Professional Documents
Culture Documents
predict angle of friction and relative density
predict angle of friction and relative density
net/publication/316327813
CITATIONS READS
35 2,045
4 authors, including:
Nagaratnam Sivakugan
James Cook University
230 PUBLICATIONS 5,688 CITATIONS
SEE PROFILE
All content following this page was uploaded by Nagaratnam Sivakugan on 01 September 2023.
··································································································································································································································
Abstract
This study is an attempt to evaluate relative density and friction angle of sands on the basis of SPT-N values. In order to develop the
relationships among relative density (Dr), friction angle (φ ) and SPT-N value, field and laboratory test results from sixty boreholes
executed in sandy deposit were used. The field tests include the SPTs conducted in the boreholes and the determination of in-situ
density at various depths in the boreholes using the pitcher sampler whereas the laboratory tests include routine classification, direct
shear box, maximum and minimum density and specific gravity tests. The SPT-N values were observed to vary between 4 and 100
and the in-situ bulk density of undisturbed samples as recovered through pitcher sampler are in the range of 13.24 ~ 18.44 kN/m . 3
The soil samples are classified as poorly graded sand (SP), poorly graded sand with silt (SP-SM) and silty sand (SM) on the basis of
Unified Soil Classification System. The values of minimum dry unit weight fall in the range of 12.22~14.95 kN/m and maximum 3
dry unit weight varies in the range of 14.64~19.17 kN/m as obtained through vibratory table. Based on the test results, correlation
3
analysis was performed to identify the parameters that affect relative density. The parameters include SPT-N value corrected for field
procedures (N60), relative density (Dr), void ratio range ( emax –emin ), effective overburden stresses ( σv ′ ) and in-situ dry density (γdf).
Based on the results of correlation analyses, it was observed that SPT-N values alone cannot be related to Dr. Therefore, multiple
regression analysis was carried out using Statistical Package for the Social Sciences (SPSS) software and relation between relative
density, corrected SPT-N value and effective overburden stress is being proposed. The variation between experimental and predicted
values falls within ± 10% at 95% confidence interval. Validation of the proposed correlation was also performed by using an
independent data set which indicated that the prediction by using the proposed correlation also falls within ± 10%. Further,
comparison of the proposed correlation with other similar relationships already available in the literature was also performed. In
addition to the above, correlation between φ and SPT-N has also been proposed. It has been observed that the experimental and
60
predicted values of friction angle fall within ± 10% with 95% confidence interval. The proposed correlations may be very useful in
the field of geotechnical engineering during feasibility/preliminary design stage for rapid estimation of relative density or friction
angle based on the field SPT-N values. 60
Keywords: relative density, void ratio range, effective overburden stress, friction angle, standard penetration test, mean grain size
··································································································································································································································
*Associate Professor, Department of Civil Engineering, University of Engineering & Technology, Lahore, Pakistan (Corresponding Author, E-mail: hassan-
mujtaba@uet.edu.pk)
**Professor, Department of Civil Engineering, University of Engineering & Technology, Lahore, Pakistan (E-mail: kfch@uet.edu.pk)
***Associate Professor, College of Science, Technology & Engineering, James Cook university , Townsville , QLD 4811, Australia (E-mail: siva.siv-
akugan@jcu.edu.au)
****Dean Emeritus, California State University, 2689 Chateau Clermont Street, Henderson, Nevada 89044, USA (E-mail: brajamdas@gmail.com)
−1−
Hassan Mujtaba, Khalid Farooq, Nagaratnam Sivakugan, and Braja M. Das
r 60 v
angle, are obtained through the regression analysis. Mahmoud
Cubrinovski and Ishihara (1999) proposed a correlation (2013) correlated shear strength parameters (c and φ) for hard and
between SPT-N blow count and relative density (Dr) of granular dry silty clay with sand soil with corrected standard penetration test
soil in the form given in Eq. (6) blow count (SPT-N) and presented empirical equations.
Rogers (2006) considered subsurface exploration using standard
0.06
N F (0.23 + ) 1.7
penetration test and cone penetrometer test and observed that
D 98 (6)
Dr (%) = [ 50
( )] × 100 each of the two tests has certain advantages over the other. SPT
9 σv'
allows a firsthand look at subsurface material (CPT does not)
σv ′ is effective vertical overburden stress and D50 is the median and can provide crucial information regarding the type of
grain size and NF is the field SPT-N value. Kibria and Masood subsurface material especially cohesionless material containing
(1998) proposed correlations relating SPT blow count, Dr and φ fines while CPT is capable of detecting discrete horizon that
for Indus sands at Chashma based on the geotechnical investigations would normally be missed during drive sampling in SPT. Thus,
conducted for Chashma Hydropower Project. They proposed employment of both tests simultaneously has the greatest
that Dr of fine and silty sands (percent passing US # 200, F200 ≤ potential for correct site characterization. Sivrikaya and Togrol
15%) can be predicted from the SPT-N blow count using the (2006) correlated SPT-N blow count and undrained shear
following correlations for Indus sands and is presented in Eq. (7) strength using statistical approach and concluded that penetration
test is found to be sufficient reliable for assessment of undrained
Dr = 16.5 (N 1 ) 60 (7)
strength. Farooq et al. (2010) collected SPT test data for various
All seven empirical correlations relating Dr and SPT-N blow sites in Lahore, Pakistan and based on this data geotechnical
count suggest that N-value depends on the effective overburden zoning of Lahore into different zones has been proposed.
stress and relative density, and that N-value is proportional to the Evaluation of allowable bearing capacity for each zone is also
square of the relative density. proposed based on the data available. Hayat (2003) developed a
Friction angle of granular soils has also been correlated to SPT- geotechnical zonation map of Pakistan based on geotechnical
N value. Hatanaka and Uchida (1996) presented relation between investigation data and geology of different zones of the country.
friction angle (φ) and SPT-N value given in Eq. (8)
φ = (20 N ) + 20
0.5 2. In-situ and Laboratory Tests
1( 60 ) (8)
The (N1)60 in Eq. (8) is SPT-N value normalized to 95.76 kN/ Extensive geotechnical investigations have been carried out at
a site near Mianwali for a Hydropower Project of National drop the SPT hammer. Rope cathead lift system consisted of a
importance in Pakistan. These investigations ranged from deep rotating drum around which the operator wrapped Manila rope to
drilling, soil sampling and performing variety of in-situ penetration lift and drop the hammer by successively tightening and loosening
tests along with laboratory testing to characterize the engineering the rope turns around the drum. 1-3/4 rope turns was used for
attributes of sandy soil deposits. Depth of boreholes varies counterclockwise rotation. Split spoon sampler was used to
between 30 and 100 m and total number of boring was 60. obtain a representative disturbed soil sample for identification
Standard Penetration Test (SPT) was carried out in the boreholes purposes and measure the resistance of the soil to penetration of
following ASTM D1586. The test was performed after every 1.0 sampler. After the borehole has been drilled to the desired depth
m interval in the borehole which is routine local practice being and excessive cuttings have been removed, the Standard Penetration
observed in Pakistan. Pitcher samples were collected at every 2.0 Test (SPT) was performed following ASTM D1586 standard.
m interval up to 20.0 m depth and at 5.0 m interval below 20 m Cutting bit was removed from the bottom of the drill rod and
depth. The interval of sampling along the depth is in line as per split spoon sampler was attached and lowered into the bore hole.
the recommendations by Tomlinson (1986). Grain size distribution, Care was taken not to drop the sampler onto the soil to be
specific gravity, maximum/minimum density and direct shear sampled. Anvil was attached to the top of the drilling rod and
tests were performed on these samples in the laboratory. The SPT hammer was lifted up using the Manila rope having not
large data bank obtained from the geotechnical investigation has more than 2 turn around the cathead. The exposed portion of the
been analyzed during this research to develop correlation drilling rods was marked with chalk in three successive 0.15 m
between SPT-N value, relative density and friction angle for (6 in) intervals so that the advance of the sampler under the
sandy soil deposits. impact of the hammer can be easily observed for each 0.15 m (6
in) increment. The sampler was driven into the ground by freely
2.1 Standard Penetration Test (SPT) Procedure and Aver- dropping 623 N (140 lbs) hammer from a height of 0.762 m (30
age Subsurface Profile in). The number of blows required for driving the sample
Straight rotary drilling rig was used for subsurface exploration. through three 0.15 m (6 in) interval was recorded. The sum of
The rig was properly anchored with the ground to minimize blows required for driving last two 0.15 m (6 in) interval is
vibration during drilling and Standard Penetration Test (SPT) referred to as SPT-N value and blows for first 0.15 m (6 in)
performance. Bentonite slurry was used for stabilization of penetration is considered as seating drive. In total, sixty (60)
boreholes. Drill rods having inner diameter of 28.5 mm and outer boreholes were drilled following the above mentioned procedure
diameter of 41.2 mm and length of 3-4 m were used. Bottom of and the SPT tests were conducted at respective depths. The
the drill rod was attached with the drill bit to drill the borehole subsurface profile of the above mentioned site based on the data
having diameter of 10.16 cm (4 in). Side discharge bits were used from 60 boreholes and the average SPT-N values along the depth
instead of bottom discharge bits to minimize the disturbance of the are shown in Fig. 1. The subsoil in general, consists of medium
underlying soils. Rope cathead lift system was used to lift and to fine sand with fines up to 45%. The Ground Water Table
Fig. 1. Average Subsurface Profile and Variation of SPT-N at the Site area up to 100 m Depth
dense state. Refusal of SPT test at any depth is represented in-situ dry density ( γdf ) . The results of the correlation analysis
by SPT-N value of 100 in Fig. 1 and Fig. 2. More specifi- between N60 and relative density values (Dr) revealed that these
cally based on SPT-N values the strata is classified as very two parameters are directly proportional to each other. However,
loose to loose for SPT-N values vary between 0~10, while it the coefficient of determination (R2) is 0.60 and correlation
is regarded as medium dense for SPT-N values falling between coefficient (R) is 0.77. These two are obtained as a result of
11 and 30, subsoil is treated as dense for SPT-N in range of 31 correlation analysis and linear regression analysis between N60
and 50 while very dense for SPT-N value above 51. and Dr. The R2 value indicates that 40% of the variation in Dr
values cannot be explained by using only N60. Also, it is observed
2.3 Data Analysis from correlation analysis that in-situ dry unit weight (γdf) is
To develop the correlation between SPT-N value and relative directly proportional to N60. Linear regression analysis was also
density, field SPT-N values are corrected for dilatancy, correction performed between ( γdf ) and N60. The R2 value of 0.62 and R
for field procedures and overburden to compensate the variations value of 0.79 are just satisfactory indicating that correlation between
in SPT blow count. Corrections are applied using the following these two variables cannot predict efficiently. It is also evident
procedures. from correlation analysis between effective overburden stress
• SPT-N values measured between any two consecutive loca- and relative density that with increase in depth, overburden stress
tions where undisturbed soil sample was collected in the increases and relative density increases. However, regression
borehole are averaged out. These average SPT-N values are statistics i.e., R2 is 0.60. This revealed that 40% of the variation
then used in the development of correlation. cannot be explained by the relation between relative density and
• Dilatancy correction was applied on average SPT-N values overburden stress so model between these two parameters is not
where SPT test was performed on saturated soils at depth reliable. The scatter diagram of void ratio range with relative
below ground water table (Meyerhof, 1956). Dilatancy density Dr indicates that no correlation exist between these two
N − 15 . parameters. It is also observed from correlation analyses and
correction is calculated using the relation N = 15 +
corr
2 linear regression analyses that any two parameters cannot be
Correction is applied when SPT-N values are greater than 15 correlated to predict relative density. The coefficient of determination
below ground water table. (R2) for different linear relations between different variables is in
• Afterwards, SPT-N values corrected for dilatancy are cor- the range of 0 to 0.62. Therefore, it can be inferred that variation
rected for field procedures. Correction was applied using the in the data which cannot be explained by respective parameter
Em C B C s C R N varies from 38 to 100%. For these reasons, multivariate regression
formula N 60 = 0.6
where hammer efficiency,
analysis was performed on the data to develop correlation which
Em = 0.5 , borehole diameter correction, CB = 1 , sampler can efficiently predict relative density. Statistical Package for the
correction, Cs = 1 , rod length correction CR = 0.75 , and N Social Sciences (SPSS) software was used to carry out multiple
is average number of blows. Hammer efficiency was not regression analysis using stepwise regression analysis technique
measured during current geotechnical investigation. How- to develop the correlation between above mentioned parameters.
ever, the value of Em is taken 0.5 based on hammer effi- As identified in the stepwise regression analyses, N60 and σv ′
ciency measurement reported by Kibria and Masood (1998) pass t test with 95% confidence interval and have a significant
while conducting geotechnical investigations for Indus effect on the relative density of sandy soil deposits as compared
sands at Chashma. with other input parameters. Therefore, only these two parameters
1 are used in model formulation and the remaining input parameters
• Liao and Whitman (1986) relation; CN = 9.78 σ ' where
v are discarded. Model calibration is carried out using regression
σv ′ is in kPa, is used for overburden correction to compen- analysis and the values of coefficients for input and output
sate the depth effects. parameters are calculated. The final best fit model obtained is
• Relative density was calculated using the expression given in Eq. (11)
γ d (max) ⎛ γ d ( field ) − γ d (min) ⎞ pa
Dr = ×⎜ ⎟ Dr (%) = 1.96 N − 19.2( ) 0.23
+ 29.2
γ d ( field ) ⎜⎝ γ d (max) − γ d (min) ⎟⎠
60
σv'
(11)
done with the prediction made by using equations of other studies Study [Eq. (11)] and Yoshida and Ikemi Eq. (5)
using the data used for the development of Eq. (11). These include
studies by Gibbs and Holtz (1957), Meyerhof (1956), Yoshida and values when experimental Dr values are between 0-60%; and for
Ikemi (1988) and Kibria and Masood (1998). It can be observed 60-70% experimental values of Dr, predicted values fall in the
from Fig. 5 that Gibbs and Holtz relationship [Eq. (1)] underestimate ±10% band. For Dr > 70% experimental versus predicted values
Dr values by about 20%. This underestimation may be attributed to by Eq. (2) underestimates. Yoshi and Ikemi equation [Eq. (5)]
the difference in gradation of samples used by Gibbs and Holtz to overestimate the relative density values when experimental Dr is
develop the correlation. The mean grain size of the samples used by up to 50% where as experimental versus predicted values falls
Gibbs and Holtz is 0.3 and 1.5 mm and where as mean grain size of within ±10% limit, when Dr value vary between 50% to 70% and
sandy deposit used in this research is 0.9 mm to 0.1 mm. beyond 70% their equation underestimates as shown in Fig. 7.
Figure 6 compares the results of predictions obtained from the The probable reasons for these variations in predictions by
present study and that proposed by Meyerhof (1956). It can be equations proposed by various researchers are explained by
seen that Meyerhof’s relationship [Eq. (2)] overestimates Dr Cubrinovski and Ishihara (1999). They suggested that these
Study [Eq. (11)] and Gibbs and Holtz Eq. (1) Study [Eq. (11)] and Kibria and Masood Eq. (7)
Data set used to correlate relative density and SPT-N value was
also used to develop the correlation between friction angle and in Fig. 10.
relative density for sandy soil. Regression analysis is carried out on
the data to develop the correlation between relative density Dr and φ = 0.70 N 60 + 18.0 (13)
friction angle (φ) determined through direct shear test. Calibration
The coefficient of determination of Eq. (13) is 0.88 indicating
of the correlation is carried out using regression analysis and the
that 88% of the variation in the data can be explained by the
values of coefficients for input and output parameters are
correlation and only 12% is unexplained by this relation.
calculated. The final best fit correlation is given in Eq. (12).
Standard error of estimate for Eq. (13) is 1.88, F–statistics for
φ = 0.276 D (%) + 18.70
r (12) output parameter is 7171.6 and t-statistics for input parameters is
76.78 and 84.68, respectively. These values indicate that variable
The correlation is presented graphically in Fig. 9. Standard
passes the F-test and t-test at 95% confidence interval. Hence,
error of estimate for Eq. (12) is 1.63, F–statistics for output
these regression statistics indicated that the correlation between
parameter is 10048.05 and t-statistics for input parameters is
friction angle and N60 is significant. The lower and upper limit of
100.58 and 100.24, respectively. These regression statistics
intercept at 95% confidence interval is 17.13 ~ 18.03 and for
indicated that the correlation between friction angle and relative
coefficient of N60 is between 0.69 ~ 0.72. Experimental versus
density is significant. The lower and upper limit of intercept at
predicted values of φ using Eq. (13) are presented in Fig. 11.
95% confidence interval is 18.34 ~ 19.06 for intercept and that of
Variation within ±10% of 90% of the predicted values indicates
coefficient of Dr is between 0.27 ~ 0.28. The coefficient of
that the relation can be used for future prediction. Relations
determination (R2) for Eq. (12) is 0.9 indicating that 10% of
presented by Hatanaka and Uchida (1996), Peck et al. (1974) and
variation in the values of φ is not accounted by the above
Japanese Road Association (1990) are employed to predict φ
mentioned regression equation.
using the same data set and the estimations using these relations
Also the same data set is used to correlate friction angle (φ) and
along with proposed equation are presented in Fig. 12, 13 and 14,
SPT-N60 values. Linear regression analysis technique is adopted
respectively. It can be inferred from Fig. 12 that Hatanaka and
to propose the relation. The best fit relation achieved through
Uchida [Eq. (8)] over estimates φ when the strata is in loose to
regression in presented in Eq. (13) and also presented graphically
Fig. 12. Experimental Versus Predicted Values of φ using Present Fig. 14. Experimental Versus Predicted Values of φ using Present
Study [Eq. (13)] and Hatanaka and Uchida Eq. (8) Study [Eq. (13)] and Japanese Road Association Eq. (10)
5. Conclusions
of Pakistan. PhD Thesis, Institute of Geology, Punjab University, Engineering, 2nd ed., Wiley, New York.
Lahore, Pakistan. Rogers, J. D. (2006). “Reliability of using Standard Penetration Test (SPT)
Hettiarachchi, H. and Brown, T. (2009). “Use of SPT blow counts to in predicting properties of silty clay with sand soil.” Environmental &
estimate shear strength properties of soils: Energy balance approach.” Engineering Geoscience, Vol. XII, No. 2, 161-179, DOI: 10.2113/
Journal of Geotechnical and Geoenvironmental Engineering, 12.2.161.
Vol. 135, pp. 25-32, DOI: 10.1061/(ASCE)GT.1943-5606.0000016. Salari, P., Lashkaripour, G. R., and Ghafoori, M. (2015). “Presentation
Japan Road Association (1990). Specification for Highway Bridges, of empirical equations for estimating internal friction angle of GW
Part IV. and GC soils in mashhad, iran using standard penetration and direct
Jianguo, C. (2012). Correlation analysis of SPT N values and cohesion shear tests and comparison with previous equations.” Open Journal
and internal angle of a clay.” Soil Engineering and Foundation, of Geology, No. 5, pp. 231-238, DOI: 10.4236/ojg.2015.55021.
Vol. 26, No. 4, pp. 91-93. Sivrikaya, O. and Togrol, E. (2006). “Determination of undrained shear
Kibria, S. and Masood, T. (1998), “SPT, Relative Density and PHI strength of fine-grained soils by means of SPT and its application in
Relationships for Indus Sands at Chashma.” Proceeding of VII National Turkey.” Engineering Geology, Vol. 86, pp. 52-69, DOI: 10.1016/
Conference of Pakistan National Society for Soil Mechanics and j.enggeo.2006.05.002.
Foundation Engineering, 169-188 Skempton, A. W. (1986). “Standard penetration test procedures and the
Kulhawy, F. H. and Mayne, P. W. (1990). Manual on Estimating Soil effect in sands of overburden pressure, relative density, particle size,
Properties for Foundation Design, Final Report (EL-6800) submitted to aging and overconsolidation.” Geotechnique, Vol. 36, No. 3, 425-
Electric Power Research Institute (EPRI), Palo Alto, Calif. 447, DOI: 10.1680/geot.1986.36.3.425.
Liao, S. and Whitman, R. V. (1986). “Overburden correction factor for Tomlinson, M. J. (1986). Foundation Design and Construction, Longman,
SPT in sand.” Journal of Geotechnical Engineering, ASCE, Vol. 112, Singapore.
No. 3, pp. 373-377, DOI: 10.1061/(ASCE)0733-9410(1986)112:3(373). Wolff, T. F. (1989). “Pile capacity prediction using parameter functions.” in
Meyerhof, G. G., (1956). “Penetration tests and bearing capacity of Predicted and Observed Axial Behavior of Piles, Results of a Pile
cohesionless soils.” ASCE Journal of Geotechnical Engineering, Prediction Symposium, sponsored by Geotechnical Engineering
Vol. 82, No. 1, pp. 866/1-866/19. Division, ASCE, Evanston, Ill., June 1989, ASCE Geotechnical
Mohammad, M. (2013), “Reliability of Standard Penetration Test (SPT) Special Publication No. 23, 96-106.
in predicting properties of silty clay with sand soil.” International Yoshida, Y. and Ikemi, M. (1988), “Empirical formulas of SPT blow-
Journal of Civil and Structural Engineering, Vol. 3, No. 3, pp. 545- counts for gravelly soils.” Proc. 1st Int. Symposium on Penetration
556, DOI: 10.6088/ijcser.201203013050. Testing, Rotterdam, pp. 381-387.
Peck, R. B., Hanson, W. E., and Thornburn, T. H. (1974). Foundation