You are on page 1of 12

- 2727 -

Universal Correlation of Shear Wave


Velocity and Standard Penetration
Resistance
Aminaton Marto
Professor
Faculty of Civil Engineering, Universiti Teknologi Malaysia,
81310 UTM Johor Bahru, Malaysia
e-mail: aminaton@utm.my
Choy Soon Tan
PhD Candidate
Faculty of Civil Engineering, Universiti Teknologi Malaysia,
81310 UTM Johor Bahru, Malaysia
e-mail: cstan8@live.utm.my
Tiong Kung Leong
PhD Candidate
Faculty of Management and Human Resource Development,
Universiti Teknologi Malaysia, 81310 UTM Johor Bahru, Malaysia
e-mail: kung_leong83@hotmail.com


ABSTRACT
Shear wave velocity is one of the important characteristics in seismic microzonation. Most of
the researchers attempted to quantify the relation between shear wave velocity and standard
penetration resistance at local scale for various countries in past few decades, others tried to
correlate in terms of various soil indexes including depth, geological epoxy and etc. Although
these empirical correlations are convincing with notable worldwide agreement, it is region
specific and not applicable to other site. The aim of this paper is to review the existing
empirical correlations and to generate universal correlation by using statistical simulation.
Statistical simulations are described in this paper and illustrated by using 60 existing
correlations which had been established worldwide. New correlations are proposed by
combining published correlations in two ways; (i) without outliner and (ii) with outliner. The
correlation with outliner data gives higher correlation coefficient and lower percentage of
error. Results show that soil type is influencing the correlation pattern. Correlation for
cohesive soils is higher than correlation for all other soil types, and followed by correlation
for cohesionless soil. This coincides with most of the researchers findings. This study
provides an advancement to incorporate all previous research findings and generate
correlations that can be applied to the sites globally.
KEYWORDS: All soils, cohesionless soils, cohesive soils
Vol. 18 [2013], Bund. M 2728

INTRODUCTION
Shear wave velocity (Vs) of near ground surface is one of the important fundamental soil
characteristics for both earthquake and geotechnical engineering. The most devastating damages
and casualties are usually caused by ground shaking and enlarged by the site effect; The 1985
Mexico Earthquake is one of the classical examples. Hence, shear wave have become key
controller in ground response, soil-structure interaction and wave amplification. Shear wave for
the top 30m also serve as key indicator in classifying site class for both building codes and design
standard. Consideration impotency of this issue, the necessity to describe subsurface
characteristics and geology stratifications through shear wave velocity is essential.
Seismic microzonation studies have been investigated globally especially the earthquake
prone countries. The determination of soil elasticity through in situ measurement has been
developed worldwide. Due to lacking of both financial and respective expertise, on-site
investigation on Vs is unfeasible in engineering practice. It is not surprising that uses of direct
correlation of SPT-N to various soil indexes are common practice world-wide. Therefore,
formulation of empirical relations in estimating the Vs through various soil indexes is proposed.
Researchers tried to establish the relationship based on soil type, depth, overburden pressure,
geological age and etc. Many regression equations are available literately; either interprets using
single soil index or multivariable analysis. The investigation to reach concrete conclusion is still
on the move. Among variety of soil indexes, the standard penetration tests resistance (SPT-N) and
depth factor are the most preferable soil indexes (Kuo et al, 2011).
Researchers had only developed their correlation at local scale in various countries. Site-
dependant studies incorporate with local geologic and soil conditions in analysis. Therefore, past
studies are independent studies without any firm theoretical formulation to support with.
Nevertheless, only limited studies try to summarize those independent studies for comparison and
verification purpose. There is no initiative study to link these correlations and comprehensively
investigate the general pattern of these empirical regressions based on soil characteristic. In this
regards, this paper intends to refine this subject to a wider context through reformulation and
reinterpretation of previous published correlations. There are two major parts in this paper. This
paper will deliver an overview of previous correlations first then establish universal correlation
using statistical simulation in second part.
EMPIRICAL REGRESSION OF SHEAR WAVE VELOCITY
Empirical regression to compute shear wave velocity had been studied extensively. Different
researchers tried to correlate Vs in terms of various soil indexes including depth, geological
epoxy and etc. There are some opinions suggested to use either one term alone or to combine few
dependent variables in describing the relationship. The following paragraphs will discuss on the
influences of each variant.
Most researchers prefer to describe Vs in term of SPT-N alone (eg: Dikmen, 2009). The
conflicting point is whether should SPT-N to be energy corrected. In general, majority
researchers (eg: Anbazhagan et al, 2013) makes use of uncorrected SPT-N while only limited
researchers (eg: Anbazhagan and Sitharam, 2008) prefer to consider energy corrected SPT-N in
formulating their relationship. Some researchers had examined the performance for both
corrected and uncorrected SPT-N but different outcomes were obtained. There are few
researchers (eg: Hasancebi and Ulusay, 2007) hold similar viewpoint that uncorrected SPT-N
Vol. 18 [2013], Bund. M 2729

gives better fit correlations. Some other researchers (eg: Uma Maheswari et al, 2010) show that
both corrected and uncorrected SPT-N give similar coefficient of correlation.
Depth factor is another favourable soil indexes besides SPT-N value. Researchers (eg:
Chapman et al, 2006) examined depth factor in development of empirical equation. Akin et al
(2011) proved the necessity of depth effect in providing better fit correlations while Brandenberg
(2010) highlighted the importance of overburden pressure effects. Although depth factor is
corresponding with the overburden pressure (Holzer et al, 2005), both Vs and SPT-N are
theoretically independent of overburden pressure effect. In most of the cases, overburden effect is
neglected by researchers. In addition, the inclusion of effective overburden pressure weakens the
spatial relationship between Vs and SPT-N (Lee and Tsai, 2008). Some geologist stated that
influence of geological age is considerable, but it is indirectly represented by the variation of
blow count; hence it is negligible.
The influence of soil types has attracted researchers interest. In general, researchers
differentiate correlations for cohesive and cohesionless soils. However, contradict findings exist
literally. For example, there were two different soil sites in Greece, Tsiambaos and Sabatakakis
(2011) stated that shear wave velocity of sandy soils is lower than cohesive soils. Conversely, the
finding of Pitilakis et al (1999) is totally different. In fact, these correlations may not be
generally and extensively used because soil deposit is a natural product which comprises of
various compositions. It is nonlinear and inseparable during boring procedure. Correlations of all
soil types will have greater coefficient of determination compare to individual soil types (Sun et
al, 2012). When the soil types are separating to test, result of Pitilakis et al (1992) shows
extraordinary curve and opp general trends that cohesiveness soils are having higher Vs values.
Some researchers believed that the combining effects of few soil indexes will give better
correlations. Ohta and Goto (1978) presented the very first multivariable analysis to develop 15
sets of empirical regression with different combination of soil type, geological epoch, depth and
SPT-N. Results reported that correlation coefficient is the highest when all four factors are
considered. Other researchers (eg: Lee, 1992) included only SPT-N and depth factor, the two
favourable factor among researchers in their analysis. Besides SPT-N and soil types, Iyisan
(1996) does also compare the variables such as mean grain size, effective overburden stress.
It is worth to mention the most pressing problem is that all these studies are site dependant.
Researchers only tried to obtain their comparative correlations without appropriate verification
process. Anbazhagan et al (2012) noticed the urgency to establish new correlation that is
applicable to any region. However, the study is focus on correlations between SPT-N and shear
modulus. There is lack of study to link these correlations of Vs and SPT-N to establish a general
pattern based on the soil characteristic.
EXISTING CORRELATIONS OF SHEAR WAVE
VELOCITY AND STANDARD PENETRATION
RESISTANCE
After considering the overview of earlier published correlations, it is reasonable to exclude
other soil characteristic indexes and include only one variant, the SPT-N in this initial study of
universal correlation development. This paper will only concentrate on the most common and
basic equation to correlate Vs and SPT while multivariable analysis is recommended to carry out
in near future. It is because the most of the studies predict the relation by using a power-law
Vol. 18 [2013], Bund. M 2730

relationship (Vs =AN
B
).There are some opinions proposed a liner relation by hatching the dataset
(Kayabali, 1996). In order to increase the concentration in statistical simulation, it seems
reasonable to employ the most favorable relation in this study. This could enhance the accuracy
during formulation of predictive relation.
Moreover there are several researchers including J afari et al (2002); Hanumantharao and
Ramana (2008); Uma Maheswari et al (2010); Kuo et al (2011); Akin et al (2011) had initiated to
summarise the available correlations in table format for comparative study.
There are total 60 established correlations which are developed globally within past 50 years
have been selected and summarised in Table 1. The published regression was divided into three
groups according to soil types (all soil, cohesionless soil and cohesive soil). These independent
correlations are established for various sites exploration from the world including J apan, USA,
Greece, Taiwan, Turkey, India, Iran, Korea, and etc. Nevertheless, most of the correlations are
comparable and in good trend. Although most of the correlations have similar trend, direct
application to other regions are disallowed because of different practice of SPT and Vs
investigation works. Besides, the variations in geotechnical engineering characteristics of studied
sites are significantly contributing to the difference.
STATISTICAL ANALYSIS
There are total 60 correlations of shear wave velocity and standard penetration resistance,
which are globally available, are used to formulate new universal correlation. This study will only
focus on correlations in power-law form, as agreed by majority of researchers as the most
preferable relationship. Hence, the data are homogenous with consistence trend. It is important to
highlight that these correlations are not applicable for SPT-N above 50 (Hanumantharao and
Ramana, 2008).
By combining the published data using statistical simulation, it is thus aimed to establish new
universal correlations which are applicable to all regions. This paper is extending the findings of a
previous study (Marto et al, 2013). The analysis will not only concentrate on the correlations
which are applicable for all soil types, but it also included correlations for both cohesionless and
cohesive soils for comparison.
Simple linear power regression analysis is carried out to develop the relationship statistically.
By assuming the dataset distribution is normally distributed; only the data located within the
range between mean standard deviation are used to develop the general correlations. In other
words, 68.2% of the dataset from published correlations used by researchers are considered. This
outliner boundary of mean standard deviation will also serve as the boundary limit to guide
future investigator in their validation process.
Coefficient of determination (R-squared) value and percentage of Vs error (PVE) are
determined to indicate the accuracy of the correlations. PVE is calculated using the following
modified relation:
Peicentage of vs Eiioi =
vs
cclculctcd
- vs
mccn
vs
mccn
(1)



Vol. 18 [2013], Bund. M 2731

Table 1: Correlation of Vs =AN
B
(after Marto et al, 2013)
Year Researcher All soil Cohesionless soil Cohesive soil
1966 Kanai Vs =19N
0.6
- -
1970 Ohba and Toriumi Vs =84N
0.31
- -
Shibata Vs =32N
0.5
-
Imai and Yahimura Vs =76N
0.33
-
1972 Ohta et al Vs =87N
0.36
-
Fujimara Vs =92.1N
0.337
-
1973 Ohsaki and Iwasaki Vs =81.4N
0.39
Vs =59.4N
0.47
-
1975 Imai and Yoshimura Vs =92N
0.329
- -
Imai et al Vs =89.9N
0.341
- -
1977 Imai Vs =91N
0.337
Vs =80.6N
0.331


Vs =102N
0.292

1978 Ohta and Goto Vs =85.35N
0.348
Vs =88N
0.34

1980 J RA Vs =80N
0.33


Vs =100N
0.33

1981 Seed and Idriss Vs =61.4N
0.5
- -
1982 Imai and Tonouchi Vs =97N
0.314
- -
1983 Seed et al - Vs =56.4N
0.5
-
Sykora and Stokoe - Vs =100.5N
0.29
-
1989 Okamoto et al - Vs =125N
0.3
-
1990 Lee - Vs =57.4N
0.49


Vs =114.43N
0.31

Imai and Yoshimura Vs =76N
0.33
- -
1991 Yokota et al Vs =121N
0.27
- -
1992 Kalteziotis et al Vs =76.2N
0.24
Vs =49.1N
0.50
Vs =76.6N
0.45

1995 Raptakis et al Vs =100N
0.24
Vs =184.2N
0.17

Athanasopoulos Vs =107.6N
0.36
- -
Sisman Vs =32.8N
0.51
- -
1996 Iyisan Vs =51.5N
0.516
- -
1997 J afari et al Vs =22N
0.85
- -
2000 Chien et al - Vs =22N
0.76
-
2001 Kiku et al. Vs =68.3N
0.292
- -
2002 J afari et al Vs =22N
0.85
Vs =19N
0.85
Vs =27N
0.73

2007 Hasancebi and Ulusay Vs =90N
0.309
Vs =90.82N
0.319
Vs =97.89N
0.269

2008 Hanumantharao and Ramana Vs =82.6N
0.43
Vs =79N
0.434

2008 Lee and Tsai Vs =137.153N
0.229
Vs =98.07N
0.305
Vs =163.15N
0.192

2009 Dikmen Vs =58N
0.39
Vs =73N
0.33
Vs =44N
0.48

2010 Brandenberg et al
Uma Maheswari et al. Vs =95.64N
0.301
Vs =100.53N
0.265
Vs =89.31N
0.358

2011 Tsiambaos and Sabatakakis Vs =105.7N
0.327
Vs =79.7N
0.365
Vs =88.8N
0.370

2012 Anbazhagan et al Vs =68.96N
0.51
Vs =60.17N
0.56
Vs =106.63N
0.39

RESULTS AND DISCUSSION
In Figure 1-3, the dataset of published correlations have been summarised in Table 1 and it
show three soil classifications: all soils, cohesive and cohesionless soils. The new developed
correlations for dataset before outliner process are having R-squared values which are ranging
from 0.6 to 0.8. Cohesionless soil is having the lowest Vs value compare with cohesive soils, but
it is similar to all soils. In order to establish boundary limit to guide future investigator in their
validation process, outliner boundary of mean standard deviation will be used.
Vol. 18 [2013], Bund. M 2732


Figure 1: Proposed correlation (without outliner) for all soil types


Figure 2: Proposed correlation (without outliner) for all cohesionless soil
Vs =69.76N
0.401
R =0.624
0
100
200
300
400
500
600
700
0 10 20 30 40 50 60
S
h
e
a
r

W
a
v
e

V
e
l
o
c
i
t
y

(
m
/
s
)
SPT-N Value
Vs =66.56N
0.420
R =0.752
0
100
200
300
400
500
600
0 10 20 30 40 50 60
S
h
e
a
r

W
a
v
e

v
e
l
o
c
i
t
y

(
m
/
s
)
SPTNValue
Vol. 18 [2013], Bund. M 2733


Figure 3: Proposed correlation (without outliner) for all cohesive soil
In Figure 4-6, the developed correlations for dataset with outliner boundaries and
corresponded R-squared values are shown. The upper boundary and lower boundary lines are
represented by the mean standard deviation of dataset. Besides, it could mark as the boundary
range of the regression correlations which no researcher has started to consider so. The R-squared
values for all three soil classifications are improved, compared with dataset without outliner
boundaries. It can be observed in Figure 7 that the PVE is less than 6% for all points, although
geotechnical and site investigation varied. This indicates that the predictive Vs are very close to
Vs mean, thus showing good agreement of the established correlations.

Figure 4: Proposed correlation (with outliner) for all soil types
Vs =87.72N
0.361
R =0.678
0
100
200
300
400
500
600
0 10 20 30 40 50 60
S
h
e
a
r

W
a
v
e

V
e
l
o
c
i
t
y

(
m
/
s
)
SPT-N Value
0
100
200
300
400
500
600
700
0 10 20 30 40 50 60
S
h
e
a
r

W
a
v
e

V
e
l
o
c
i
t
y

(
m
/
s
)

SPT-N Value
Vs=93.67N
0.389
Vs =77.13N
0.377
R =0.876
Vs =53.87N
0.403
Vol. 18 [2013], Bund. M 2734


Figure 5: Proposed correlation (with outliner) for all cohesionless soil


Figure 6: Proposed correlation (with outliner) for all cohesive soil

All generated correlations from this study are shown in Table 2, including both dataset with
and without outliner. The correlation with outliner data gives higher correlation coefficient and
lower percentage of error. The soil type is influencing the correlation pattern. Results show that
Vs of sandy soils are lower compared with cohesive soils (Figure 8), which are coincided with
most of the researchers findings. However, because of the soil non-linearity, correlations which
are applicable to all soil types are most user-friendly and more economically to be generated. This
0
100
200
300
400
500
600
0 10 20 30 40 50 60
S
h
e
a
r

W
a
v
e

V
e
l
o
c
i
t
y

(
m
/
s
)
SPT-N Value
Vs=89.18N
0.390
Vs =75.05N
0.388
R =0.911
Vs =50.79N
0.444
0
100
200
300
400
500
600
0 10 20 30 40 50 60
S
h
e
a
r

W
a
v
e

V
e
l
o
c
i
t
y

(
m
/
s
)
SPT-N Value
Vs=128.8N
0.303
Vs =91.87N
0.361
R =0.937
Vs =59.55N
0.420
Vol. 18 [2013], Bund. M 2735

is an advancement step in incorporating earlier research findings to draft an initiate boundary on
empirical correlation between Vs and SPT-N.

Figure 7: Percentage of error (PVE) for all proposed correlation

Table 2: Summary of the results
Dataset Correlations
Correlation
Coefficient
Without
outliner
All Soil Types Vs =69.79N
0.401
0.624
Cohesionless Soil Types Vs =66.56N
0.420
0.752
Cohesive Soil Type Vs =87.72N
0.361
0.678
With
outliner
All Soil Types
Vs =53.87N
0.403
-
Vs =77.13N
0.377
0.876
Vs =93.67N
0.389
-
Cohesionless Soil Types
Vs =50.76N
0.444
-
Vs =75.05N
0.388
0.911
Vs =89.18N
0.390
-
Cohesive Soil Type
Vs =59.55N
0.420
-
Vs =91.87N
0.361
0.937
Vs =128.8N
0.303
-
0%
2%
4%
6%
8%
10%
0 10 20 30 40 50
P
e
r
c
e
n
t
a
g
e

o
f

V
s

e
r
r
o
r


(
%
)
SPT-N Value
PVE (all) PVE (cohesionless) PVE (cohesive)
Vol. 18 [2013], Bund. M 2736


Figure 8: Comparison between cohesionless and cohesive soils correlations
CONCLUSION
This paper review existing correlations of shear wave velocity and standard penetration
resistance. Then universal correlations are established by incorporating all earlier research
findings. There are some conclusions have been listed as below:
1. The correlation with outliner data gives higher correlation coefficient and lower
percentage of error.
2. At constant SPT-N, cohesive soils have greater Vs value than all soils followed by
cohesionless soils. This kind of behaviour is an general agreement. This trend also
justifies the applicability of correlations of all soil types, which is laying between
cohesive and cohesionless soils.
3. Considering soil nonlinearity- correlations which are applicable to all soil types are the
most user-friendly and economical.
4. New correlations presented in this paper must be applied together with proper
engineering judgement.
5. The initial purpose of this paper is to generate universal correlations based on SPT-N and
soil types sorely. Multivariable analysis which included other soil parameters should be
investigated in future to promote the usability of these general correlations.
6. Further investigations which include the influences of different shear wave investigation
method are warranted for verification and validation purposes.
ACKNOWLEDGEMENT
The authors gratefully acknowledge the financial supports by Universiti Teknologi Malaysia
through Research University Grant (RUG) and UTM Zamalah program.
0
50
100
150
200
250
300
350
400
0 10 20 30 40 50 60
S
h
e
a
r

W
a
v
e

V
e
l
o
c
i
t
y

(
m
/
s
)

SPTNValue
Sand
Clay
Vol. 18 [2013], Bund. M 2737

REFERENCES
1. Akin, M., Kramer, S., and Topal, T. (2011) Empirical correlations of shear wave
velocity (Vs) and penetration resistance (SPT-N) for different soils in an earthquake-
prone area (Erbaa-Turkey), Engineering Geology, 119, 1-17.
2. Anbazhagan, P., and Sitharam, T. (2008) Mapping of average shear wave velocity
for Bangalore Region: a case study, J ournal of Environmental and Engineering
geophysics, 13(2), 69-94.
3. Anbazhagan, P., Parihar, A., and Rashimi, H.N. (2012) Review of correlations
between SPT N and shear modulus: A new correlation applicable to any region, Soil
Dynamics and Earthquake Engineering 36, 52-69.
4. Anbazhagan, P., Kumar, A. and Sitharam, T. (2013) Seismic site classification and
correlation between standard penetration test N value and shear wave velocity for
Lucknow City Indo-Gangetic Basin, Pure and Applied Geophysics, 170 (3), 299-
318.
5. Brandenberg, S., Bellana, N., and Shantz., T.(2010) Shear wave velocity as function
of standard penetration test and vertical effective stress at California Bridge Sites,
Soil Dynamics and Earthquake Engineering 30, 10261035.
6. Chapman, M.C., Martin, J .R. Olgun C.G., and Beale, J .N. (2006) Site-response
models for Charleston, South Caroline, and vicinity developed from shallow
geotechnical investigations, Bulletin of Seismological Society of America, 96(2),
467-489.
7. Dikmen, U. (2009) Statistical correlations of shear wave velocity and penetration
resistance for soils, J ournal of Geophysics and Engineering, 6(1), 6172.
8. Hamumantharao, C. and Ramana, G. (2008) Dynamic soil properties for
microzonation of Delhi, India, J ournal of Earth System Science, 117 (S2), 719-730.
9. Hasancebi, N. and Ulusay, R. (2007) Empirical correlations between shear wave
velocity and penetration resistance for ground shaking assessments, Bulletin of
Engineering Geology and the Environment, 66(2), 203-213.
10. Holzer, T.L., Bennett, M.J ., Noce T.E. and Tinsley III, J .C. (2005) Shear-wave
velocity of surficial geologic sediments in Northern California: statistical
distributions and depth dependence, Earthquake Spectra, 21(1), 161-177.
11. Iyisan, R. (1996) Correlations between shear wave velocity and in-situ penetration
test results, Digest, 371-374.
12. J afari, M, Shafiee, A. and Ramzkhah, A. (2002) Dynamic properties of the fine
grained soils in South of Tehran, J ournal of Seismology and Earthquake
Engineering, 4(1), 2535.
13. Kayanali, K. (1996) Soil liquefaction evaluation using shear wave velocity,
Engineering Geology, 44, 121-127.
14. Kuo, C.H.., Kuo, K.W., Hsieh, H.H., Chang, T.M., Wen K.L. and Chen, C.T. (2011)
Evaluating empirical regression equations for Vs and estimating Vs 30 in
northeastern Taiwan, Soil Dynamics and Earthquake Engineering, 31, 431-439.
Vol. 18 [2013], Bund. M 2738

15. Lee, S.H.H. (1992) Analysis of the multicollinearity of regression equations of shear
wave velocities, Soils and Foundations, 32(1), 205-214.
16. Lee, C.T. and Tsai, B.R. (2008) Mapping Vs30 in Taiwan, Terrestrial, Atmospheric
and Oceanic Sciences, 19(6), 671682.
17. Marto, A. Tan, C.S., Kasim, F., and Suhatril, M. (2013) A correlation of shear wave
velocity and standard penetration resistance. Electronic J ournal of Geotechnical
Engineering, 18c, 463-471.
18. Ohta, Y. and Goto, N. (1978) Empirical shear wave velocity equations in terms of
characteristic soil indexes, Earthqquake engineering and structural dynamics 6, 167-
187.
19. Pitilakis, K., Anastassiadis, A., and Raptakis, D. (1992) Field laboratory
determination of dynamic properties of natural soil deposits, Earthquake
Engineering Tenth World Conference, Balkema, Rotterdam.
20. Pitilakis, K., Raptakis, D., Lontzetidis, K., Tika-vassilikou, T.H. and J ongmans, D.
(1999) Geotechnical and geophysical description of Euro-seistest, using field,
laboratory tests and moderate strong motion recordings, J ournal of Earthquake
Engineering 3(3), 381-409.
21. Sun, C-G., Cho, C-S., Son, M. and Shin J .S. (2012) Correlations between shear
wave velocity and in-situ penetration test results for Korean soil deposits, Pure and
Applied Geophysics, 170(3), 271-281.
22. Tsiambaos, G. and Sabatakakis N. (2011) Empirical estimation of shear wave
velocity from in situ tests on soil formations in Greece, Bulletin of Engineering
Geology and the Environment, 70:291297.
23. Uma Maheswari, R., Boominathan, A., Dodagoudar, GR. (2010) Use of surface
waves in statistical correlations of shear wave velocity and penetration resistance of
Chennai soils, Geotechnical and Geology Engineering, 28(2):119-137.
2013 ejge

You might also like