCP test, the distance of the bottom of the anvil from the soil surface is measured. The
distance from the bottom of the anvil to the top of the cone shoulder is 98.0 cm in the
Kessler’s DCP. Other parameters, including DCP-PR are obtained from this data. List of
variables available in a DCP MAT-file containing the experiment’s original measurements is
presented in Table 5-2. DCP test data are presented and discussed in “DCP data” section and
the necessary information is extracted from the original measurements.
Table 5-2: The Variables in a DCP MAT-file Containing Original Data.
Variable Description
DCP The distance between bottom of anvil and soil surface - The original
recorded data (cm)
Test_ID A character variable which contains Test-ID in “AAA-nn-D-S#” format.
P a g e | 91
RapSochs MAT-Files
Table 5-3 provides description of the variables that include original data and measurements
for any RapSochs test. The raw measured signals of string potentiometer, acceleration and
two strain gauges are added together in an n-by-1 array. Later in data analysis, other
variables are added to this database. RapSochs test data are presented and discussed in
“RapSochs data” section.
Table 5-3: List of Variables Containing Original Data in a RapSochs MAT-file.
Variable Description
Job Test information used in the lab for test and sample identification.
Later Test-ID is obtained from this information.
Test_ID A character variable shows Test-ID in “AAA-nn-R-S#” format.
moisture_Depth Position of the cone below the soil surface. The zero level is at the
sample surface and values below the surface are negative (mm).
moisture_Frequency Moisture Sensor’s frequency of operation (Hz). The data in each raw
belong to moisture sensor reading at the corresponding
moisture_Depth.
moisture_Magnitude Moisture Sensor’s measured magnitude (dB). The data in each raw
belong to moisture sensor reading at the corresponding
moisture_Depth.
moisture_Phase Moisture Sensor’s measured phase (degree). The data in each raw
belong to moisture sensor reading at the corresponding
moisture_Depth.
Spfm Output of string potentiometer, filtered with order 5 median filter
(which does not add any time delay), added together for all the blows
of a test (mm).
Accel Output of accelerometer, added together for all blows of a test (g)
str1 Output of the tip strain gauge, added together for all the blows of a test
(kN)
P a g e | 92
str2 Output of the strain gauge in the sleeve, added together for all the
blows of a test (kN)
drop_heights Hammer drop height (in.)
Data Examination, Verification, and Correction
The DCP procedure and data recording is simple, and there was no concern about the validity
of the measurements. However, in RapSochs revision 0.0, measurements are complex and
depend on operator’s accuracy and instrument’s functionality.
In the database development, several issues including data accuracy, different testing
procedures and problems, which occurred during testing needed to be addressed. For
instance, some blows did not exactly follow the procedure due to operational problems (e.g.,
hammer fall was not a free fall or the hammer was dropped from an unknown height) or
instrument mechanical failure (e.g., string potentiometer holder broke during a test). In
addition, due to hardware problems (e.g., electronic miscommunication) the data length for
some drops was not 2048 data points long as it should be. Other issues were rooted in data
analysis codes and individual mistakes. There were other defects that occurred during
generation of MAT-files. For example the length of variables in MAT files (str1, str2, …)
did not match with the sum of individual data files.
To address these issues, a comprehensive data validation procedure is conducted. Two new
variables are added to the database as a result of data validation. The drop_penetration_
reliability is a binary variable that may have a value of 0 or 1. If in a specific blow due to any
reason, the data is not reliable then this variable is 0 otherwise it is 1. The drop_data_length
includes the number of recorded data points of the string potentiometer, accelerometer, and
strain gauges for any individual blow. When a communication error happens, the value for
the corresponding blow is less than 2048. In addition, in cases that the instrument fails to
trigger, the variable has a value of zero. The RapSochs system is designed in a way that data
of all sensors are saved in <JOB_ID>_tip_nn.dat and <JOB_ID>_tip_nn_converted.dat
simultaneously. Therefore, the drop_data_length presents the number of data points of a
specific blow in spfm, accel, str1, and str2 variables. For any further analysis these variables
are checked first. It should be mentioned that in cases that drop_data_length is less than 2048
but not zero, no biased or unusual data in sensor measurements were observed. In most cases,
P a g e | 93
it is believed that data are missed from the beginning or end of the recorded data. Table 5-4
lists the variables added to RapSochs MAT-files following the comprehensive data
validation.
Table 5-4: Variables Added to RapSochs MAT-files Containing Data Examination Information and
Reliability of Each Blow.
Variable Description
drop_data_length Number of data points for any individual blow. It is supposed to be
2048 exactly but for some blows it may be less than 2048 due to
electronical errors or miscommunication. Data are arranged in one
column. The nth value corresponds to blow number n.
drop_penetration_reliability Reliability of drop data (1= reliable, 0 = not reliable), for any
reason, such as hammer drop from an unknown height, system’s
crash, a problem in triggering, or etc. data of a blow may not be
reliable.
5-4- DCP Test Results
There is no consistent terminology in reporting DCP test results in the literature, and this can
cause some confusion for users. In this work, DCP-PR (Dynamic Cone Penetrometer –
Penetration Rate) is used for the penetration of the instrument due to one blow and DCPi
(Dynamic Cone Penetrometer Index) is used as the average calculated index of DCP-PRs.
These two terms are not used distinctively in most of the literature. DCPI or DCPi (Dynamic
Cone Penetrometer Index), penetration per blow, penetration ratio, penetration rate,
penetration index, DCP-PR, DCP index, DN (DCP Number), and DPI are other terms used in
literature for DCP-PR or DCPi.
Penetration Rate and Depth of DCP
The DCP test measurements are saved in “DCP” variable in DCP MAT-files. The other
calculated information is the depth of the cone after each blow and penetration rate or DCP-
PR. The added variables, including calculated parameters, are listed in Table 5-5.
P a g e | 94
Table 5-5: Variables Added into the DCP MAT-files Regarding Penetration Rate and Depth.
Variable Description
dcp_sinkage_depth The depth of the top of the cone shoulder below the surface penetrated
into the soil under the instrument’s own weight (mm) before starting
the test. This distance is positive if the cone is below the surface.
dcp_drop_depth The depth of the top of the DCP cone shoulder after each blow. The
zero level is at the soil sample surface and values below the surface are
negative (mm). Data are arranged in one column. The nth value
corresponds to blow number n.
dcp_drop_penetration The penetration per blow (mm/blow). This is basically DCP-PR. Data
are arranged in one column. The nth value is corresponding to the
penetration caused by blow number n.
Presenting DCP Data
There is no unique method in the literature to plot DCP test results. Sometimes the
penetration per blow is presented in regular or logarithmic scale versus depth. The
cumulative number of hammer blows has also been plotted versus depth.
ASTM D 6951-03 suggests plotting penetration per blow versus the scale reading or total
depth. However, to define any layering in the soil profile, it is suggested to plot the
cumulative number of blows versus depth to examine changes in the slope. To define the
layer thickness the intersection of the lines representing the average slope of adjacent layers
should be found.
De Henau (1982) and Webster et al. (1994) presented DCP-PR as a broken vertical line from
each set of penetration on a logarithmic scale versus depth. Chua (1988), Siekmeier et al.
(1998), Gabr et al. (1999), and Siekmeier et al. (2009) plotted DCP-PR as data points versus
depth. Van Vuuren (1969) suggested plotting values versus the mid-point (average) depth.
Burnham and Johnson (1993) plotted the calculated CBR versus average depth. Edil and
Benson (2005) used two different methods for data presentation. The first method is similar
to Webster et al (1994) and the second method is to plot the penetration per blow as data
points versus the depth after the penetration. Abu-Farsakh et al. (2005) plotted DCP-PR
P a g e | 95
versus depth as data points and the cumulative number of blows versus depth for determining
layer thickness. Kleyn (1975), Kleyn et al. (1982), Livneh and Ishai (1987), Chen et al.
(2001), Rahim and George (2002), and Ampadu and Arthur (2006) plotted the cumulative
number of blows versus depth.
Figure 5-2 shows different data presentation methods for an example data presented in the
Table 5-6. In Figure 5-2-A the cumulative number of blows is plotted versus depth. In Figure
5-2-B, DCP-PR is plotted as vertical broken line versus depth as it is constant for the
increment covered by each penetration. In this method, the entire depth of penetration is
considered to have the constant strength associated with the soil. Figure 5-2-C is the same as
Figure 5-2-B but in logarithmic scale. Figure 5-2-D shows DCP-PR as data points versus the
depth after penetration. The red line in these figures is from Figure 5-2-B plotted for
comparison only. Figure 5-2-E is DCP-PR versus the mid-point (average) depth before and
after penetration as data points. Figure 5-2-F is similar to Figure 5-2-E but without any line
connecting data points.
P a g e | 96
Table 5-6: Example DCP Data Used to Illustrate Different DCP Data Presentation Methods.
NOTE: * Blow number 1, 2, and 9 are excluded. † No data point is excluded. ‡ Blow number 25 is excluded.
The DCPi for each DCP test is obtained by fitting a line to the linear part of cumulative blow
numbers versus depth and excluding transient zones at the top or bottom of the graphs. The
slope of this line is reported as DCPi. In this project, since the soil samples are prepared in
the lab, they are expected to be very uniform. Except for the confinement effects close to the
surface and boundary effects at the bottom of the container, no other transient zones are
expected.
A code is developed to plot number of blows versus depth from the DCP data in the database
and to fit the best line to selected data points. The presentation method used is similar to
Figure 5-2-E. As an example, Figure 5-4 shows the resultant figure and calculated DCPi for
the test SIL-06-D-S9.
P a g e | 101
Figure 5-4: Number of Blows versus Depth in SIL-06-D-S9 DCP Test and the Best Fitted Line where the
Slope is DCPi.
The graphs showing the best line fitted to data of all DCP tests results are also presented in
Appendix H and the calculated DCPi are summarized in Table 5-8 also presents DCPi values
obtained by averaging the DCP-PRs in parentheses. In Table 5-8 the average DCPi of tests
performed in a similar condition is also calculated. The DCPi in GRV and SIL are not
averaged for all the test results. For GRV, the GRV-01-D-S8 test was performed before
draining the sample and GRV-03-D-S4 was performed after allowing the sample drain for 24
hours. In addition, data points at a depth lower than 600 mm are excluded when calculating
DCPi because the tip was very close to the bottom of the container and they are affected by
container’s boundary. For SIL, the SIL-02-D-S3 test was performed 3 months before other
tests in SIL sample and the effect of aging on cementation could be detected in this case.
0 20 40 60 80
-700
-600
-500
-400
-300
-200
-100
0
No. of Blows
Dep
th (m
m)
SIL-06-D-S9
DCPi = 6.8 (mm/blow)
DataFitted Line
P a g e | 102
Table 5-8: DCPi Obtained by Line Fitting to Number of Blows versus Depth and Averaging.
Sample-ID Test-ID DCPi
(mm/blow)
Average DCPi
(mm/blow)
BSC
BSC-04-D-S3 131.5 (131.5)
129.7 (129.7) BSC-05-D-S6 110.0 (110)
BSC-07-D-S7 166.0 (166.0)
BSC-08-D-S4 111.5 (111.5)
C6S C6S-02-D-S6 21.1 (21.1)
21.3 (21.2) C6S-03-D-S4 21.5 (21.3)
CSB CSB-02-D-S2 72.5 (75) 72.5 (75)
GRV GRV-01-D-S8 11.8 (11.9) 11.8 (11.9)
GRV-03-D-S4 8.0 (7.9) 8.0 (7.9)
RSC
RSC-04-D-S7 44.5 (44.5)
44.1 (51.2) RSC-05-D-S8 42.0 (52.0)
RSC-06-D-S4 47.0 (55.0)
RSC-07-D-S6 43.0 (53.5)
RSD
RSD-01-D-S1 42.5 (48.3)
42.1 (46.2) RSD-07-D-S2 42.0 (47.7)
RSD-08-D-S3 41.5 (41.5)
RSD-09-D-S4 42.5 (47.3)
SIL
SIL-02-D-S3 9.0 (9.1) 9.0 (9.1)
SIL-06-D-S9 6.8 (6.8)
6.6 (6.7) SIL-07-D-S4 6.6 (6.6)
SIL-08-D-S2 6.5 (6.6)
SIS
SIS-03-D-S7 13.8 (14.5)
13.8 (14.6) SIS-04-D-S8 12.8 (13.2)
SIS-06-D-S2 13.8 (14.5)
SIS-08-D-S3 14.6 (16.3)
NOTE: Numbers in parentheses are DCPis obtained by averaging the DCP-PRs. GRV-01-D-S8 was performed before draining the sample and GRV-03-D-S4 performed
after draining the sample. SIL-02-D-S3 is performed 3 months before other tests in SIL sample. The effect of
cementation is obvious in this case.
P a g e | 103
5-5- RapSochs Test Results
RapSochs is a newly developed instrument without any specific data presentation strategies.
This section describes how basic parameters (e.g., depth of the RapSochs) are obtained from
the original data together with a procedure to access and plot RapSochs original data.
Examples of plots are also provided in each subsection.
Penetration Rate and Depth of RapSochs
The initial position of the rod, captured by the moisture sensor, is used to measure the cone
sinking under the RapSochs’s own weight. Similar to DCP, a term RS-PR (RapSochs
Penetration Rate) is introduced for the penetration per blow in the RapSochs tests. To extract
the RS-PR for an individual blow, the averages of the first and last 50 data points of the
string potentiometer signal are subtracted. In cases that the instrument did not trigger during
a blow, the penetration is calculated based on the previous and the following blow. The depth
of the RapSochs cone after any blow is also calculated by adding the sinkage depth to the
calculated depth from string potentiometer signals. The string potentiometer itself did not
exhibit drift or hysteresis as the strain gauges and the accelerometer. However, summation of
RS-PRs may not result the same depth due to the numerical errors. The new variables added
to RapSochs MAT-files are listed in Table 5-9.
Table 5-9: Variables Added to RapSochs MAT-files Regarding Penetration Rate and Depth.
Variable Description
sinkage_depth The distance of the RapSochs cone below the surface, penetrated into the
soil under the instrument’s own weight (mm). This distance is positive if the
cone is below the surface.
drop_depth Depth of the RapSochs cone shoulder after the penetration (mm). The zero
level is at the sample surface and values below the surface are negative.
Data are arranged in one column. The nth value corresponding to the depth
of the RapSochs after blow number n.
drop_penetration Penetration per blow (mm/blow). This is basically RS-PR. Data are
arranged in one column. The nth value corresponding to the penetration rate
due to the blow number n.
P a g e | 104
The sinkage depth data is modified during configuration of the database to address different
test procedures. For example in granular soil samples a metal plate was placed on the soil
surface to obtain the zero depth reference. The sinkage depth could be negative, which occurs
when the shoulder of the cone is resting above the soil surface. For example in tests on SIL
sample, the sinkage depth is negative because the tip penetrated just for few millimeters
when it was put on the soil and the shoulder was still above the soil surface. Different codes
and software revisions resulted in a shift or different sign convention for string potentiometer
data which are corrected in the presented database.
Displacement, Acceleration, and Force Data
As indicated, the measured signals of string potentiometer, accelerometer and two strain
gauges are added together in an n-by-1 array. MATLAB codes are developed to access data
of any specific blow for a specified test. These codes produce a figure that include the test
and blow information and plots of displacement (measured by string potentiometer),
acceleration (measured by accelerometer), tip + sleeve force (measured by upper strain
gauge), and tip force (measured by lower strain gauge) for one drop. The test information
includes sample name, depth after the penetration, penetration per blow and hammer drop
height.
Figure 5-5 presents typical results of RapSochs tests. Figure 5-5 shows data of the 22nd drop
in RSC-02-R-S2, the 99th drop in SIL-03-R-S5, the 35th drop in C6S-01-R-S8, and the 65th
drop in GRV-02-R-S5 tests, respectively. The header for each plot consists of Test-ID, depth
after the penetration, penetration per blow, and hammer drop height.
P a g e | 105
Figure 5-5: Typical Signal Output of RapSochs Sensors for a Blow.
Where the parameter θ is (µ, σ), and if it is desired to estimate the mean, that is, τ(θ)=µ, then = (1/ ) ∑ is a possible point estimator of τ(θ) = µ. When we speak of estimating θ,
we are speaking of estimating the fixed yet unknown value that θ has. That is, we assume
that the random sample X1, X2, …, Xn came from the density f(x ; θ), where θ is unknown but
fixed.
There are two problems with point estimation: the first, to devise some means of obtaining a
statistic to use as an estimator; the second, to select criteria and techniques to define and find
a “best” estimator among many possible estimators.
One of the methods of finding the point estimator and probably the most important is the
maximum likelihood method. If L(θ) = L(θ; x1, x2, …, xn) is the likelihood function for the
random variables X1, X2, …, Xn and = ( x1, x2, …, xn) (a function of the observations x1,
x2, …, xn) is the value of θ which maximizes L(θ), then = (X1, X2, …, Xn) is the
maximum-likelihood estimate of θ for the sample x1, x2, …, xn. The most important cases are
those in which X1, X2, …, Xn is a random sample from some density f(x ; θ), so that the
likelihood function is L(θ)= f(x1 ; θ). f(x2 ; θ)… f(xn ; θ). Sometimes finding the maximum of
the Log L(θ) is easier than finding it for L(θ). We know that in this case Log L(θ) and L(θ)
have their maxima at the same value of θ.
If the likelihood function contains k parameters, that is, if,
And dynamic cone resistance qd multiplied by A is equal to soil resistance force:
Fd = qd. A .............................................................................................................. 6-56
After substitution of Equation 6-56 in Equation 6-55 and rearranging, the following equation
is obtained:
Fd = Ex MM+m ................................................................................................... 6-57 Fd. x = E MM+m ................................................................................................... 6-58
If an ideal free fall of hammer is assumed, without any friction between hammer and guide or
any resistance due to air entrapped between hammer and anvil, then the imparted hammer
energy equals to the kinetic energy of hammer before the first collision and equals to
Fd: Average dynamic force applied by soil against the DCP cone in N
Even though the force in DCP test was not measured, it is expected that in a similar soil and
in the same depth, the two instruments will face similar soil resistance. In that way, the above
mentioned argument for RapSochs is also valid for DCP. Therefore for the same reason, this
term is neglected in the calculations.
The Energy Model for DCP Data
In DCP the force during the penetration is not measured. However both instruments have a
similar cone dimensions and penetration mechanism. Therefore we expect that the measured
tip force by RapSochs at a certain depth be the same for the DCP.
The average stresses of RapSochs tests versus depth are obtained for each sample. Figure
6-16 shows a plot of tip stresses of several RapSochs tests in RSD and the calculated average
stress versus depth. The weighted average tip stress for the increment depth of each blow of
DCP is calculated. As an example, Figure 6-17 shows the estimated average tip stress for
RSD-01-D-S1 test.
P a g e | 157
Figure 6-16: Average of Tip Resistance of
RapSochs tests in RSD sample.
Figure 6-17: Average of Tip Resistance estimated
for RSD-01-D-S1 test.
Total mass of DCP (as tested at Northeastern University) is 12.52 kg. Based on ASTM 6951-
03, hammer mass is 8 kg (17.6 lb). Therefore mass of rod, guide and connections are 4.52 kg.
DCP has a fixed hammer drop height of 575 mm (22.6 in.).
Using data from DCP tests and the average force obtained by the method explained, the DCP
data are plotted in the energy balance diagram using Equation 6-43. Since the coefficient of
restitution is unknown, the regression is used to calculate the Cr. Figure 6-18 shows the
results for Cr= 0.73. It is believed that the smaller diameter of the penetrometer in DCP and
the lubrication of the penetrometer with oil will prevent and minimize the skin friction
resistance on the DCP. However, examination of the test data from C6S and CSB showed
that the assumption that cohesive soils do not stick to the DCP penetrometer and
consequently do not cause any circumferential frictions is not correct. Therefore, the data
points from these tests are excluded in calculating Cr and are not plotted in Figure 6-18.
0 1 2 3 4
-700
-600
-500
-400
-300
-200
-100
0De
pth
(mm
)Cone Resistance (MPa)
RSD-02-R-S5
RSD-03-R-S6
RSD-04-R-S7
RSD-05-R-S9
RSD-06-R-S8
Average
0 1 2 3 4
-700
-600
-500
-400
-300
-200
-100
0
Dep
th (m
m)
Cone Resistance (MPa)
RSD-01-D-S1
Average from RapSochs
Estimated for DCP
P a g e | 158
These data were significantly below the equality line. It was an indication that not all the
resisting force that is acting on the penetrometer is taken into account. The possible effect of
errors in measuring DCP advancement in accuracy of the energy balance is also shown in the
figure.
Figure 6-18: Energy Balance Diagram of DCP Test Results Using the Energy Model Formula for Cr =
0.73 and the Average Force Obtained from RapSochs Test Results.
For comparison, DCP data are also plotted using Equation 6-61 (Dutch equation). Figure
6-19 shows the result. C6S and CSB data are also excluded in this figure. The possible effect
of errors in measuring DCP advancement in accuracy of the energy balance is also shown in
the figure. As shown, the Dutch equation shows an average error of 9%. Note that in neither
situation the coefficient of determination (R-squared) is a suitable parameter to evaluate the
accuracy of models since the regression was restricted to pass through the origin.
y = 1.00xR² = 0.42
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
0 10 20 30 40 50 60 70
Out
put E
nerg
y, (
N.m
)
Input Energy, E (N.m)
Equality Line±1 mm Constrained RegressionLinear (Constrained Regression)
P a g e | 159
Figure 6-19: Energy Balance Diagram of DCP Test Results using Dutch Equation and the Average Force
Obtained from RapSochs Test Results.
For DCP data and based on presented figures, although it is not feasible to verify that the
energy balance model is more accurate than Dutch formula, it is definite that the Dutch
equation has a 9% error in predicting the hammer-instrument-soil interaction. Nonetheless,
one could conclude that assuming the coefficient of restitution about 0.73 for DCP, the
energy model predict the DCP interaction with soil more accurately than Dutch formula.
Distribution of Data
A total of 42 RapSochs tests and 29 DCP tests were conducted. Table 6-2 summarizes the
number of RapSochs and DCP tests and the range of CBR of the materials of each sample.
Number of recorded data in this table refers to total number of hammer blows. For example
in CSB sample it means that it took 14 hammer-blows to complete one RapSochs penetration
test and it took 3 hammer-blows to complete the only DCP test conducted on this sample.
y = 1.09xR² = 0.00
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
0 10 20 30 40 50 60 70
Out
put E
nerg
y, (
N.m
)
Input Energy, E (N.m)
Equality Line±1 mm Constrained RegressionLinear (Constrained Regression)
P a g e | 160
Table 6-2: Number of Tests for each Sample and Range of CBR Values.
Sample
ID
# of Tests # of Recorded Data CBR**
RapSochs DCP RapSochs DCP Min Max
BSC 5 4 114 11 0.5 1.6
BSL 0 0 0 0 0 0
C6S 2 2 80 24 4 23
CSB 1 1 14 3 2.3 2.7
CSR 1 0 2 0 - -
GRV 3 2 158 98 2.4 35
RSC 5 4 124 17 1 4.7
RSD 5 4 116 20 1 5
RSL 5 4 6 0 - -
SIL 4 4 404 258 20 50
SIS 4 4 205 121 9 20
MS0-6* 7 - 183 - - -
*Includes Sample MS0, MS1, MS2,MS3,MS4, MS5 and MS6. **CBR is estimated using correlation based on DCP index.
It is clear from Table 6-2 that some of the soil types have more weight than other soil types
in the energy balance model regression due to number of data points. Any generalization
should consider this fact for future tests and analysis.
Discussion
Dynamic energy formulas that are developed for pile driving as a mean to predict the “static
resistance of piles” are discredited by many. However, our goal was not to calculate the
ultimate capacity or static resistance of soil (i.e. base, sub-grade or natural soil). The main
reason is to develop a model to predict the soil-instrument interaction during a RapSochs test
which could ultimately be used to relate the measurements of RapSochs to standard DCP.
The presented analytical model showed to be able to predict soil-instrument interaction with
acceptable accuracy. The model was applied to RapScohs data and the only unknown
parameter in the equation was obtained by regression analysis. Comparison of the model with
widely used Dutch formula showed that our model yields better results. The estimated
coefficients of restitution from results of MLE analysis showed a good agreement with the
parameter obtained from regression analysis. Comparison of calculated coefficient of
restitution versus variable parameters in soil-instrument interaction (different hammer drop
P a g e | 161
height and different soil resistance) showed that Cr is independent of hammer impact
velocities and soil strength, and the assumption that Cr is constant is valid. For RapScoch Cr
is estimated to be 0.39.
DCP test data also used to compare Dutch formula and our energy model. Although it is not
feasible to examine the validity of the energy model for DCP data, it was obvious that the
Dutch equation carries an incorporated error for predicting soil-instrument interaction. The
coefficient of restitution for DCP instrument is estimated to be 0.73.
The sensitivity of the energy balance to penetration rate shows that any minor error in
measuring displacement has a greater error in the accuracy of transferred energy. The
recommended procedure by some researchers, to use the average penetration rate for DCP-
CBR (or any other) correlation, is believed to help reduce the corresponding error.
All the RapSochs and DCP tests data used for this analysis are performed in not fully
saturated soil samples. It is well known that in fine-grained saturated soils, pore pressure
build up during penetration influence the penetration resistance. How the dynamic resistance
may influence the energy balance relationship under this condition was not assessed in this
study. Therefore, the validity for using the proposed energy model for tests in fully saturated
soil is not verified.
P a g e | 162
6-3- RapSochs Correlation to CBR
A vast number of DCP-CBR empirical equations exist, which makes correlation of RapSochs
penetration per blow (RS-PR) to DCP-PR more reasonable than its direct correlation to in-
situ or laboratory CBRs. The problem is that in addition to the different geometry and
hammer weight of RapSochs, different drop height should be considered in developing a
RapSochs-DCP correlation.
In this study, the measured index of soil strength is DCPi which could be used to obtain the
CBR for each soil sample. The main purpose of this section is to correlate RapSochs data to
estimated CBR obtained from DCP tests.
Available DCP- CBR Correlations
The DCP is used as a benchmark for in-situ evaluation of shallow soil strength properties.
Extensive research has been performed and reported in literature to empirically relate DCPi
to CBR. While some of the relationships are graphical, others are formulated in the forms of
equations. The most used equation is in the form of:
logCBR = A – B LogDCPi .................................................................................. 6-66
where;
A and B are real numbers.
ASTM D6951/D6951M (2009) suggested that the selection of the appropriate correlation is
up to the professional judgment of the user. However, the USACE suggests using the
equations developed by Webster et al. (1992) and then modified for CL and CH soils by
Webster et al. (1994). Table 6-3 presents a list of some of these correlations.
P a g e | 163
Table 6-3: List of Some DCPi to CBR Correlations.
Equation Reference, Description and Comments
logCBR = 2.73-1.3(logDCPi)
logCBR = 2.7-1.3(logDCPi)
Kleyn (1975), DCPi is the slope of number of blows versus
depth. The anvil is a long hexagon nut*. Cone angle is 60°.
Laboratory compacted soaked CBR is tested in the standard
molds (D= 150 mm). DCP tests are conducted immediately after
the CBR test. Different pavement materials are tested. The first
equation is obtained from the TPA correlation presented in figure
9 and the second equation is the average from figure 10 of that
report. (Initial calibration was done with a 30° cone and later
converted to the 60° cone.)
logCBR = 2.555-1.145(logDCPi)
Smith and Pratt (1983), DCPi is the average of DCP-PR to
penetrate 50 mm into the surface, 30° angled cone, 9.08 kg
hammer falling 508 mm used, Correlated versus in-situ CBR. The
soil tested are clayey materials having LL around 40 and PL
around 15.
lnCBR =5.80-0.95(lnDCPi)
lnCBR =5.93-1.1(lnDCPi)
lnCBR =6.15-1.248(lnDCPi)
lnCBR =5.70-0.82(lnDCPi)
lnCBR =5.86-0.69(lnDCPi)
Sampson (1984), DCPi is the slope of number of blows versus
depth but the last reading before hitting the base plate is excluded.
Cone angle is 60°, CBR is done on the laboratory compacted
soaked samples and the DCP test are conducted immediately after
the CBR test. The presented equations are for; a) All tests, b)
Plastic materials only, c) Materials with PI> 6, d) Materials with
PI<6, e) Materials with PI=0.
logCBR = 2.81-1.32(logDCPi)
logCBR = 2.70-1.12(logDCPi)
logCBR = 2.56-1.16(logDCPi)
logCBR = 3.03-1.51(logDCPi)
logCBR = 2.55-0.96(logDCPi)
logCBR = 2.76-1.28(logDCPi)
logCBR = 2.83-1.33(logDCPi)
Harrison (1986), DCP is performed on samples prepared in
standard CBR molds while a circular steel plate which produce
the same surcharge weight as in CBR test was placed on the
surface. DCPi is the average of DCP-PR in the top 50 mm of the
sample surface while the first penetration is excluded. Cone angle
is 60°, CBR is done on the laboratory compacted soaked and
unsoaked samples, Tested on clay-like soils, well-graded sand,
and well-graded gravel materials, presented equations are for:
1) All materials. Application for clay, SW, and GW is limited to
CBR=2-17, CBR=17-45, and CBR=55-100 respectively.
P a g e | 164
Equation Reference, Description and Comments
2) Granular materials with DCPi < 10 mm/blow (the equation is
reported in Harrison, 1989)
3) Cohesive soils (MH) with DCPi = 10-70 mm/blow
4) Sand (SW) with DCPi =5-15 mm/blow
5) Gravel (GW) with DCPi =4-10 mm/blow
6) Soaked CBR (all materials)
7) Unsoaked CBR (all materials)
logCBR=2.20-0.71(logDCPi)1.5 Livneh (1987) and Livneh and Ishai (1987) / DCPi is the slope of
the number of blows vs. depth at a given linear depth segment.
Cone angle is 30°. From the presented figure it seems that the
anvil is a long hexagon nut*. CBR tests include soaked and
unsoaked laboratory and field testing. Laboratory tests are on
granular soils.
logCBR=2.55-1.14(logDCPi) Harrison (1989), DCPi is the average of DCP-PR in the top 50
mm of the sample surface while the first penetration is excluded.
Cone angle is 60°, CBR is done on the laboratory compacted
soaked and unsoaked samples but the correlation is corrected to
take into the account the confinement effects. The correlation is
for all types of materials.
CBR = 292 / DCPi1.12
Webster et al. (1992) / DCPi is the average of DCP-PR for 20, 10,
5, 3, 2, or 1 hammer blows. The field in-place CBR tests were
conducted following MIL-STD-621A. Based on data in SW, SC,
SM-SC, SP-SM, CL, CH, and GC materials.
CBR = 1/(0.017019×DCPi)2
Webster et al. (1994) / DCPi is the average of DCP-PR where
transition zone data at the top of the surface layer and between
soil layers are excluded and for the soil zone extending 15 cm
below the CBR test. The field in-place CBR tests were conducted
following MIL-STD-621A. For CL materials only when DCPi >
18 mm/blow.
P a g e | 165
Equation Reference, Description and Comments
CBR = 1 / 0.002871×DCPi Webster et al. (1994) / DCPi is the average of DCP-PR where
transition zone data at the top of the surface layer and between
soil layers are excluded and for the soil zone extending 15 cm
below the CBR test. The field in-place CBR tests were conducted
following MIL-STD-621A. For CH materials only. Based on data
DCPi > 20 mm/blow.
LogCBR=2.669-1.065(logDCPi)
LogCBR=2.438-1.065(logDCPi)
Ese et al. (1994), DCPi calculation method is not reported. Cone
angle is 30°. Materials consisted of well-graded gravel with 9 to
19% fines. CBR tests performed on samples prepared in
laboratory at optimum moisture content and saturated condition
(soaked samples) following modified AASHTO. CBR testing was
carried out from the top of the specimen followed by DCP test
from the bottom of the same specimen. They reported that the
first equation results higher CBR with DCP tests in the field. The
second equation is corrected for this issue. In the second equation
the DCPi form the field will yield a laboratory derived CBR.
CBR=320/DCPi0.943
Truebe et al. (1995) / The method to calculate DCPi is not
mentioned. The DCP is the standard DCP used by USACE. The
CBR is in-situ CBR. The materials tested include aggregate
surface (20 < CBR < 86) and MH or ML subgrade (6 < CBR <
22). The equation is valid for 4 ≤ DCPi ≤ 40.
logCBR=2.14-0.69(logDCPi)1.5 Livneh et al. (1995) / DCPi is the slope of the number of blows
vs. depth at a given linear depth segment. Cone angle is 30°.
From the presented figure it seems that the anvil is a long
hexagon nut*. CBR tests include laboratory and field testing.
LogCBR=3.24-1.50(logDCPi)
LogCBR=2.80-1.46(logDCPi)
LogCBR=2.54-1.23(logDCPi)
LogCBR=2.50-1.07(logDCPi)
Al-Refeai and Al-Suhaibani (1997) / Cone angle is 60°. From the
presented figure it seems that the anvil is a long hexagon nut*.
Soil samples obtained from Riyadh area of Saudi Arabia. Test
performed in the laboratory prepared samples in CBR molds. The
result of DCP test and CBR test, performed in similar prepared
samples with similar surcharge load, are compared. DCPi is the
P a g e | 166
Equation Reference, Description and Comments
average of DCP-PRs needed to penetrate 50 mm in the soil while
the first blow is excluded.
1) The first equation: for poorly graded sand, 10 ≤ DCPi ≤ 50
mm/blow,
2) The second equation: for silty sand, 4 ≤ DCPi ≤ 35 mm/blow,
3) The third equation: for CL or ML, 4 ≤ DCPi ≤ 35 mm/blow,
4) The forth equation for all materials, 4 ≤ DCPi ≤ 50 mm/blow.
LogCBR=2.53-1.14(logDCPi)
Coonse (1999) / DCPi is the average of DCP-PR after the first
blow to the 116 mm deep in 150 mm diameter mold and to 200
mm deep in 250 mm diameter mold. The DCP test is performed
after conducting CBR test in the same mold following AASHTO
T-193 while the 44 N surcharge weight was on top of the mold.
Cone angle is 30°. From the presented figure it seems that the
anvil is not flat*. This equation is corrected to consider the
confinement effect of standard CBR molds, so it could be used
for estimating in-situ CBR.
Materials tested included remolded residual clayey soils from
Piedmont Geologic Region, North Carolina.
CH consists of 90% fines, LL= 55, PI =27.
CL consists of 65% fines, LL=33, PI=13.
CL consists of 63% fines, LL=30, PI=12.
The equation is valid for 25 ≤ DCPi ≤ 80. For lower DCPis it
overestimates the CBR.
LogCBR=2.53-1.14(logDCPi)
LogCBR=2.40-0.55(logDCPi)
Gabr et al. (1999) / Cone angle is 30°. Test program included
tests in lab and in the field in Davidson County, North Carolina.
In the laboratory, the DCP test is performed after conducting
CBR test in 150 mm diameter molds. In-situ tests conducted on
Piedmont subgrade soil and aggregate base coarse material
The first equation is for Piedmont subgrade soil. See the
description of Coonse (1999)’s equation for more details.
The second equation is for aggregate base coarse material
obtained from rock quarry (Before compaction: Cc= 0.75, Cu =
25, Passing sieve number 4 = 42%)
P a g e | 167
Equation Reference, Description and Comments
CBR=410 DCPi-1.27
CBR=66.66 DCPi2-330DCPi+563
CSIR Transportek (2000), The DCPi is the average or weighted
average of DCP-PR.
The first equation is for DCPi > 2 mm/blow and,
The second equation is for DCPi ≤ 2 mm/blow.
LogCBR=2.182-0.872(logDCPi)
LogCBR=1.145-0.336(logDCPi)
LogCBR=1.671-0.577(logDCPi)
LogCBR=1.966-0.667(logDCPi)
Karunaprema & Edirisinghe (2002) / The DCP is tested in a 280
mm diameter specimen prepared in the lab while a surcharge of
147 kg/m3 was on the sample. The method to calculate DCPi is
not mentioned, Cone angle is 60°.
For the first, second and third equations, materials tested include
residual clayey and silty sand (according to British Soil
Classification System) and for the fourth and fifth equations
materials are very clayey or silty gravel from Sri Lanka.
1) The first equation is correlated to remolded unsoaked CBR.
The samples were prepared at optimum moisture content and
maximum density following British Standard procedure.
2) The second equation is correlated to partially undisturbed
unsoaked CBR following British Standard procedure.
3) The third equation is correlated to remolded soaked CBR. The
samples were prepared at optimum moisture content and
maximum density following British Standard procedure.
The first, second and third equations are valid for DCPi between
7 to 75 mm/blow.
4) The forth equation is correlated to remolded unsoaked CBR,
following British Standard procedure, the equation is valid for
DCPi between 11 to 386 mm/blow.
CBR= 1161.1/DCPi1.52 Abu-Farsakh et al. (2005) / DCPi is the average of DCP-PR of
the top 300 mm. Cone angle is 60°. The laboratory CBR test is
performed on unsoaked samples according to ASTM D 1883.
Materials tested include CL, CL-ML, SP, GP, GW-GC, and
cement treated soil. The equation is valid for DCPi between 7.5
to 70 mm/blow.
NOTE: * A long hexagon nut which connects the hammer guide and the rod, also acts as anvil, which may not be totally flat at hammer-anvil contact surface.
P a g e | 168
In Figure 6-20 some of the correlations in Table 6-3 are plotted for comparison. As shown,
the majority of the proposed equations are in power form (e.i. log-log). Although,
comparison of different available relations shows that a universal correlation exists between
DCP and CBR, it is clear from the graph that the use of different correlations may result wide
range of CBRs especially in lower DCPi values. Therefore, the selection of a proper equation
is very important.
Figure 6-20: Plots of Some of DCPi-CBR Correlations.
Meanwhile, the data which are used to derive any of the above relations are usually scattered
even more than the variation which is showed in Figure 6-20. As an example data points of
works done by Webster et al. (1992 and 1994) are presented in Figure 6-21.
1
10
100
1 10 100
CBR
(%)
DCPi (mm/blow)
Kleyn (1975) Equtaion 2Smith and Pratt (1983)Sampson (1984) Equation 1Harison (1986), Equation 1Livneh (1987)Harison (1989)Ese et al. (1994), Equation 2Truebe et al. (1995) Livneh et al (1995)Al-Refeai and Al-Suhaibani (1997) Coonse (1999) Gabr et al. (1999), Equation 2, Aggregate BaseCSIR Transportek (2000)Abu-Farsakh et al. (2005)Webster et al. (1992)Webster et al. (1994) , CH soilsWebster et al. (1994) , CL soils
P a g e | 169
Figure 6-21: Plot of DCP and CBR Test Data versus Correlation Equations (after Webster et al., 1994).
The difference in pore pressure and the confining effect of the rigid mold in the CBR
test is one of the known reasons that the laboratory CBR tests generally give higher values
than in-situ CBR (Smith and Pratt, 1983; Harison, 1989). It is also recognized that laboratory
CBR values may be greater than in-situ values due to the idealized compaction process
(Booth et al., 2008). The soaked and unsoaked process is shown to affect CBR results
significantly as well. The results of DCPs with 60° and 30° cone are different (Livneh, 1991).
Due to these considerations, the correlation equation proposed by Webster et al. (1994) is
selected to be used to find the in-situ CBR for CH and CL soils and the equation proposed by
Webster et al. (1992) for other materials. These relations are highlighted in Figure 6-20.
P a g e | 170
CBR of Soil Samples
In previous chapter the DCPi of each DCP test and the average DCPi in each sample
were obtained. Table 6-4 summarizes the estimated CBR based on each DCP test and also for
the average DCPi in each sample. The table also shows the level of variability in the
estimated CBR values.
Table 6-4: CBR of Samples and Tests Estimated from DCPi Values.
NOTE: CBR of C6S is calculated from CBR = 1 / 0.002871×DCPi (Webster et al., 1994) and the rest is obtained from CBR = 292 / DCPi1.12 (Webster et al., 1992).
P a g e | 171
In the discussion presented in Chapter 5, how the DCPi is differentiated from DCP-PR was
explained. In Figure 6-22, the CBR values derived from DCP-PR are compared with the
average CBR values derived from DCPi in each soil samples. Since the CBR values, which
are derived from DCPi, are estimated for the entire sample they are plotted for the entire
depth of the test.
Figure 6-22: Estimated CBR Profile Derived from DCP-PR and DCPi in Soil Samples.
0 1 2 3 4 5 6
-700
-600
-500
-400
-300
-200
-100
0
Dept
h (m
m)
CBR (%)
BSC
BSC-04-D-S3
BSC-05-D-S6
BSC-07-D-S7
BSC-08-D-S4
Averaged
Average CBR
0 10 20 30 40 50 60
0
0
0
0
0
0
0
0
CBR (%)
C6S
C6S-02-D-S6
C6S-03-D-S4
Averaged
Average CBR
0 1 2 3 4 5 6
0
0
0
0
0
0
0
0
CBR (%)
CSB
CSB-02-D-S2
Average CBR
0 10 20 30 40 50 60
0
0
0
0
0
0
0
0
CBR (%)
GRV
GRV-01-D-S8
Average CBR
0 10 20 30 40 50 60
0
0
0
0
0
0
0
0
CBR (%)
GRV
GRV-03-D-S4
Average CBR
0 1 2 3 4 5 6
-700
-600
-500
-400
-300
-200
-100
0
Dept
h (m
m)
CBR (%)
RSC
RSC-04-D-S7
RSC-05-D-S8
RSC-06-D-S4
RSC-07-D-S6
Averaged
Average CBR
0 1 2 3 4 5 6
0
0
0
0
0
0
0
0
CBR (%)
RSD
RSD-01-D-S1
RSD-07-D-S2
RSD-08-D-S3
RSD-09-D-S4
Averaged
Average CBR
0 10 20 30 40 50 60
0
0
0
0
0
0
0
0
CBR (%)
SIL
SIL-02-D-S3
Average CBR
0 10 20 30 40 50 60
0
0
0
0
0
0
0
0
CBR (%)
SIL
SIL-06-D-S9
SIL-07-D-S4
SIL-08-D-S2
Averaged
Average CBR
0 10 20 30 40 50 60
0
0
0
0
0
0
0
0
CBR (%)
SIS
SIS-03-D-S7
SIS-04-D-S8
SIS-06-D-S2
SIS-08-D-S3
Averaged
Average CBR
P a g e | 172
Theoretical Relation between RapSochs and DCP Penetration per Blow The energy models, developed to predict the soil-instrument interaction in RapSochs and
DCP, could be used to correlate RS-PR to DCP-PR. Known parameters of DCP instrument in
Equation 6-43, are:
mDCP = 4.52 (kg)
MDCP = 8.0 (kg)
hDCP = 0.575 (m)
Cr,DCP = 0.73
Replacement of these parameters into Equation 6-43 and simplifying results in the following
DCP-PRRS is the estimated DCP-PR from RapSochs tests results.
A code is developed to use Equation 6-100 to estimate the equivalent DCP-PR based on RS-
PR. It is also used to plot the measured RS-PR, DCP-PRRS, averaged DCP-PR which is
obtained from DCP tests, and hammer drop height corresponding to RS-PRs for each
RapSochs test. Figure 6-29 shows an example of outputs for test RSC-08-R-S3.
P a g e | 184
Figure 6-29: Estimated DCP-PR from RS-PR in RSC-08-R-S3.
For all the RapSochs tests, including tests in clayey materials, the DCP-PRRS is obtained.
DCP-PRRS profiles of tests in similar conditions are plotted along with the calculated average
DCP-PRRS. The averaged DCP-PR which was calculated from DCP tests are also overlaid for
comparison. Figure 6-30 shows the profiles of estimated DCP-PRRS, averaged DCP-PRRS,
and averaged DCP-PR.
0 8 16 24 32-700
-600
-500
-400
-300
-200
-100
0
Hammer Drop Height (cm)
0 100 200 300 400
-700
-600
-500
-400
-300
-200
-100
0
Penetration Rate (mm/blow)
Dep
th (m
m)
RSC-08-R-S3
Drop HeightRS-PRDCP-PRRSAveraged DCP-PR
P a g e | 185
Figure 6-30: DCP-PRRS profile, Averaged DCP-PRRS and Averaged DCP-PRRS in the Same Test
Conditions.
Figure 6-30 shows that in non-cohesive soils the method yields estimation of DCP-PR with
acceptable accuracy. In these soil types, while DCP is giving one penetration increment close
to soil surface, the RapSochs provides a smoother transient zone. In CSB the DCP profile is
not complete up to the soil surface and no comments could be made regarding the accuracy
of the estimation. However it shows that RapSochs can provide data where DCP cannot. In
C6S, the model underestimates the DCP-PR. In this project, the number of tests conducted in
cohesive materials is limited, and not all soil conditions are tested. Therefore, at this time,
deriving a relationship between RS-PR and DCP-PR is not statistically possible in cohesive
soils.
0 100 200 300 400
-700
-600
-500
-400
-300
-200
-100
0D
epth
(mm
)Penetration Rate (mm/blow)
BSC
BSC-01-R-S1
BSC-02-R-S2
BSC-03-R-S9
BSC-06-R-S8
Averaged DCP-PRRS
Averaged DCP-PR
0 25 50 75 100
0
0
0
0
0
0
0
0
Penetration Rate (mm/blow)
C6S
C6S-01-R-S8
C6S-04-R-S2
Averaged DCP-PRRS
Averaged DCP-PR
0 100 200 300 400
0
0
0
0
0
0
0
0
Penetration Rate (mm/blow)
CSB
CSB-01-R-S8
Averaged DCP-PRRS
Averaged DCP-PR
0 25 50 75 100
0
0
0
0
0
0
0
0
Penetration Rate (mm/blow)
GRV
GRV-02-R-S5
Averaged DCP-PRRS
Averaged DCP-PR
0 25 50 75 100
0
0
0
0
0
0
0
0
Penetration Rate (mm/blow)
GRV
GRV-04-R-S2
GRV-05-R-S6
Averaged DCP-PRRS
Averaged DCP-PR
0 100 200 300 400
-700
-600
-500
-400
-300
-200
-100
0
Dep
th (m
m)
Penetration Rate (mm/blow)
RSC
RSC-01-R-S1
RSC-02-R-S2
RSC-03-R-S9
RSC-08-R-S3
RSC-09-R-S5
Averaged DCP-PRRS
Averaged DCP-PR
0 100 200 300 400
0
0
0
0
0
0
0
0
Penetration Rate (mm/blow)
RSD
RSD-02-R-S5
RSD-03-R-S6
RSD-04-R-S7
RSD-05-R-S9
RSD-06-R-S8
Averaged DCP-PRRS
Averaged DCP-PR
0 15 30 45 60
0
0
0
0
0
0
0
0
Penetration Rate (mm/blow)
SIL
SIL-01-R-S1
Averaged DCP-PRRS
Averaged DCP-PR
0 15 30 45 60
0
0
0
0
0
0
0
0
Penetration Rate (mm/blow)
SIL
SIL-03-R-S5
SIL-04-R-S6
SIL-05-R-S7
Averaged DCP-PRRS
Averaged DCP-PR
0 15 30 45 60
0
0
0
0
0
0
0
0
Penetration Rate (mm/blow)
SIS
SIS-01-R-S5
SIS-02-R-S1
SIS-05-R-S4
Averaged DCP-PRRS
Averaged DCP-PR
P a g e | 186
In the next step, DCPiRS (DCPi estimated from RapSochs test) is calculated for each
RapSochs test by averaging the estimated DCP-PRRS after the transient zone data points are
excluded. To find the transient zone, data points from the soil surface are checked one by
one. If the DCP-PRRS is smaller than the average of DCP-PRRSs plus one standard deviation
it is included. Otherwise it is excluded from data points and average and standard deviation
are calculated for the rest of data and the next DCP-PRRS is checked to verify if it is in the
transient zone or not. The iteration is continued until the first data point within the acceptable
range of average plus one standard deviation is found. The blows above that point are
considered to be in transient zone. A similar procedure is also followed to find the transient
zone close to the bottom of the container. Starting from the last penetration measurement, if
the data point is greater than average minus one standard deviation of DCP-PRRSs, it is
included. Otherwise it is considered to be in transient zone.
The calculated DCPiRSs are summarized in Table 6-6. Average DCPiRS for tests in similar
test conditions are also calculated. Average DCPi which were obtained from DCP tests are
also presented for comparison.
P a g e | 187
Table 6-6: DCPi Obtained by Averaging the Estimated DCP-PR from RS-PR.
Sample-ID Test-ID DCPiRS
(mm/blow)
Average DCPiRS
(mm/blow)
Average DCPi
(mm/blow)
BSC
BSC-01-R-S1 150.0
122.3 129.7
BSC-02-R-S2 99.0
BSC-03-R-S9 131.8
BSC-06-R-S8 108.5
BSC-09-R-S5* 74.8
C6S C6S-01-R-S8 13.6
14.1 21.3 C6S-04-R-S2 14.5
CSB CSB-01-R-S8 129.2 129.2 72.5
GRV
GRV-02-R-S5* 9.3 9.3 11.8
GRV-04-R-S2 8.5 8.2 8.0
GRV-05-R-S6 8.0
RSC
RSC-01-R-S1 48.0
45.2 44.1
RSC-02-R-S2 50.0
RSC-03-R-S9 43.0
RSC-08-R-S3 41.0
RSC-09-R-S5 44.0
RSD
RSD-02-R-S5 40.0
44.2 42.1
RSD-03-R-S6 41.3
RSD-04-R-S7 47.3
RSD-05-R-S9 44.6
RSD-06-R-S8 48.0
SIL
SIL-01-R-S1* 10.5 10.5 9.0
SIL-03-R-S5 7.3
8.1 6.6 SIL-04-R-S6 8.3
SIL-05-R-S7 8.6
SIS
SIS-01-R-S5 15.0
17.1 13.8 SIS-02-R-S1 17.9
SIS-05-R-S4 17.9
SIS-07-R-S6 17.6
NOTE: BSC-09-R-S5 is excluded in calculation of Average DCPiRS due to concerns about the test spacing which affects the
boundary conditions of tests and test interference. GRV-02-R-S5 was performed before draining the sample and GRV-04-R-S2 and GRV-04-R-S2 were performed after
draining the sample. SIL-01-R-S1 was performed 3 months before other tests in SIL sample. The effect of cementation is obvious in this case.
P a g e | 188
Estimation of CBR from RapSochs
CBR could be calculated by the equation developed by Webster et al. (1992) or other
relationships listed in Table 6-3. The calculated DCPiRS should be used instead of DCPi in
those formulas. Table 6-7 summarizes the estimated CBR for each RapSochs test and also for
the average DCPiRS of each sample. The average CBR obtained from DCP tests are also
reported for comparison.
For measured CBR in laboratory compacted soil samples, one data set shows the coefficient
of variation to be 6.9% (compacted per ASTM D698 test method) or 9.2% (compacted per
ASTM D1557 test method) based on seven repetitions (ASTM D1883, 2007). For in-situ
CBR test, no precision is reported in ASTM D4429 (2009). The coefficient of variation for
DCP tests in the same material or same location is reported to be 10% by Harrison (1989),
40% by Smith and Pratt (1983), and 23% by Livneh (1987). The interesting point is they all
reported higher coefficient of variation for CBR tests than for DCP tests. Therefore it could
be concluded that DCP test is more reproducible than CBR.
ASTM D6951-03 Method for DCP indicates that the within-field-laboratory repeatability,
standard deviation has been determined to be less than 2 mm/blow. In Figure 6-31, the effect
of plus and minus 2 mm error in DCP measurement on the estimation of CBR using
equations of Webster et al. (1992) and Webster et al. (1994) are shown. For instance, if the
real CBR is about 50, penetration variation in the range of plus and minus 2 mm/blow means
CBR range of 35 to 85 percent. In this respect, the estimated CBRs for non-cohesive soils are
within the acceptable range of repeatability reported for DCP.
P a g e | 189
Table 6-7: Average CBR of Samples and Tests Estimated from RapSochs and DCP.
Sample-
ID
Test-ID DCPiRS
(mm/blow)
CBRRS
(%)
Ave. of
DCPiRS
(mm/blow)
Ave. CBR
from RS Tests
(%)
Ave. CBR from
DCP Tests
(%)
BSC
BSC-01-R-S1 150.0 1.1
122.3 1.3 1.3
BSC-02-R-S2 99.0 1.7
BSC-03-R-S9 131.8 1.2
BSC-06-R-S8 108.5 1.5
BSC-09-R-S5* 74.8 2.3
C6S C6S-01-R-S8 13.6 26
14.1 25 16 C6S-04-R-S2 14.5 24
CSB CSB-01-R-S8 129.2 1.3 129.2 1.3 2.4
GRV
GRV-02-R-S5* 9.3 24 9.3 24 18
GRV-04-R-S2 8.5 27 8.2 28 28
GRV-05-R-S6 8.0 28
RSC
RSC-01-R-S1 48.0 3.8
45.2 4.1 4.2
RSC-02-R-S2 50.0 3.7
RSC-03-R-S9 43.0 4.3
RSC-08-R-S3 41.0 4.6
RSC-09-R-S5 44.0 4.2
RSD
RSD-02-R-S5 40.0 4.7
44.2 4.2 4.4
RSD-03-R-S6 41.3 4.5
RSD-04-R-S7 47.3 3.9
RSD-05-R-S9 44.6 4.2
RSD-06-R-S8 48.0 3.8
SIL
SIL-01-R-S1* 10.5 21 10.5 21 25
SIL-03-R-S5 7.3 32
8.1 28 35 SIL-04-R-S6 8.3 27
SIL-05-R-S7 8.6 26
SIS
SIS-01-R-S5 15.0 14
17.1 12 15 SIS-02-R-S1 17.9 12
SIS-05-R-S4 17.9 12
SIS-07-R-S6 17.6 12
NOTE: CBR of C6S is calculated from CBR = 1 / 0.002871×DCPiRS and the rest is obtained from CBR = 292 / DCPiRS
1.12 .
P a g e | 190
Figure 6-31: DCPi-CBR Correlations and Range of CBR due to ± 2mm DCPi Variability.
1
10
100
1 10 100
CBR
(%)
DCPi (mm/blow)
Webster et al. (1992)
Webster et al. (1992) ± 2 mm
Webster et al. (1994) , CH soils
Webster et al. (1994) ± 2 mm, CH soils
Webster et al. (1994) ± 2 mm , CL soils
Webster et al. (1994) ± 2 mm , CL soils
P a g e | 191
CBR Profile of Soil by RapSochs
There is also interest to obtain CBR profiles for soils similar to graphs presented in Figure
6-22. An equation for estimation of CBR from RapSochs test results could be developed by
substitution of the Equation 6-100 that relates RS-PR to equivalent DCP-PR in the equation
proposed by Webster et al. (1992). The notation CBRRS is used to refer to the CBR value that
is estimated by RapSochs test results. Since Equation 6-100 is a function of RS-PR and hRS
in SI units and Webster et al. (1992)’s equation for DCP is in mm, Equation 6-100 is
multiplied by 1000 to be consistent with Webster et al. (1992)’s convention. Substituting and
Narasimha Rao, S. and Veeresh, C., (1996) "Behaviour of Batter Anchor Piles in Marine
Clay Subjected to Vertical Pull Out", 6th International Offshore and Polar.
P a g e | 253
Narasimha Rao, S., R. Ravi, and C. Ganapathy (1997). “Pullout Behaviour of Model Suction
Anchors in Marine Clays”, 7th International Offshore and Polar Enginering
Conference, ISOPE - 97, Honololu, Hawaii, USA, Vol.1, pp.740-743.
Narasimha Rao, S., K. Hema Lathaa, B. Pallavia, and S. Surendrana (2006). “Studies on
pullout capacity of anchors in marine clays for mooring systems”, Applied Ocean
Research, Volume 28, Issue 2, April 2006.
Prakash, S. (1961). Behavior of pile groups subjected to lateral loads, PhD thesis, University
of Illinois, USA.
Shin, E. C., B. M. Das, V. K. Puri, S. C. Yen, E. E. Cook (1993). Ultimate uplift capacity of
model rigid metal piles in clay, Geotechnical and Geological Engineering, Vol. 11,
No. 3, pp 203-215, 1993-09-01.
Supachawarote C., M.F. Randolph, and S. Gourvenec (2004). “Inclined Pull-out Capacity of
Suction Caissons”, Proceedings of the 14th International Offshore and Polar
Engineering Conference, Toulon, France, pp 500-506.
Veeresh C. and Narasimha Rao S. (1996). "Vertical pullout capacity of model batter anchor
piles in marine clays", Marine Georesources & Geotechnology, Volume 14, Number
3, 1996, pp. 205-215.
Zimnik, A.R.; L.R. van Baalen, P.N.W. Verhoef, D.J.M.Ngan-Tillard (2000). The adherence
of clay to steel surfaces. GeoEng 2000 an International Conference on Geotechnical
& Geological Engineering, Melbourne, Australia, 19-24 November 2000. p.
UW0838:1-6. Article on CDrom. ISBN: 1-58716-068-4.
P a g e | 254
Appendix B: RapSochs Rev. 0 Operating
Instructions
Introduction
The Rev 0. RapSochs will differ from the deliverable version (Rev. 1) in a number of ways.
1. Battery power will be provided by a 12 volt battery array where BB2590 Lion 15 volt batteries will be used in Rev 1.
2. The Rev. 0 system will be tethered to a laptop computer for operator interface and control. In the Rev. 1 system a single board computer internal to the system will be used.
3. Hammer actuation will be manually powered instead of controlled by a servo system. 4. A string potentiometer will be used for displacement measurement. In the final
version the displacement may be developed from double integration of the acceleration signal, eliminating the need for the external string pot.
5. The recorded data will not have identifying header information; it will just be a set of columns of ASCII data in engineering units.
General Test Sequence
1. The system is assembled, checked, and initialized as described in the Operation Section below.
P a g e | 255
2. The operator raises the hammer by a predetermined amount and allows it to drop. The drop height will be proportional to the estimated strength of the material. The drop height should be adjusted to produce a penetration between 0.5 and 1 inch per blow.
3. For each hammer blow a dynamic data sequence will be triggered by the impact of the hammer. A data block 200 milliseconds long acquired at 10K samples/sec. will be taken. The total block is then 2048 samples long, with 100 of the samples taken prior to the trigger event. The data for each blow will be stored in an ASCII file with filename = <JOB ID>::_tip_nn.dat where nn is the file index which starts with 1 and is automatically incremented after each blow. The moisture sensor file name is the same as the dynamic data except that the string “tip” is prelaced by the string “moisture”. A second file is produced with filename = <JOB ID>::_tip_nn_converted.dat that contains the tip strain, sleeve strain, acceleration, string pot, and TIP temperature in engineering units. All files are stored in the C:\technical\testing\rapsochs\rawdata\ folder.
4. After completion of the dynamic sequence, a moisture sensor reading is taken by the operator after the penetrometer has stopped moving.
5. Steps 2-5 are repeated until the desired penetration depth has been achieved. The MatLab script “RapSochs_Data_Dump.m” can be run at any time to show progress. In order to prevent damage to the penetrometer the sequence must be stopped just prior to reaching the bottom of the sample. Based upon the actual sample depth, a mark should be placed on the penetrometer rod (or a tape measure or other measuring means) to insure that the penetrometer does not impact the floor.
Operation Procedure
1. Insure the battery in the Processor module is charged. To charge connect the charger to the connector labeled “charger”. To fully charge dead batteries will take 2.5-3 hours. When the batteries are fully charged the “Charging” LED will go out. Remove the charger cable from the Processor module. The instrument will not function if the charger is left plugged in. Battery life will be 4-8 hours of continuous operation. To provide maximum battery life, turn off the system after each test.
2. Insure the system is completely assembled and inspected prior to each test. (see RapSochs Assembly and Operability Checklist). It is important that this procedure be followed. Damage to the system could result from loose mechanical joints or failure to clean and inspect the tip and moisture sensor.
3. Insures all cables are connected. The “Ext Comm” cable must be plugged into the upper USB port on the back of the laptop.
4. Press the “ON” button on the Processor Module to power up the processor. 5. Activate the ‘RapSochs GUI’ on the laptop Desktop. 6. Depress “Continue” on the splash screen. 7. Depress “Setup” on the Main Menu screen. 8. On the Setup screen enter the JOB ID for the current test. The JOB ID will become
the root of the name of the data files created during the test. Only characters legal in windows file names may be used. Select “Operate” when done.
9. For the first blow of a test only, depress the “Reset Test Indexes” button.
P a g e | 256
10. Depress “ON/RESET” to enable power to the TOP and TIP modules. The Acquire Tip and Moisture buttons should become available.
11. Depress the “Moisture” button to acquire a moisture sensor reading, while the tip is resting on the plate. This gives us the value of the stringpot at the surface which will later be used to determine the sinking penetration due to the weight of the penetrometer.
12. Depress “Acquire Tip” to arm the data acquisition sequence. After several diagnostic lines, the message “BIG Acquisition armed…” should appear in the data window.
13. Raise the hammer by an amount that will cause 0.5-1 inch of penetration. This value will have to be determined empirically. For very hard materials, the maximum penetration achievable for the maximum 20 inch drop will be less than 0.5 in. For low CBR materials, such as clay, drop heights of ~4-5 inches are appropriate.
14. The raw data will be echoed to the screen and the file written to the disk. 15. After the first blow of a test it is strongly recommended to display the raw data (using
the procedure described in Data Display below) to insure the penetrometer is working properly. Display of the data will also provide the penetration per blow to allow determination of the drop height necessary to produce 0.5-1 inch penetration. Do not close the RapSochs GUI to activate Matlab.
16. Check the total penetration to insure that the maximum established for the test has not been exceeded.
17. Repeat steps 11-15 until the test is complete. Moisture readings are not necessary after every dynamic blow. It is suggested that they are taken only when the sensor is submerged. Note that at any time during the sequence it is realized that an incorrect action has been performed, such as depressing “Acquire TIP”, when it was desired to depress “Moisture”, the operation can be aborted by pressing “ON/RESET”. This operation will close and files and then re-enable the Moisture and Acquire Tip buttons.
18. Depress “OFF” to remove power from the TOP/TIP modules 19. Depress “OFF” on the processor module to power down the system. 20. Remove the penetrometer from the hole immediately, especially in saturated
materials. The system does have a complete o-ring system to keep moisture from the electronics and sensors but I’d rather not push our luck.
21. Clean the tip as described below before the soil dries out. This is especially important in clays where the material can become like stone if it dries out.
Data Display
The following files can be run at any time during a test as the data files are written immediately after each data block is acquired. To check the results during or after any test, activate Matlab and open the file RapSochs_data_dump.m. Click ‘Run’. At the prompt enter the root (JOB ID from the test) of the filename. Then enter the thickness of the plate used to support the penetrometer during the initial moisture test. Two figures are produced. Figure 1 is a plot of the conductivity and susceptance vs depth from the moisture sensor. Generally, the conductance and susceptance
P a g e | 257
increase with increasing volumetric moisture content. Thus a soil sample with constant gravimetric moisture but with increasing density with depth would show increasing volumetric moisture. Figure 2 is a concatenation of the dynamic data (in engineering units). After plotting the data, the Matlab command window will ask if the test is completed. If not, continue taking data. Typing “n” will replot any new data that has since been gathered, and re-prompt “Is the test complete?”. This can be done between every hammer blow if need be. When the test is complete, reply to the prompt with “y” and a new prompt will ask about generating a .MAT file. If a .MAT file is not necessary, reply to the prompt with an “n”, and the program will complete. If a .MAT file is desired, reply with a “y”. This will bring about a long list of prompts asking for key information specific to the test:
1. A description of the test can be written. Any pertinent details or problems involved in the process or conditions of the experiment.
2. The hammer drop height in inches. 3. The USCS soil classification of the soil. 4. The soil density in pcf. 5. The soil moisture in % gravimetric. 6. The plastic limit of the soil. 7. The liquid limit of the soil. 8. The optimum moisture of the soil in % gravimetric (from the proctor test). 9. The peak dry density of the soil in pcf (from the proctor test). 10. The coefficient of curvature of the soil (from the gradation). 11. The coefficient of uniformity of the soil (from the gradation). 12. The maximum particle size of the soil in mm.
If any of this information is unknown or unimportant, a return carriage will bring about the next prompt.
Operating Modes
In addition to the above described normal operating mode, dynamic and moisture sensor scans can be taken independently. For example repeated moisture sensor scans could be taken to determine repeatability and/or stability of the moisture level in the soil.
Converted File Creation
If for any reason the “Converted” file is not created after a blow (due to for example a crash) and the raw data file is intact, the following procedure can be used.
1. Start the RapSochs GUI as described above in Operation Procedure steps 1-7.
P a g e | 258
2. On the Setup screen enter the JOB ID for the file to be created using only the root part of the tip raw data file. Only characters legal in windows file names may be used. Select “Operate” when done.
3. On the Operate screen enter the file index of the tip file to be converted in the text box to the right and just above the “Convert Tip Data” button.
4. Depress Convert Tip Data to create the converted file. Verify in the \rawdata\ directory that file has been created.
5. Repeat steps 2-4 to convert additional files.
Raw Data File format
Data will be saved in columns separated by at least one space in signed ASCII format. The columns will be 1) tip strain, 2) sleeve strain, 3) axial acceleration, 4) string potentiometer and 5) temperature in °F. Moisture sensor data will be saved in four column format with the frequency in Hz. In column 1, magnitude in dB in column 2, phase in degrees in column 3, and the string potentiometer reading (in volts) in column 4. The file could be simply 5 columns of data as below. Column1 XXXX.x lbf – tip strain Column2 XXXX.x lbf – sleeve strain Column3 XX.xx g – acceleration Column4 XX.xx inches – string pot Column5 XX.XX degree F – temperature
.MAT Data File format
When the .MAT file is opened in Matlab, it will save the following variables to the workspace: ‘job’ = The job i.d. entered in the RapSoChS.exe (name of test). ‘test_description’ = A description of the test entered by user. ‘hammer_height’ = Hammer drop height in inches. ‘uscs_class’ = USCS classification of the soil. ‘density’ = Bulk density of the soil in pcf. ‘moisture’ = Bulk gravimetric moisture of the soil in %. ‘plastic_limit’ = Plastic limit of the soil. ‘liquid_limit’ = Liquid limit of the soil. ‘moist_opt’ = Optimum moisture of the soil from the proctor test in % gravimetric. ‘max_dd’ = Peak dry density of the soil from the proctor test in pcf. ‘cc’ = Coefficient of curvature from the gradation. ‘cu’ = Coefficient of uniformity from the gradation. ‘max_part_size’ = Maximum particle size of the soil in mm. ‘str1’ = Output of the strain gauge in the tip. Force in kN.
P a g e | 259
‘str2’ = Output of the strain gauge in the sleeve. Force in kN. ‘accel’ = Output of the accelerometer in multiples of g. ‘sp’ = Output of the string potentiometer in mm (median filtered). ‘vel’ = Velocity as calculated from the string potentiometer in m/s. ‘sinkage_depth’ = Distance below the surface that the tip sank under the weight of the
penetrometer.
RapSochs Assembly and Inspection Checklist
Pre-Test Inspection 1. At the completion of each test the portion of the penetrometer that was in the soil
should be cleaned with a cloth moistened with distilled water. It is not necessary to remove all soil from the x-ring grooves. The x-rings are tight and could be damaged by attempting to pick soil out of the grooves.
2. All mechanical joints must be tight. Using a wrench (channel locks, vice grip, or similar) above the penetrometer to sensor head connection and one on the cone tip, insure all three joints are tight. The force is transmitted through the mechanical joists on flat loading lands. If the joint is loose, the force will be transmitted instead through the threads which will cause thread distortion. Further there is a hidden mechanical joint just above the upper x-ring. If this joint becomes loose, the fine wires from the sensors could become twisted and fail. This will necessitate returning the unit to TransTech for repair.
3. Inspect the moisture sensor (blue area) to insure the plastic cover is intact. The cover has been coated with acrylic clearcoat to provide abrasion strength and sealing. Grooves have been filled with epoxy. Any exposed copper electrodes must be sealed prior to testing using five minute epoxy or super glue.
4. Insure the TOP module is centered on the purple foam pieces and the clamps are tight.
Assembly Generally in order to place the penetrometer into the soil test fixture, the upper section (above the anvil) must be removed. Do not remove the anvil! The system will be supplied with all cables removed. Generally, only the string potentiometer cable need be removed (at the string pot), and the TOP to Processor cables at the processor to move the system around. The TOP to penetrometer cables should always be left attached as the coax connectors are somewhat fragile.
Pre-Test Assembly 1. Install a guide in the desired position in the test cell. 2. Lower the penetrometer through the guide to the soil surface. Using a level, insure the
penetrometer is perpendicular to the surface. Locate the string potentiometer below the hook on the anvil and attach the string. Attach the string potentiometer cable.
P a g e | 260
3. Place the hammer (small side up) onto the slide and screw the slide rod into the anvil. During the installation do not allow the anvil to rotate as this could damage the cables from the penetrometer to the TOP module. Insure it is tight using a wrench.
4. Attach the TOP to processor cables. 5. Follow the Operation procedure to conduct the test.
Post-Test Dis-assembly Dis-assembly is the reverse of assembly. Again insure the anvil does not rotate when removing the slide rod from the anvil as this could damage the cable between the penetrometer and the TOP module.
Maintenance X-ring replacement. If an x-ring fails, remove the broken ring from the groove. Clean the groove with distilled water and a suitable soft cloth and dry. Stretch a new x-ring over the tip and slide into the grove. Insure the ring is not twisted in the groove (use a magnifier to inspect). A twisted ring will fail quickly.
RapSochs VBTERM Command Interface
If data is to be saved, a log file must be opened prior to using <dump> commands. Except for moisture sensor data, which is provided in engineering units, the results of the dump commands are in raw A/D counts (0-4095). If conversion to engineering units is required, the sensor calibration factors, excitation voltages, and channel gains must be known. Prior to issuing commands insure the battery is charged, the system is on, and the “RAPSOCHS SW Version 1.0” announcement is shown in the VBTERM command window. The following commands are supported: General commands: <MBON><CR> Turn power on to the measurement board <MBOFF><CR> Turn power off to the measurement board Moisture sensor commands: <f1=><nnn><CR> where nnn is the start frequency in KHz. <f2=><nnn><CR> where nnn is the stop frequency in KHz. Note if f2<f1, then the system runs continuously at f1…good for troubleshooting <st=><nnn><CR> where nnn is the # of frequency steps per decade (logarithmic spacing) <tx=><nnn><CR> where nnn is the TX level (0-1023) where 1023 ~=8VPP <rf=><nnn><CR> where nnn is the reference level. Should be <200 to avoid overvoltage at the 8302 input
P a g e | 261
<ph=><nnn><CR> where nnn is the phase offset of the reference (0-359.9°). Not sure whether lead or lag. <de=><nnn><CR> where nnn is the interfrequency delay (in ms?). Default = 80, range (10-300). Not functional at this time. <go><CR> take 1 moisture spectrum and display results on screen Dynamic sensor commands: <acq0><CR> do TOP A/D acquisition (temp, stringpot, moisture sensor data) single sample <dump0> display TOP acquisition results on screen <acq1> do complete acquisition (TIP/TOP). 0.2 seconds of data saved at 10 KHz. (2000 samples, 10 channels) <dump1> dump 2000 sample buffer to screen (or logfile). Sample organization is: 0a 0b 1a 1b 2a 2b 3a 3b 4a 4b 5a 5b where the numeric is the channel number and a,b designate TOP and TIP respectively.
P a g e | 262
Appendix C: List of MATLAB Codes
This appendix presents the list of MATLAB codes used for database establishment and data
analysis. All the codes developed in MATLAB are restored as M-files which are source
codes in MATLAB. They are files in ASCII format with extension ".m". An m-file may be
either a list of commands or a function with input and output variables. All the codes are
available in the folder “\m-files” on the CD-ROM along this document. In the following table
only main codes are listed. In a main code different functions may be called to do some of
the analysis. Those functions are not listed in this table.
Name and Path of the Code / Description \m-files\01-Data to Mat\RAPSOCHs_Data_Dump_Rev2.m
This code reads data-files of a RapSochs test (raw data in DAT format) and put them
together in a MAT-file. The code does not put "moisture data" in MAT-files.
\m-files\01-Data to Mat\Adding Moisture Data to database\RAPSOCHs_Moisture_Dump2.m
This code reads moisture measurements data-files of a RapSochs test (raw data in
DAT format) and put them together in the corresponding MAT-file.