LOWLAND TECHNOLOGY INTERNATIONAL Vol. 13, No.

2, 41-51, December 2011

International Association of Lowland Technology (IALT), ISSN 1344-9656

ON USE OF DYNAMIC PROBING IN SANDY SOILS

H. Abuel-Naga 1, A. Bouazza 2 and M. Holtrigter 3

ABSTRACT:The Dynamic Cone Penetration Tests (DCPT) is a rapid inexpensive field test that can be used to assess
the compactness of soils. However, correlation between the results of DCPT and soil properties or any other trusted
field test is not well established yet. In this paper, a General Method of Data Handling (GMDH) approach was utilized
to investigate the correlation between Dynamic Probing Super Heavy (DPSH) and Standard Penetration Tests (SPTs)
using test results obtained from an intensive site investigation study carried out on sandy soils in Egypt. Linking these
two tests will enable DPSH to make use of well-established correlations between SPT and soil properties. The
developed GMDH model indicates that the relation between the results of the two penetrometer tests is nonlinear for
sandy soils and is a function of soil relative density and effective overburden pressure. The validity of the proposed
correlation was verified using test results on sandy soils from different sites.

Keywords: field test, cone, dynamic, penetration.

INTRODUCTION designing shallow and deep foundations (BS EN ISO

22476-2:2005, DIN 4094-1 (1974), DIN 4094-2 (1980)),
The Dynamic Cone Penetration Test (DCPT) is a only limited research work has been conducted in
simple ground investigation technique in which a solid relation to foundation engineering (Sanglerat 1972;
penetrometer is driven into the ground and the number of Kayalar 1988; Butcher et al. 1996). The advantages of
blows required to drive it to a given depth is recorded. DCPT over other penetration tests are its low cost and
Dynamic cone penetrometers were originally designed to portability. However, although DCPT is a blind
obtain qualitative and quantitative data on the soil investigation tool which does not give direct information
resistance to penetration and in particular to determine on soil type; its data can be exploited in combination
the compactness of cohesionless soil which are usually with the results of borehole or trial pit investigations to
difficult to sample. Four different dynamic probing types provide an economic method of assessing the ground
are recommended for use by the international reference characteristics between each investigation location. It is
test procedures (Stefanoff et al. 1988). These are: particularly useful to detect different strata horizons,
Dynamic Probing Light (DPL), Dynamic Probing strength characteristics, obstructions or voids.
Medium (DPM), Dynamic Probing Heavy (DPH) and The relationship between soil strength parameters
Dynamic Probing Super Heavy (DPSH). They differ in and DCPT results is not well established yet. An
cone size, hammer weight and drop height to fit different approach that can be used to find a link between the
soils conditions and various investigation purposes. DCPT and soil strength parameters is to establish a
In the pavement construction field, light DCPT under correlation between DCPT results and the results of a
ASTM standard specification (ASTM D6951-03) has widely recognized penetrometer test such as the
been used extensively. Moreover, several investigations Standard Penetration Test (SPT). Such approach has
were conducted to establish a correlation between the been used by some researchers (Bergdahl and Eriksson
results obtained from the DCPT and the pavement 1983; Cearns and Mckenzie 1988; Butcher et al. 1996;
material properties such as CBR, unconfined Spagnoli 2007). However, a reliable correlation does not
compressive strength, resilient modulus and shear yet exist (Spagnoli 2007). It is thought that the main
strength (Jayawickrama et al. 2000; Gabr et al. 2000; reason behind the difference and uncertainty of the
Chen et al. 2005). On the other hand, although DCPT is available DCPT-SPT correlations are their failure to
internationally recognized as a soil investigation tool for

1
School of Mechanical, Aerospace and Civil Engineering, University of Manchester, M13 9PL, UK, [email protected]
2
Department of Civil Engineering, Monash University, Melbourne, Australia, [email protected]
3
Ground Investigation Ltd, Auckland, New Zealand, [email protected]
Note: Discussion on this paper is open until June 2012
 Abuel-Naga, et al.

account for the effect of hammer efficiency, soil strength, Table 1 Technical data of the DPSH (BS EN ISO 22476:
effective overburden pressure, and soil type. Part 2)
The aim of this paper is to develop a correlation
between DPSH-SPT results that take into consideration Factor DPSH
these effects using the results obtained from an intensive
site investigation program on sandy soils conducted in Hammer mass, kg 63.5 ± 0.5
Egypt. The approach of general method of data handling
Height of fall, m 0.75 ± 0.02
(GMDH) was used to model such relationship. The
developed GMDH model was used to investigate the Mass of anvil and guide rod
effect of soil strength, effective overburden pressure, and (max), kg 30
soil type on the DPSH-SPT correlation. Furthermore, the
validity of the developed DPSH-SPT correlation was Max mass of rod, kg/m 8
tested using the results obtained from Egyptian sites that
were not used in developing the correlation; two sites Rod OD, mm 35
located in Auckland, New Zealand; one site located in
the south of Portugal; and the correlation equation Apex angle, deg 90
proposed by Cearns and McKenzie (1988).
Nominal area of cone cm2 20
STUDY DATABASE
Cone diameter, new, mm 50.5 ± 0.5

The database of this study was drawn from 14 Mantel length of cone, mm 50.5 ± 2
different sites located in Egypt. The soil investigation
program for each site included one SPT borehole and Cone taper angle, upper, Deg. 11
three to five adjacent DPSH tests to depths of 15.00 m.
The distances between SPT and DPSH tests locations Length of cone tip, mm. 25.3 ± 0.4
varied between 2.0 to 6.0m. SPT tests, at one meter
Number. of blows/cm penetration 20 cm; N20
depth intervals, were conducted according to ASTM
standard specifications (ASTM, D1586-99) where the Standard range of blows 5 - 100
first 15 cm of penetration was disregarded and the
number of blows for the next 30 cm penetration was Specific work per blow; Mgh/A,
238
recorded as N30. kJ /m2
DPSH test equipment and procedure followed the
guidelines recommended by British standard (BS EN
ISO 22476-2:2005) as listed in Table 1. The number of
blows required to drive the cone penetrometer 0.20 m The SPT-borehole samples were described and
into the ground, DC20, was measured continuously classified according to ASTM standard specifications as
throughout the test depth (15.0 m). The DPSH rods were shown in Fig. 1 for some of these sites. In general, the
rotated one and half turns every 0.20 m to keep the rods sites’ stratigraphy consists mainly of sandy soils with
straight and vertical. The torque necessary to turn the occasional silty clay lenses. For the soil classified as SM
driving rods is usually measured to evaluate the effect of or SC, the percentage of passing sieve no. 200 (0.075
the skin friction on DPSH blow count measured during mm) is less than 26%. Soil description was converted
probing. The skin friction could develop along the rod into a continuous numerical value using a soil
length due to the annulus around the rods squeezing in or classification index, Ic, proposed by Robertson (1990) as
collapsing. However, in this site the torque was not shown in Fig. 2. The ground water level was determined
measured since according to the international reference from boreholes and the effective overburden pressure (p)
test procedures for dynamic probing (Stefanoff et al. at penetration test depth was calculated.
1988) dynamic cone test equipment with cone/rod To eliminate the hammer efficiency effect on DPSH-
diameter ratio exceeding about 1.3 leads to results that SPT relation, the standard SPT blow count (N1)60,
are little or not at all influenced by skin friction in corrected to overburden stress equal to 1.0 atmospheric
cohesionless soils. This justification is also supported by pressure and normalized to an effective energy delivered
the experimental results reported by Waschkowski to the rods at 60% of free fall theoretical energy, was
(1982) which shows that in sandy soils the dynamic skin used instead of N30 as follows:
friction is negligible.
 On Use of Dynamic Probing in Sandy Soils

14 14
GL
12 GL 12
GL
SP-SM GL
12 12 SM 10 10
SP-SM GWL GWL
SC
10 10 8 8
GWL GWL SC SP-SM
8 8 6 6
SC
CH

Eleva tion (m)

Eleva tion (m)
CH

Eleva tion (m)

Elevation (m)

6 6 4 4 SC
SP-SM SC SC
4 SP 4 SM 2 2
CH
2 2 SC 0 SP-SM 0 SP
SP-SC

0 0 -2 CH -2 SC
SM SC
SP-SM
-2 -2 -4 SP-SM -4
CH SP-SM
SM SC
-4 -4 -6 -6

BH8 BH52 BH82 BH102

Fig.1 Soil profile at the Egyptian sites

1
( N1 ) 60  N30  C1  C2  C3  C4 (1)

where C1, C2, C3, and C4 are correction factors for 4

hammer energy, anvil, rod length, and overburden
pressure, respectively. In this study, the correction 3.5
factors C1, C2, and C3 were determined by McGregor and
Duncan (1998) whereas C4 can be estimated based on 3
Tokimatsu and Seed (1987) as listed in Table 2. As a
Range of soil
tested in this

pulley safety hammer was used for both tests, correction 2.5
study

factors C1 and C2 can be also used for DPSH test results.

Ic

Furthermore, since the SPT rod length correction is only 2

Clayey SILT to silty CLAY
Silty SAND to sandy SILT

applied for very shallow depths, it can be concluded that

CLAY

C3 is an insignificant correction factor. Therefore, 1.5

assuming the validity of SPT rod length correction factor
Clean SAND to

for DPSH test results will not produce measurable errors. 1

Silty SAND

The validity of SPT overburden pressure correction, C 4,

Gravely

for DPSH test results will be investigated in the 0.5

SAND

following section. Therefore, the standard blow count of

DPSH test, DC60, normalized to an effective energy 0
delivered to rod at 60% of2 free fall theoretical energy can
Soil description
be computed from the measured DC20 as follows:
Fig. 2 Soil classification index (Robertson, 1990)
DC60  DC20  C1  C2  C3 (2) (2)

The data base of this study consists of 201 records. between 1.5 and 2.0 and 35 % of the patterns have I c
The maximum, minimum and average of the database between 2.0 and 2.5. Therefore, the developed
variables are listed in Table 3. However, it should be correlation reported in this study will be only valid for
mentioned that about50% of the patterns haveIc sandy soils.
 Abuel-Naga, et al.

Table 2 SPT correction factors learning procedure, this method allows finding
functional dependence of the output on the most
Factor Variable Correction significant inputs of the system. The algorithm of
GMDH model was originally developed by Madala and
Ivakhnenko (1994). Comprehensive testing of GMDH
Trip or Automatic
proves that it is a powerful tool for mathematical
1.67
Hammer modeling of a wide variety of different real-life problems
Energy (Dolenko et al. 1996). Recently, several geotechnical
Ratio1 Rope and Pulley studies have used GMDH algorithm to interpret the
(c1) 1.0 results of field penetration tests (Ardalan et al. 2009;
Safety Hammer* Kalsntary et al., 2009). NeuroShell II (1996) code
produced by Ward System Group was used in this study
Donut Hammer 0.75 to develop, train, and test the GMDH model.
The algorithm of GMDH model involves generating
Small (4.4 lbs) 0.85
a set of model-candidates in accordance with the specific
Anvil1 Large (26.5 lbs) 0.7 iterative rule. These models compete between
(c2) themselves for a chance to be selected for the next step
Safety (5.5 lbs)* 0.9 of the procedure as shown in Fig. 3. Selection of the
model-candidates is based on the external selection
0 to 3.0 m 0.75 criterion. For example, as shown in Fig. 3, a particular
Rod
Length1 description in the form is used in the first layer,
Over 3.0 m 1.0
(c3)
Layer-1
2.90 t/m2 1.60
Y1
4.80 t/m2 1.30
x1 S
Y2
Over- 9.60 t/m2 1.00 2 E
burden L
x2
pressure2 19.15 t/m2 0.70 Y3 E
(c4) 2
C
28.75 t/m2 0.55 x3 T
Y4
2
I
38.30 t/m 0.50 O
x4 Y5 N
1
Mcgregor and Duncan (1998)
2
Tokimatsu and Seed (1987) Y6

Table 3 Range of database variables Selected neuron (survive)

N60 DC20 p/Pa Ic Not selected neuron (die)

Layer-1 Layer -2
Average 21 15 0.85 -
x1 x1
Y2 Z1 S Z12
Y
Maximum 47.5 49.5 1.64 2.9 E
x2 L
Minimum 5 1.33 0.04 1.5 Y3 E x2
Z2 C 2 Z2
x3 T x3 W=f(x)
I
DESIGN OF GMDH MODEL O
x4 Y5 Z x4
W3 3 N Y35
Z
The GMDH modelling approach is based on sorting Explicit analytical
Selected neuron available optimum
of layer-1 complex model
Selected neuron of layer-2
out procedure, which implements consequent testing of
Selected neuron (survive) Explicit analytical available
models chosen from a set of model-candidates in Fig. 3 Not
Theselected
architecture
neuronof(die)
GMDH optimal complex model
accordance with the given criterion. Using a supervised
 On Use of Dynamic Probing in Sandy Soils

Y  a 0  a1x i  a 2 x j  a 3 x i x j (3) Table 4 Evaluation of GMDH model output results

Training Testing
Then, a particular description in the second layer can be Statistical measurement
data set data set
used as follows, and so on: Mean squared error 0.1523 0.3271

Z  b 0  b1Yi  b 2 Yj  b3 Yi Yj (4) Mean absolute error 0.2899 0.3577

Min. absolute error 0.0009 0.0043

In other words, the output values of a preceding layer
serve as arguments in the succeeding layer. This process Max. absolute error 1.3922 3.2253
continues until the net stops getting better according to a
pre-specified selection criterion. Actually, the algorithm Correlation coefficient r 0.972 0.9586
checks (2d-1) possible models for each two variables xi
and xj, where d is the number of terms in a particular within 5% error 35.938 28.767
description (in our example d=4 as the set of basic

Percentage of data
functions is {1,xi, xj, xixj}). For all possible pairs of input within 5% to 10% error 28.906 28.767
variables, (2d-1)[m(m-1)/2] models must be evaluated,
within 10% to 20% error 17.969 28.767
where m is the number of input variables.
The available database in this study was divided into within 20% to 30% error 7.031 10.959
two groups. The first group, including 9 sites (128
records), was used as a training set whereas the over 30% error 10.156 2.74
remaining 5 sites (73 records) were used for testing the
robustness of the developed GMDH model. The
designed GMDH model has three inputs (DC60, p, and Ic), 3.0
a)
and one output ((N1)60/DC60). The iterative multilayered p= 10 t/m2
algorithm was used. The design control criterion used in 2.5
this network was as follow: Max. Variable in connection
2.0
was set to (x1, x2), which means two variables are
(N1 )60 /DC60

selected. Max. Product term in connection was set to 1.5

(None), which means no covariant allowed. Max.
Variable degree in connection was set to x. Selection 1.0
criterion of the model-candidates was set to regularity Ic=1.5
that minimizes the normalized mean square error of the 0.5 Ic=2.0
model on the test set. The maximum number of survivor Ic=2.5
models in each layer was set to 3. 0.0
0 20 40 60
DC60
DISCUSSION OF RESULTS b) 3.0
I c=2.0
The accuracy of the developed DPSH-SPT 2.5
correlation using the GMDH model was assessed
2.0
statistically using the training and testing data sets, as P t/m2
= 5t/m2
P=5
(N1 )60 /DC60

listed in Table 4. High coefficients of correlations were 1.5 P t/m2

= 10t/m2
P=10
obtained for both of the training and testing data sets.
P t/m2
= 15t/m2
P=15
The developed GMDH model was used to investigate the 1.0
effect of soil type and overburden pressure on (N1)60/ D60
as shown in Fig. 4. In general, for sandy soils, the results 0.5
indicate that (N1)60/ D60 has a non-linear relation with
DC60. Moreover, the overburden pressure has more 0.0
influence on (N1)60/ D60 than the soil type. However, for 0 20 40 60
DC60>20, the effect of both parameters can be ignored as DC60
shown in Fig. 4. Therefore, for simplicity the parameter Fig. 4 The effect of soil type (a), and overburden
Ic can be given a constant value of 2.0 for sandy soils. pressure (b) on the ratio (N1)60/DC60
 Abuel-Naga, et al.

The results of Fig. 4b can be used to estimate the a) 60

overburden correction factor (C4) for DPSH test. As
(N1)60 values are corrected for overburden pressure, the 50 (N1 )60 =27
change of DC60 with overburden pressure at constant
40 (N1 )60 =26
(N1)60 can be used for this purpose as shown in Fig. 5a.

DC60
The DPSH overburden correction factor can be 30
calculated by normalizing DC60, at each (N1)60, for its (N1 )60 =24
value at 1.0 atmospheric pressure as shown in Fig. 5b. 20
The SPT overburden correction factor proposed by (N1 )60 =20
Tokimatsu and Seed (1987) was also plotted in Fig. 5b. 10
(N1)60=14
It shows close agreement with the DPSH results.
0
Therefore, it can be concluded that the SPT overburden
0 5 10 15 20
correction factor can also be used for DPSH test. To p (t/m2 )
check the soundness of this finding, the Tokimatsu and b) 2
Seed (1987) correction factor was applied to the DC60 (N1)60
results given in Fig.4b to calculate (DC1)60; the standard 14
DPSH blow count corrected to overburden stress equal 1.5 SPT 20
to 1.0 atmospheric pressure and normalized to an 24
effective energy delivered to the rods at 60% of free fall 1
26
theoretical energy. The results at different overburden C4 27
pressures, shown in Fig. 4b, have collapsed into one line,
as shown in Fig. 6, confirming the validity of Tokimatsu 0.5
and Seed (1987) correction factor for DPSH test.
Therefore, if similar hammer system is used for both
0
tests, the following equation is valid:
0 5 10 15 20
p (t/m2 )
N30/ DC20 = (N1)60/ (DC1)60 (5)
Fig. 5 Determination of overburden pressure correction
Moreover, as (DC1)60 can reflect the soil compactness factor for DPSH results
(relative density), it can be concluded that (N1)60/
(DC1)60 decreases non-linearly as soil relative density
increases. 3.5

VALIDATION OF THE PROPOSED MODEL

P=5 5 t/m2
P = t/m2
3.0
P=10
P = 10 t/m2
t/m2
The validity of the developed DPSH-SPT correlation 2.5
P=15
P = 15 t/m2
t/m2
using GMDH model was tested using the five testing
(N1 )60 /(DC1 )60

sites in Egypt which were not used in developing the

2.0
proposed DPSH-SPT correlation. A typical comparison
between the model predictions and the field results is
1.5
shown in Fig. 7 where an acceptable agreement between
predicted and measured (N1)60 values can be observed.
The proposed model was also assessed using SPT 1.0
and DPSH penetration results from two different sites in
Auckland, New Zealand as shown in Figs. 8 and 9. The 0.5
stratigraphy at these sites consists mainly of a sandy soil
layer underlying a silty clay soil layer. The distance 0.0
between SPT and DPSH was between 3.0 to 5.0 m. The 0 10 20 30 40 50 60 70 80
torque necessary to turn the DPSH driving rods was (DC1)60
measured due to the presence of the upper cohesive soil
layer. Such measurements are usually used to assess the Fig. 6 DPSH results corrected for overburden pressure
using Tokimatsu and Seed (1987)
amount of skin friction developed on the driving rods.
 SC
-6
Dp80 Measured N60
Dp81
Predicted N60
On Use of Dynamic Dp83
Probing in Sandy Soils
Avrg

(NN
1)60
60
0 3 8 25 42 58

Very loose

Medium

Dense

Dense
Loose

Very
SPT (N30) DPSH (N20)
DC20
0 40 80 0 30 60
12
GL
10
GWL
8
SP-SM
6
Predicted
Elevation (m)

4 SC
Measured
2

0 SP

-2 SC

-4
SP-SM
-6
Dp99
Measured N60
Dp101
Dp103
Predicted N60
Avrg

Fig. 7 Comparison between predicted and measured (N 1)60 for atypical testing site in Egypt

No. Of
of blows
blows No.
No.of
Ofblows
blows
0 20 40 60 0 20 40 60
0 0
Silty Clay to clayey Silt

1 1

2 2

3 3
(m)
Depth (m)

Depth (m)
Depth

4 4

5 5
Silty Sand

6 6

7 7

8 8
DC20
DC20 Predicted(N1)60
Predicted (N 1 )60
SPT (N 30 )
N30 Measured(N1)60
Measured (N1 )60

Fig. 8 Comparison between predicted and measured (N1)60 for site no.1 in Auckland
 Abuel-Naga, et al.

No. Of blows Mv (Nm) (N1 )60

0 20 40 60 0 100 200 300 0 20 40 60
0.0 0.0 0.0

2.0 SPT (N30) 2.0 2.0 Measured

Predicted, no M v
4.0 4.0 4.0

Silty Clay
correction

DC20 Predicted,
Depth (m)

6.0 6.0 6.0

with M v
correction
(Eq.
(Eq. 5)
7)
8.0 8.0 8.0

10.0 10.0 10.0

Clayey Sand
12.0 12.0 12.0

14.0 14.0 14.0

2
Fig. 9 Comparison between predicted and measured (N1)60 for site no.2 in Auckland

No.Of
No. ofblows
blows No.
No. Of
of blows
0 10 20 30 0 20 40 60
0.0 0.0

1.0 1.0

2.0 2.0

3.0 3.0
Depth (m)

Silty Sand
Depth (m)

4.0 4.0

5.0 5.0

6.0 6.0

7.0 7.0

8.0 8.0
DC20
Series1 Predicted (N 1 )60
Series1
SPT (N 30 )
N30 Measured(N1)60
Measured (N 1 )60

Fig. 10 Comparison between predicted and measured (N1)60 for site in Marvao, Portugal (Duarte et al. 2004)
 On Use of Dynamic Probing in Sandy Soils

3.5 have suggested a correction for the effect of skin friction

P=5 5 t/m2
P =t/m2 as follows:
3.0
P=10 10 t/m2
P =t/m2 2M v e
2
N skin  (6)
2.5 P =t/m2
P=15 15 t/m DMo gh

Cearns McKenzie (1988)

2.0 where Nskin is the number of blows required to overcome
N30 /DC20

skin friction resistance, e is the standard depth increment,

1.5 D is rod diameter, Mo is hammer mass, h is hammer drop
height, Mv is torque measurement on rod. For DPSH this
formula gives Nskin= 0.0244Mv.
1.0

i.e. 41 Nm torque = 1 blow/20 cm (7)

0.5

Fig. 9 shows a comparison between (N1)60 field

0.0
measurements and the proposed model prediction
0 10 20 30 40 50 60 considering no torque correction, and with torque
DC20 correction according to Eq. (7). Better agreement can be
obtained with torque correction made according to Eq.
Fig. 11 Comparison between the proposed DPSH-SPT
(7). In fact the observed deviation between the predicted
correlation in this study and the correlation by Cearns
and McKenzie (1988) and the measured results through the silty clay layer can
be also explained in light of the fact that the data used in
the development of the proposed model in this study was
For the first site (Fig. 8), the torque measurements mainly from sandy soils and extrapolating the
were very low through the soft clay and the sandy layer modelworking range to include cohesive soils is not
(less than 10 Nm). Therefore the effect of developed skin recommended before investigating the effect of soil type
friction can be ignored. Such insignificant dynamic skin on the correlation between SPT and DPSH test results.
friction of soft clay soils was also reported by The proposed correlation in this study was also
Waschkowski (1982). Consequently, the measured tested using the field test results reported by Duarte et al.
DPSH penetration resistance at this site is totally due to (2004) for a sandy soil from Marvao in the south of
the cone penetration resistance and consequently no skin Portugal. The Marvao soil was classified as SM with
friction correction is required. Taking into consideration 84.72 % and 17.05% passing from sieve no. 4 (4.75 mm)
the possible effect of the spatial soil heterogeneity at the and 200 (0.075 mm), respectively. The distance between
testing points of DPSH and SPT, the comparison SPT and DPSH test location was between 0.5 and 1.0 m.
between the measured and predicted (N1)60 shown in Fig. 10 shows very good agreement between (N1)60 field
Fig. 8 can be considered reasonable from the practical measurements and the proposed model prediction.
view point. Finally, the soundness of the proposed DPSH-SPT
For the second site, the torque measurements show correlation can be also supported by comparing it with
different behaviour as it builds up through the upper clay the correlation proposed by Cearns and McKenzie
layer reaching to 200 Nm which is the maximum (1988) as shown in Fig. 11. Both correlations have
capacity of the utilized torque wrench as illustrated in approximately similar non-linear pattern in N30/DC20-
Fig. 9. However, as the sandy layer is not expected to DC20 plane. However, since Cearns and McKenzie
add significant skin friction to the DPSH driving rods, it (1988) did not consider the influence of effective
was assumed that the torque measurements through the overburden pressure and it wasn’t clear whether similar
sandy layer are equal to 200 Nm. In fact such high hammering system was used for both tests, their
torque measurements indicate that the measured DPSH correlation can be considered as a special case of the
penetration resistance is not purely due to cone general correlation proposed in this study.
penetration resistance and consequently skin friction
correction is required.
Based on the simplified assumption that average skin CONCLUSIONS
friction along the rod is the same when the rod is driven
down by the hammer, as it is when the rod is rotated and As Standard Penetration Test (SPT) is well
the torque is measured, Dahlberg and Bergdahl (1974) developed for the sandy soils, correlations between SPT
 Abuel-Naga, et al.

and other test results (SCPT, DCPT, etc.) are best the European symposium on penetration testing,
established for sands only. In this study, GMDH model Sttockholm, No. 2.2: 93-101.
was used to predict the SPT results from the results of DIN 4094 part 1. (1974). Dynamic and Statics
DPSH test for sandy soils. The salient conclusions that pentrometers, Dimensions of apparatus and method
can be drawn from this study are: of operation.DeutschesInstitu fur Normung e. V.
Berlin.
 GMDH approach is a useful tool for establishing a DIN 4094 part 2. (1980). Dynamic and Statics
correlation between SPT and DPSH test results. pentrometers, Application and
 The SPT overburden pressure correction factor can evaluation.DeutschesInstitu fur Normung e. V. Berlin.
also be used for DPSH test results. Dolenko, S. A., Orlov, Y. V. and Persiantsev, I.
 DPSH-SPT correlation is function of soil relative G.(1996). Practical implementation and use of Group
density and overburden pressure. Method of Data Handling (GMDH): Prospects and
problems. Proceedings of ACEDC’96, PEDC,
University of Plymouth, UK
REFERENCES Duarte, I. M. R., Pinho, A. B. and Ladeira, F.
L.(2004).Penetrometer testing in residual soils from
Ardalan, H., Eslami, A.and Nariman-Zadeh, N.(2009). granitic rocks in the south of Portugal. Geotechnical
Piles shaft capacity from CPT and CPTu data by and Geophysical Site Characterization, Vol. 2:1279-
polynomial neural networks and genetic algorithms. 1284.
Computers and Geotechnics. 36:616–625 Gabr, M. A., Hopkins, K., Coonse, J. and Hearne,
ASTM D1586-99: American Society of Testing T.(2000). DCP Criteria for performance evaluation of
Materials. Standard Test Method for Penetration Test pavement layer. Journal of performance of
and Split-Barrel Sampling of Soils, ASTM Constructed facilities ASCE. 14(4):141-148.
International, West Conshohocken, PA. Jayawickrama, P.W., Amarasiri, A. L. andRegino, P. E.,
ASTM D6951-03: American Society of Testing (2000).Use of dynamic cone penetrometer to control
Materials. Standard Test Method for Use of the compaction of granular fill. Trans. Res. Board, Trans.
Dynamic Cone Penetrometer in Shallow Pavement Res. Rec., No. 1736.
Applications, ASTM International, West Kalantary, F., Ardalan, H. and Nariman-Zadeh,
Conshohocken, PA. N.(2009).An investigation on the Su–NSPT
Bergdahl, U. and Eriksson, U.(1983). Soil properties correlation using GMDH type neural networks and
with Penetration Tests – A Literature study. Swedish genetic algorithms.Engineering Geology. 104:144–
Geotechnical Institute. Linkoping, Sweden, Report 155
No. 22. Kayalar A. S.(1988). Statistical evaluation of dynamic
BSI British Standard.Geotechnical investigation and cone penetration test data for design of shallow
testing. BSI, 2005, BS EN ISO 22476: Part 2. foundations in cohesionless soils. Proceeding of 1st
Butcher, A. P., McElmml, K. and Powell, J. J. M. International Symposium on Penetration Testing,
(1996).Dynamic probing and its use in clay soils. Orlando, Vol.1: 429- 434.
Proceedings of the international conference on Madala, H. R. and Ivakhnenko, A. G.(1994).Inductive
Advances in site investigation practice, (Craig C. Learning Algorithm for Complex Systems Modeling,
(eds)). Thomas Telford, London: 383-395. CRC Press Inc., Boca Raton.
Cearns, P. J. and Mckenzie, A. (1988).Application of McGregor, J.A. and Duncan, J.M.(1998).Performance
dynamic cone penetrometer testing in East Anglia. and Use of the Standard Penetration Test in
Proceeding of Geotechnology Conference on Geotechnical Engineering Practice.Center for
Penetration Testing In the UK, Thomas Telford, Geotechnical Practice and Research, Virginia
London: 123-127. Polytechnic Institute, Blacksburg, VA.
Chen, D-H., Lin, D-F., Liau, P-H.andBilyeu, J. A. NeuroShell II Manual (1996), Neural Network Computer
(2005). Correlation between dynamic cone Program, Ward Systems Group, Inc., Frederick, Mass,
penetrometer values and pavement layer USA.
moduli.Geotechnical Testing Journal ASTM. Robertson, P. K.(1990). Soil classification using the
28(1):42-49. cone penetration test.Canadian Geotechnical Journal.
Dahlberg, R. and Bergdahl, U.(1974). Investigations of 27: 151-158.
the Swedish ram sounding method. Proceedings of Sanglerat G. (1972). The Penetrometer and Soil
Exploration. Elsevier, Amsterdam.
 On Use of Dynamic Probing in Sandy Soils

Spagnoli G.(2007). An empirical correlation between Tokimatsu, K. and Seed, H.B.(1987). Evaluation of
different dynamic penetrometers. The Electronic Settlementsin SandsDue to Earthquake
Journal of Geotechnical Engineering, 12, Bundle C. Shaking.Journal of Geotechnical Engineering ASCE.
(http://www.ejge.com/2007/Ppr0729/Abs0729.htm) 113:861-878.
Stefanoff G., Sanglerat G., Bergdahl U. and Melzer K- Waschkowski, E.(1982). Dynamic probing and site
J.(1988). Dynamic probing (DP): International investigation. Proceeding of the Second European
reference test procedure. Proceeding of 1st Symposium on Penetration Testing, (Verruijt et al.
International Symposium on Penetration Testing, (eds)). A.A. Balkema/ Rotterdam, 1, 363-368.
Orlando, 1:53- 70.

View publication stats