For Volume 1:

Geotechnical and Geophysical Site Characterisation 5 – Lehane, Acosta-Martínez & Kelly (Eds)
Relative density prediction based
© 2016 Australian on in-situ
Geomechanics andSydney,
Society, laboratory measurements
Australia, ISBN 978-0-9946261-1-0
of shear wave
Relative velocity
density prediction based on in-situ and laboratory measurements
Relative density
of shear wave prediction
velocity
For Volume 2: based on in-situ and laboratory measurements
of shear wave velocity
T. Biryaltseva
Fraunhofer InstituteGeotechnical and Geophysical
for Wind Energy and Energy Site Characterisation
System Technology, 5Bremerhaven,
– Lehane, Acosta-Martínez
Germany & Kelly (Eds)
T. Biryaltseva © 2016 Australian Geomechanics Society, Sydney, Australia, ISBN 978-0-9946261-2-7
T. Biryaltseva
T. Lunne Institute for Wind Energy and Energy System Technology, Bremerhaven, Germany
Fraunhofer
Norwegian Institute
Fraunhofer Geotechnical Institute,
for Wind EnergyOslo,
andNorway
Energy System Technology, Bremerhaven, Germany
T. Lunne
S. Lunne
T. Kreiter Geotechnical
Norwegian & T. Mörz Institute, Oslo, Norway
MARUM Center for marine and environmental science, Bremen University, Bremen, Germany
Norwegian Geotechnical
S. Kreiter & T. Mörz Institute, Oslo, Norway
S. KreiterCenter
MARUM & T.forMörz
marine and environmental science, Bremen University, Bremen, Germany
MARUM Center for marine and environmental science, Bremen University, Bremen, Germany

ABSTRACT: Extensive laboratory and in-situ testing has been carried out in an onshore test site on the North
Sea Coast near Cuxhaven, Germany. The test site was chosen due to initially high cone resistance (>30 MPa)
ABSTRACT:
in sand and closeExtensive laboratory
stratigraphic linksand in-situSea
to North testing has been
offshore windcarried out in an onshore
soil conditions. The in-situtestprogram
site on the North
included
Sea Coast
ABSTRACT: near
cone penetration Cuxhaven,
Extensive Germany.
testing,laboratory
geophysical The
anddownholetest site
in-situ testing was chosen
hasand
logging been due to initially
carried out
sampling. high cone
in an onshore
A correlation resistance
test site
between (>30 MPa)
on thedensity,
relative North
in sand
Sea
shear and
Coast
wave close
near
velocitystratigraphic
Cuxhaven,
and mean links
Germany. to
effectiveTheNorth
test Sea
stress site offshore
was wind
chosen
is proposed duesoil
based conditions.
to laboratory
on initially high The
conein-situ program
resistance
measurements with(>30included
MPa)
bender el-
cone
in penetration
sand and close testing, geophysical
stratigraphic links todownhole
North Sea logging and
offshore sampling.
wind soil A correlation
conditions. The
ements under anisotropic stress conditions. The proposed correlation is applied on in-situ shear wave velocitybetween
in-situ relative
program density,
included
shearpenetration
cone
data wave
and velocity
compared and geophysical
testing,
with mean effective
relative density stress
downhole islogging
estimated proposed
overandbased on laboratory
sampling.
density logs measurements
A correlation
and recognized between
empirical with
relative
CPT bender el-
density,
correla-
ements
shear under
wave anisotropic
velocity and stress
mean conditions.
effective The
stress is proposed
proposed correlation
based on is applied
laboratory on in-situ
measurements
tions. Taking the density logging as a reference, the new correlation provides comparable accuracy range as shear wave
with velocity
bender el-
dataexisting
ements
the and compared
under empirical with
anisotropic relative
stress
CPT-based density
conditions.
relationsestimated
The over
proposed
although density
correlation
with a slightlogstendency
isand recognized
applied empirical
toonunderestimation.
in-situ CPT
shear wave correla-
velocity
This novel
tions.
data Taking
and compared
correlation the density
could bewith logging
usedrelative as a reference,
density estimated
as an alternative the new correlation
over density
and conservative provides
logs andprocedure
verification comparable
recognizedforempiricalaccuracy range
CPTdensity
the relative as
correla-
of
the
tions.existing
sands. Taking empirical
the densityCPT-based
logging asrelations
a although
reference, the with
new a slight
correlation tendency
provides to underestimation.
comparable accuracy This novel
range as
correlation
the existing could be used
empirical as an alternative
CPT-based relations and conservative
although with a verification
slight tendency procedure for the relativeThis
to underestimation. density of
novel
sands.
correlation could be used as an alternative and conservative verification procedure for the relative density of
sands.

1 INTRODUCTION
2007). Wave propagation velocities depend on the
1
In INTRODUCTION
geotechnical practice it is very difficult and fre- mechanical properties of the media and therefore are
1quently
INTRODUCTION
impossible to obtain undisturbed sand sam- 2007).
intrinsicWave propagation velocities depend on the
soil parameters.
In geotechnical
ples. As the geotechnical practice itdesign is veryparameters
difficult and relyfre- on mechanical
2007). Wave properties
propagation
In this study a correlation of the media
velocities and therefore
depend
developed among are
on the
D r,
quently
In impossible
geotechnical to
practice obtain
it is undisturbed
very
laboratory tests, sand specimens have to be reconsti- difficult sand
and sam-
fre- intrinsic
mechanical soil parameters.
properties of the media and
Vs and effective stresses in laboratory is applied and therefore are
ples.
tuted As
quently thelaboratory
in impossible
the geotechnical to andesign
to obtain parameters
undisturbed
estimated rely
sandrelative
in-situ sam- on In this
intrinsic
validated study
soil
at a correlation
parameters.
a location with available among Dr,
developedP-S-logging,
laboratory
ples.
density. tests,
As Reconstitution sand
the geotechnical specimens
ofdesign have
the samples to
parameters be reconsti-
has inrelyturnon to V
CPTs In effective
andthis
and study
density stresses
a correlationin laboratory
log data. The developed is
questionapplied
among and
Dr,
whether
tuted in
laboratory the laboratory
tests, sand to an
specimens estimated
be based on interpretation of in-situ tests. The most have in-situ
to be relative
reconsti- Vvalidated
and at a
effective location
stresses with
in available
laboratory
cone resistance (qt)-based and the Vs-based correla-
s is P-S-logging,
applied and
density.
tuted
common in the Reconstitution
laboratorytotoof
technique anthe
estimate samples
estimated has inrelative
in-situ
the in-situ turn to
relative CPT
validated and atdensity
tions provide a comparable
location log data.
with The andquestion
available
results whether
P-S-logging,
the reasons in
be
density (Dr) is to interpret the cone penetrationmost
based
density. on interpretation
Reconstitution of of
the in-situ
samples tests.
has The
in turn to
test cone
CPT resistance
and density (q )-based
log data.
case of discrepancies are discussed.
t andThethe V -based
question
s correla-
whether
common
be based ontechnique to
interpretation estimate
of
results using internationally recognized empirical the
in-situ in-situ
tests. The relative
most tions resistance
cone provide comparable
(qt)-based and results
theand the reasons
Vs-based correla-in
common
correlations r) based
density (Dtechnique is to interpret
ontolarge the cone
estimate the penetration
calibration in-situ
chamber relativetest
tests case of discrepancies are discussed.
tions provide comparable results and the reasons in
results
density using
(e.g Jamiolkowski internationally
(Dr) is to interpret et al. 2003, recognized
the cone Baldi penetrationempirical
et al. 1986). test case
2 STUDY of discrepancies
AREA are discussed.
correlations
results using based on large
internationally calibration
recognized
Uncertainties in the Dr -estimation cause a wide var- chamber tests
empirical
(e.g Jamiolkowski
correlations
iation of expected etlarge
based onin-situ al. 2003, Baldichamber
calibration
soil behavior, et which
al. 1986).
tests
can 2
TheSTUDY
study areaAREA is a sand pit in operation located
Uncertainties
(e.g Jamiolkowski in the D
et -estimation
al. 2003,
lead to a conservative and uneconomic foundation
r cause
Baldi et a wide
al. var-
1986). 2south
STUDY AREA
of the town Cuxhaven in Northern Germany,
iation of expected
Uncertainties
design. in the in-situ soil behavior,
Dr -estimation cause awhich wide var- can The
closestudy
to theareaNorth is aSeasand pit (Fig.
coast. in operation
1). Priorlocated
to the
lead
iation to
Research a conservative
of expected is being in-situ and
carried uneconomic
soilout behavior, foundation
which can
for establishing re- south
The of
study
testing the town
area ishere
described Cuxhaven
a sand
initialpit in
CPT Northern
operationGermany,
in screening located
was per-
design.
lead
liabletomethods
a conservative and uneconomic
for estimating relative density foundationfrom close
south
formed tototheidentify
of NorthCuxhaven
town Sea coast.
sand layersin(Fig. 1).
Northern
with Prior
cone to the
Germany,
resistance
Research
design.
other in-situ istests. beingOne carried out forisestablishing
approach the measure- re- testing
close described
valuestooverthe 30North
MPa.here
Sea initial CPT
coast.further
During (Fig.screening
1). Prior
sand wasto per-
recovery thein
liable methods
Research is for
being estimating
carried
ments of in-situ shear wave velocities. Such meas-out relative
for density
establishing from
re- formed
testing to identify
described heresand
initiallayers
CPT
the pit 18 m of sand were excavated and the de- with cone
screening resistance
was per-
other methods
liable
urements in-situcan tests.
beformade Oneusing
estimating approach relative
suspension is density
thelogging
measure-
from (P- values
scribedover
formed to 30was
MPa.
identify
work During
sand
carried layers
outfurther
with sand
from cone recovery
resistance
the floor of thein
ments
other of
in-situ in-situtests.shear
One wave velocities.
approach
S-logging) which provide velocities of compression- is theSuch meas-
measure- the
valuespit 18
over m
30 of
MPa. sand were
During excavated
further
excavated area, in the following referred to here as sand and the
recovery de-
in
urements
ments
al (Vp)ofand canshear
in-situ be shear
made
(Vs)waveusing
waves, suspension
velocities. logging
Such
other techniques meas- (P-
in- scribed
the
the pit 18
groundworkm wassand
of
level. carried
wereoutexcavated
from the andfloorthe of de-
the
S-logging)
urements
clude seismic which
can becone provide
madepenetration velocities
using suspension of compression-
test (SCPT)logging(e.g. (P- excavated
scribed area,
worktowas
In order in the
carried
verify following referred
out from the floor
a laboratory-based to here
of the
correlation as
al (Vp) andwhich
S-logging)
Robertson etshear (Vs) and
provide
al. 1986) waves,
velocitiesotherof techniques
measurement compression-
of surface in- the
withground
excavated
in-situ level.
area,
datainthethechosen
following fieldreferred
site hastotohere as
fulfill
clude
al (V ) seismic
and shear cone(V penetration
) waves,
wave velocities (e.g. MASW) (Long and Donohue
p s test
other (SCPT)
techniques (e.g.
in- the In order
ground to
level. verify a laboratory-based
the following criteria: i) grain size distribution and correlation
Robertson
clude seismic et al.cone 1986) and measurement
penetration test (SCPT) of surface
(e.g. with
In in-situ
order todata the achosen
verify field site has
laboratory-based to fulfill
correlation
wave velocities
Robertson et al. (e.g.1986)MASW) (Long and Donohue
and measurement of surface the following
with in-situ data criteria: i) grain
the chosen sizesite
field distribution and
has to fulfill
wave velocities (e.g. MASW) (Long and Donohue 389 the following criteria: i) grain size distribution and
grain shape and void ratios comparable to laboratory
samples ii) water-saturation iii) bulk density varia-
tions over depth as a function of packing.
Campaign 2015 included grain size analyses and
emin, emax measurements according to DIN. The data
shows three uniformly graded deposits with con-
sistent grain size distributions and near uniform in-
dex void ratios. (see Table 1).

Figure 1. Study area, locations investigated in 2013 and 2015

with an example of the spatial configuration of CPT, downhole
logging and sampling boreholes. Campaign 2013 included first
screening CPTs and sampling of 2 m3 of sand for additional Figure 2. CUX. Grain size distributions of samples used for la-
Bender Element investigation. Campaign 2015 included CPT, boratory testing. Testing was carried out in two rounds in 2013
downhole velocity and density logging, as well as sampling for and 2014.
application of the new proposed correlation.
3.2 Downhole logging
3 METHODS The 40 m deep flush drilled borehole for geophysi-
cal logging was surrounded by three CPTs carried
out at a distance of 2 m (Fig. 1). The sampling bore-
3.1 Sampling hole was situated 13 m from the logging borehole.
Two sampling campaigns were carried out in the Downhole logging in the open, mud-supported
sand pit. During the Campaign 2013 (see Fig. 1) 2 boreholes included caliper, natural and active gam-
m3 of sand were recovered from an outcrop in the ma as well as neutron porosity measurements and
sand pit corresponding to the layer with initial qt=40 velocity logging.
MPa. After sand excavation this layer can be seen
between 0 and 4 m below ground level (mbgl) in 3.2.1 Density logging
Fig. 3. The sand was carefully mixed and two batch- The applied active gamma logging probe utilizes a
es were taken for subsequent laboratory testing. Cesium-137 source emitting gamma rays of about
Grain size distributions (Fig. 2) and minimum and 0.66 MeV. Two scintillation crystal detectors are
maximum void ratios (emin, emax) were measured ac- situated at distances of 15 and 35 cm from the
cording to NGI and DIN procedures. (Table 1) source, which sets the vertical resolution to 20 cm.
A shield prevents source photons from reaching the
Table 1. Grain size properties of CUX sand. * marked samples detectors without scattering. The depth of investiga-
were used for Vs measurements (Campaign 2013). Parameters tion is considered to be up to 15 cm behind the
according to NGI procedures are used for establishing the cor-
relation. Grain density was measured to 𝜌𝜌 = 2,675 g/cm3
borehole wall. The in-situ data were captured every
(2013). DIN values are given for comparison. 5 cm. Prior testing the tool was calibrated with two
_________________________________________________
Sample, Method ___ d10 d50 d___ Cu ___ emin ____
emax
known bulk densities of fresh water 𝜌𝜌 = 1.0 𝑔𝑔/
___ 60 ___ 𝑐𝑐𝑐𝑐3 and a sand 𝜌𝜌 = 2.05 𝑔𝑔/𝑐𝑐𝑐𝑐3 . The quality of the
mm mm mm / / /
__________________________________________________
calibration of the active gamma probe was tested
CUX, DIN* 0.125 0.23 0.26 2.08 0.55 0.84 with wet bulk densities derived from samples of co-
CUX, NGI* 0.12 0.255 0.26 2.17 0.55 0.81 hesive layers from the same borehole. The maxi-
5-17 m, DIN 0.15 0.25 0.28 1.79 0.54 0.89 mum deviation in wet bulk densities of this test is
22-30m , DIN 0.37 0.93 1.12 2.96 0.37 0.67 4.5 %. Derived wet bulk densities from the neutron
30-38m, DIN 0.25 0.37 0.4 1.57 0.5 0.88
__________________________________________________ porosity log showed to be less accurate and therefore
were excluded from the further investigation.

390
 Figure 3. LOG1. Core description, index void ratios, in-situ CPT, P-S-logging and density log. Ground water level corresponds
to the upper boundary of first till unit (approx. 4 mbgl) but may vary with precipitation. The till layers in the depth 4-5 mbgl and
15-19 mbgl are related to the older Saalian and Elsterian glaciation respectively.

respectively based on oedometer lab tests on strati-

3.2.2 Velocity logging graphic equivalent material. The assumed ice shield
The downhole P-S-tool of Robertson Geologging thickness and the weight of original overlaying stra-
Ltd contains a directional seismic source and two ta were applied to estimate the OCR. Further the co-
seismo-acoustic receivers separated by acoustic efficient of the earth pressure at rest (K0) was calcu-
damping tubes. The seismic source produces a tube lated according to Mayne (2007):
wave at the borehole wall with velocity close to the
shear velocity of the formation together with a com- 0.31
𝑞𝑞 0.22 𝜎𝜎
pressional wave. The waves are recorded with two 𝐾𝐾0 = 0.192 ( 𝑡𝑡⁄𝜎𝜎𝑎𝑎𝑎𝑎𝑎𝑎 ) ( 𝑎𝑎𝑎𝑎𝑎𝑎⁄
𝜎𝜎′
) 𝑂𝑂𝑂𝑂𝑂𝑂0.27 (1)
𝑣𝑣0
2D hydrophones spaced 1 m apart from each other.
Sonic data were captured every 20 cm. (See Fig. 3.)
Exemplary results are shown in Table 2.
3.3 CPTu Table 2. Estimation of OCR. p’c – preconsolidation stress, 𝜎𝜎′𝑣𝑣0
– vertical effective stress. Ground water level is given in Fig.3.
______________________________________________
The CPT readings were recorded every 1 cm with a
depth p’ 𝜎𝜎′ OCR qt K0
15 cm2 van den Berg fully digital 24 bit compression _____ ___
c __𝑣𝑣 ___ ____ ___
ICONE. Details of the CPT campaign are described mbgl kPa kPa / MPa /
______________________________________________
in the companion paper by Biryaltseva et al. (2016). 10 1700 160 11 28 1.08
The data are shown on Fig. 3. CPT, P-S-log and the 15 1700 210 8 49 1.05
23 3500 306 11 47 1.02
density log were depth correlated with the core de-
27 3500 352 10 34 0.87
scription. The depth velocity logging profile was 30 3500 384 9 40 0.86
taken as a reference. 32.5 3500 412 9 53 0.88
_____________________________________________

4 ESTIMATION OF FIELD OCR 5 LABORATORY CORRELATIONS BETWEEN

CPTu and cores reveal two till layers which are re-
lated to the older Saalian (Drenthe stadium) and El-
sterian glaciation (Sindowski, 1965). Due to high
sand content in the till material oedometer testing
could not be applied to estimate the overconsolida-
tion ratio (OCR). Therefore the minimum ice shield
thickness during these glaciations was estimated as
by Otto et al. (2011) to 150-180 m and 350-400 m
VS, DR AND 𝜎𝜎𝑣𝑣′
5.1 Triaxial testing
The laboratory program was carried out at NGI and
consisted of four static isotropically consolidated
and six static anisotropically consolidated drained
triaxial tests. Specimens for isotropic consolidation
were prepared loose (Dr = 50 %), medium dense

391
(65 %), dense (80 %) and very dense (95 %) using
the moist tamping (MT) technique (Ladd, 1978).
Specimens for anisotropic testing were prepared at
Dr = 65 %, Dr = 80 % and Dr = 95 %.

Figure 5. Proposed correlation among relative density, shear

wave velocity and mean effective stress based on laboratory
data. The derived empirical constants C0, C1 and C2 are given.

The empirical constants were found as a solution

of the optimization problem for the equation:
Figure 4. Laboratory test results. Measured shear wave velocity
as a function of relative density and mean effective stress for ′ )𝐶𝐶1
isotropic and anisotropic consolidation. Two groups: dense 𝑉𝑉𝑆𝑆 = 𝐶𝐶0 ∙ (𝜎𝜎𝑚𝑚 exp(𝐶𝐶2 𝐷𝐷𝑟𝑟 ) (3)
(square and triangle) and medium dense (circle and diamond)
can be identified.
At three relative densities (Dr ≈ 95, 78 and 68 %)
eight measurements were selected for data fitting.
A shear wave was triggered and received using
The lowest density (approx. Dr = 49%) was not used
piezoceramic bender elements (BE) placed at the top
due to small number of measurements. Measured da-
and bottom of the specimens. Detailed description of
ta and fits are shown in Fig. 5.
the bender element technique is given in Dyvik &
Madshus (1985). Vs measurements were performed
during consolidation at effective vertical stress of
6 APPLICATION TO IN-SITU DATA
50, 100, 200 and 400 kPa. To account for in-situ
overconsolidation anisotropic consolidation in the
The proposed Dr(Vs) correlation as well as the Dr(qt)
laboratory was achieved with K0=0.7 and K0=1.3.
correlation after Baldi et al. (1986) for overconsoli-
dated sand with C0=181, C1=0.55, C2=2.61 were ap-
5.2 Data fitting plied to the in-situ shear wave velocity and the cone
resistance data respectively (Fig. 5). K0=1 and
Analogous to the empirical formulation between
K0=0.85 were assumed for the upper and lower part
cone resistance and relative density proposed by
of the profile. Averaged emin and emax values (Table
Schmertmann (1976) the dependency of the relative
1) were used to convert measured wet bulk density
density on the shear wave velocity was assumed to
values to relative densities.
be governed by the same structural type of equa-
Pairwise comparison of the measured and pre-
tions:
dicted data over the profile is summarized in Fig. 6.
1 𝑉𝑉S Very high in-situ measured relative density values
𝐷𝐷𝑟𝑟 = 𝑙𝑙𝑙𝑙 [ ] (2)
𝐶𝐶2 ′ )𝐶𝐶1
𝐶𝐶0 ∙(𝜎𝜎𝑚𝑚 and an overall large variability in the data should be
noted. At 5-17 mbgl depth the predicted qt-based
and measured relative density are in good agree-
′
With 𝑉𝑉𝑆𝑆 shear wave velocity [m/s], 𝜎𝜎𝑚𝑚 mean effec- ment. The Dr(Vs) slightly underestimates the meas-
tive stress [kPa], Dr relative density in decimal rep- ured values.
resentation. In the depth 22-30 mbgl both empirical relations
show comparable deviations against the in situ data.
In the depth below 30 m both correlations underes-

392
timate the measured values. However the Vs meas-
urements below 33mbgl show a tendency towards
overestimation.
Figure 6. Comparison of measured (X-axis) and predicted (Y-
axis) relative densities by Baldi et al. (1986) (above) and this
study (below) for different sand layers throughout the profile
with assumed K0 values. Grey symbols represent the depth be-
low 33 mbgl at which VS but no qt data are available. The de-
rived relative density from gamma-gamma logging is based on
assumption of fully saturated sand with grain density of
𝜌𝜌=2,675 g/cm3. Residual sum of squares (RSS) is shown for
every data set.
7 DISCUSSION
The discrepancies between the results of the Dr(Vs)
correlations with the measured data in the upper part
of the profile may be explained with the stress de-
pendency of the empirical relations. The partial de-
′
rivative of Eq. (2) with respect to 𝜎𝜎𝑚𝑚 is equal to:
𝜕𝜕𝐷𝐷𝑟𝑟 𝐶𝐶1
′
𝜕𝜕𝜎𝜎𝑚𝑚
= − 𝐶𝐶 ′ (4)
2 ∙𝜎𝜎𝑚𝑚
The derivative is inversely proportional to stress,
meaning that the dependency decreases with increas-
ing stress. Since the derivative also depends on the
empirical constants the stress dependency of the
proposed velocity correlation is higher compared to
qt-based correlation after Baldi et al. (1986). This
can be explained with the different mechanisms of
the cone penetration and Vs propagation. The cone
penetration includes the stages of punching at shal-
low depth and extrusion at greater depth according
to Emerson et al. (2008). These mechanisms pro-
voke volume as well as shape changes in the sur-
393
rounding soil. The shear wave velocity in turn de-
pends on the grain-to-grain configuration.
Figure 7. Left: Stress dependency of the proposed correlation
(circle) and the CPT-based after Baldi et al. (1986). Right: De-
rivative of the relative density with respect to stress for both
correlations.
For the proposed Vs-based correlation the choice
of the K0 is therefore shown to be more important
than for qt-based one, especially for low stresses un-
der 100 kPa. The assumption about the coefficient of
earth pressure at rest was made in this study based
on Mayne (2007). This empirical relation was estab-
lished using calibration chamber measurements on
freshly pluviated sands and is considered to be a
first estimate. K0=1 leads to a good agreement
among measured data and Dr(qt) in the pre-Saalian
deposit. However the Dr(Vs) underestimates the
measured data. As discussed earlier this can be asso-
ciated with different test mechanisms especially at
shallow depth. Furthermore the form of the equation
may also be not fully suitable for describing the
Dr(Vs) dependency.
Accuracy of the gamma density measurements in
situ is found to highly influence the results. Accord-
ing to information from the subcontractor the differ-
ence between two consequently repeated measure-
ments along the same profile, may be as high as
0.05 g/cm3. For this sand this would lead to the dif-
ferences of 20 percentage points in Dr. As the inves-
tigation depth is considered to be up to 15 cm behind
the borehole wall even in a carefully prepared bore-
hole the drilling can influence the density of the sur-
rounding soil. Mud-cake and tool standoff are com-
pensated using short and long transmitter receiver
spacing but larger drilling induced irregularities may
still have local effects on the measurement. Fur-
thermore the void ratios calculated from the geo-
physical logging are sensitive to the specific gravity.
Increase in specific gravity leads at constant bulk
density to an increase in the void ratio and therefore
to a decrease in the relative density.
 Blaker et al. (2015) showed that the method for
determining index void ratios influences the results
of laboratory testing and therefore also causes the
inconsistencies when determining relative density
in-situ. In this study DIN standard provides a broad-
er range of possible void ratios in comparison to
NGI method. In the case of the same void ratio the
correlation based on NGI method provides lower
relative densities for loose sands and slightly higher
relative densities for dense sands than a correlation
based on DIN method. The ASTM method used for
establishing the qt-based laboratory correlations may
also provide a narrower void ratio range than the
DIN. However no measurements according to
ASTM methods were carried out in this study, there-
fore no quantitative assessment is possible here.
The proposed Vs-based correlation was estab-
lished on a medium-to-fine poorly graded sand. As a
Vs-dependency on the Cu and the amount of angular
particles was shown by Shin & Santamarina (2013),
this correlation can strictly speaking be applied on
the comparable soils only. Like any laboratory cor-
relation the new relationship does not account for
cementation and ageing effects. As discussed in
Robertson and Cabal (2012) the shear wave velocity
increases with age and cementation. Consequently
the relative density of older deposits can be overes-
timated.
Due to probe configuration the shear wave ve-
locity provides an average value over 1 m soil col-
umn, the cone resistance is measured every 2 cm.
However as mentioned by Lunne et al. (1997) un-
derlying layers may influence the cone resistance.
Therefore none of the methods provide the punctual
information about the soil and the results should be
considered as an averaged value around the point of
interest. In the investigated area neither CPT nor P-
S-logging fully reflects thin loose layers.
8 CONCLUSIONS
A correlation among Vs, mean effective stress and Dr
in form of an empirical equation based on laboratory
bender element tests is proposed here for a medium-
to-fine Cuxhaven sand. The correlation is found to
be sensitive to the stress state, but an application of
K0 estimations provides a fairly acceptable slightly
conservative estimation of the relative density in-
situ. Compared to recognized CPT-based empirical
relations applied to the same site the new found cor-
relation is on the safe side.
The estimation of the site specific Vs-Dr correla-
tions can be integrated into the laboratory program
without high additional effort. The proposed correla-
tion can easily be rewritten in terms of the small
strain shear modulus Gmax to provide an estimation
of the Gmax changes with relative density. Although
some shortcomings such as not accounting for the
394
ageing effects and dependency of the predicted rela-
tive density on the stress state have to be mentioned,
a potential of using the shear wave velocity as a
backup procedure for estimating the relative density
is shown. The significance and applicability of the
proposed correlation has to be further investigated
on more data with comparable sets of in-situ meas-
urements.
9 REFERENCES
Baldi, G., R. Belotti, N. Ghionna, M. Jamiolkowski, & E.
Pasqualini . 1986. Interpretation of CPT and CPTU; 2nd
part: drained penetration of sands. In Fourth International
Geotechnical Seminar: 143–156. Singapore.
Biryaltseva, T., T. Moerz, S. Brandt, S. Kreiter, U. Gerdes &
B. Ossig. 2016. Relative densities and void ratios derived
from CPT data using in situ field calibration. ISC’5. In
Press.
Blaker, O., T. Lunne, T. Vestgarden, L. Krogh, N. V. Thom-
sen, J. J. M. Powell, & C. F. Wallance. 2015. Method de-
pendency for determining maximum and minimum dry unit
weights of sands.
Dyvik, R. & Madshus, C. 1985. Lab measurements of Gmax us-
ing bender elements. Advances in Art of Testing: soils un-
der cyclic conditions, ASCE Annual Convention: 186-196.
Detroit, Michigan.
Emerson. M., Foray, P., Puech, A. & Palix, E. 2008. A global
model for accurately interpreting CPT data in sands from
shallow to greater depth. In Huang & Mayne(eds.) Ge-
otechnical and Geophysical Site Characterization.:687-694
London: Taylor & Francis Group
Ladd, R.S. 1978. Preparing specimens using undercompaction;
GTJOAD, 1(1), 16-23.
Lunne, T., Robertson, P.K., Powell, J.J.M. (1997). Cone pene-
tration testing in geotechnical practice. Oxon: Spon Press.
Mayne, P. W. (2007). NCHRP synthesis 368. Cone Penetration
Testing. A synthesis of highway practice. Washington,
D.C.: Transportation research board.
Otto, D., Kreiter, S., Fan, W., Hepp, D.A., Mörz, T., 2011. Es-
timation of minimum Quarternary ice sheet thicknesses and
relative chronology of intersecting buried valleys in the
Southern North Sea. XVIII. INQUA-Congress Quaternary
sciences: the view from the mountains, July 20-27, 2011,
Bern.
Robertson, P., R. Campanella, D. Gillespie, & A. Rice. 1986.
Seismic CPT to measure in-situ shear wave velocity. J. of
Geotechnical Engineering Division, ASCE 112(8), 791–
803.
Robertson Geologging. PS Logger Operational Manual. Rob-
ertson Geologging Ltd.
Robertson, P. & K. Cabal. 2012. Guide to cone penetration
testing for geotechnical engineering. Signal Hill: Gregg
Drilling and Testing, Inc.
Shin, H. & Santamarina, J.C. 2013. The role of particle angu-
larity on the mechanical behavior of granular mixtures.
J. Geotech. Geoinviron. Eng., ASCE, 139(2);353-355
Sindowski, K.-H., (1965). Die drenthestadiale Altenwalder
Stauchmoräne südlich Cuxhaven. Zeitschr. Dt. Ges. Ge-
owiss. 115, 158-162.
Schmertmann, J. H. 1978. Guidelines for cone penetration test,
performance and design. Report FHWA-TS-78-209, U.S.
Department of Transportation. Federal Highway Admin-
istration.