Missouri University of Science and Technology Missouri University of Science and Technology

Scholars' Mine Scholars' Mine

International Conference on Case Histories in Geotechnical Engineering

(1993) - Third International Conference on Case Histories in Geotechnical Engineering

02 Jun 1993, 9:00 am - 12:00 pm

Prediction of Drilled Shafts Axial Capacities Using CPT Results Prediction of Drilled Shafts Axial Capacities Using CPT Results

O. M. Alsamman University of Illinois, Urbana-Champaign, Illinois

J. H. Long University of Illinois, Urbana-Champaign, Illinois

Follow this and additional works at: https://scholarsmine.mst.edu/icchge

Part of the Geotechnical Engineering Commons

Recommended Citation Recommended Citation Alsamman, O. M. and Long, J. H., "Prediction of Drilled Shafts Axial Capacities Using CPT Results" (1993). International Conference on Case Histories in Geotechnical Engineering. 47. https://scholarsmine.mst.edu/icchge/3icchge/3icchge-session01/47

This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

This Article - Conference proceedings is brought to you for free and open access by Scholars' Mine. It has been accepted for inclusion in International Conference on Case Histories in Geotechnical Engineering by an authorized administrator of Scholars' Mine. This work is protected by U. S. Copyright Law. Unauthorized use including reproduction for redistribution requires the permission of the copyright holder. For more information, please contact scholarsmine@mst.edu.

II Proceedings: Third International Conference on Case Histories in Geotechnical Engineering, St. Louis, Missouri, June 1-4, 1993, Paper No. 1.30

==

Prediction of Drilled Shafts Axial Capacities Using CPT Results 0. M. Alsamman Civil Engineer, Graduate Research Assistant, University of Illinois, Urbana-Champaign, Illinois

J.H.Long Associate Professor, Department of Civil Engineering, University of Illinois, Urbana-Champaign, Illinois

SYNOPSIS Results from nineteen full-scale axial load tests are reviewed to compare methods for predicting axial capacities of drilled shafts using results of Cone Penetration '!ests (CP'I). The three methods to estimate failure loads are: (1) Nottingham method, (2) Laboratoire des Ponts et Chaussees method (LPC), and (3) the Poulos and Davis method. Comparisons are made to assess the accuracy and dependability of each predictive method. Analyses of results indicate that the LPC method provides the most reliable predictions for axial capacities of drilled shafts in clay.

INTRODUCTION

Cone penetration test results provide a detailed representation of specific subsurface conditions and have been used successfully for determining axial capacities for driven pile foundations. However, limited information exists to demonstrate how well capacities of drilled shafts can be predicted using cone results. This paper identifies and quantifies the ability to predict axial capacity of drilled shafts in sand and in clay for methods that employ results of static cone penetration tests.

After reviewing details of 722 axial load tests on drilled shaft foundations, nineteen well-documented load tests on large straight-sided drilled shafts were selected for this study. Results and data from each load test were analyzed to determine the accuracy of CPT methods for predicting axial capacity. The three methods are evaluated by comparing measured and predicted capacities.

DATA ANALYSIS

Vertical forces acting on a drilled shaft are shown in Fig. 1. At the maximum axial load,

(1)

where Or is the dn1led shaft total ultimate capacity, Oeb is the net end-bearing or tip capacity, Q, is the side capacity along the shaft perimeter.

load '!ests

Details of a database containing 7Zl. axial load tests on drilled shafts were reviewed to concentrate on predictive methods that use results of cone penetration tests. Nineteen load tests satisfied the requirements of a simple soil profile (clay only, or sand only), available cone penetration results, and a well-documented load test

Selected details of the nineteen load tests are given in 'Thble 1. and include shaft depth, shaft diameter, DIB ratio, cone type and the original reference for each load test A total of 10 load tests are used in the analysis for shafts in sand Measured capacities are in the range of 160 to 360 tons, except for load test no. 713, which exhibits an a:xial capacity of 1640 tons. Shaft lengths range from 20 to 138 feet, and diameters range between 24 and 60 inches.

In clays, nine load tests are analyzed. Shaft lengths vary from 23 to 89 feet with diameters between 18 and 72 inches. Axial capacities of the drilled shafts varied from 144 tons to 922 tons.

113

z

Fig. 1 Forces Acting on a 1YPical Vertical Drilled Shaft

The type of cone used for testing at the different sites is also given in 'Thble 1. where M represents mechanical cones (Begemann-type), and Sis used to indicate electric cones (Fugro-type).

Measured Failure loads

In this study, the failure load was determined from the load-deflection relationship measured at the top of the shaft. The failure load is defined as the load corresponding to a settlement equal to 5 percent of the shaft diameter, plus the elastic shortening of the pile.

PREDICTIVE ME'IHODS

1\vo methods are used herein to predict axial capacity of shafts in clay. The Nottingham method (Nottingham, 1975) which is described in Schmertmann (1978), and the LPC method (Bustamante and Gianeselli, 1982). In addition to these two methods, the Poulos and Davis method (Poulos and Davis, 1980) is also used for analyzing drilled shafts in sand

The three methods correlate field cone resistance with shaft capacity in different ways. The Nottingham method uses both cone tip and side resistances while the LPC approach uses only cone tip values to predict shaft axial capacity. In the Poulos and Davis method in sand, cone results are converted to strength, which is then used to estimate shaft capacity. The methods are described below.

Nottingham Method

Nottingham method was developed for driven piles and based on results of 108 load tests on large-scale model piles. The method can be applied

Third International Conference on Case Histories in Geotechnical Engineering Missouri University of Science and Technology http://ICCHGE1984-2013.mst.edu

Table 1 Load Test Data

Soil Load Depth 'IYpe Thst no.

89

91

93

~ 95

302

404

406

407

628

713

17

20

33

~ 35

37

u 158

159

652

653

a M = mechanical S =electric

(ft)

42.6

45.9

19.7

19.7

23.6

60

39.4

39.4

32.1

138

23

45

33

42.6

32.8

82.2

88.7

50

50

Diametei (ft)

3.61

4.92

3.61

4.92

1.97

2.07

1.97

1.97

1.97

3.44

2.5

2.5

2.89

4.26

5.9

3.56

3.61

1.5

1.5

bassumed

a D!B Cone Reference Ratio 'l}rpe

11.8 M Franke and Garbrecht (1977)

9.3 M Franke and Garbrecht (1977)

5.5 M Franke and Garbrecht (1977)

4 M Franke and Garbrecht (1977)

12 M Martins and Martins (1989)

29 s Kruizinga and Nelissen (1985

20 s Maertens (1985)

20 s Maertens (1985)

16.3 Mh Burch et al. (1988)

40 Mh Caputo and Viggiani (1988)

9.2 M O'Neill and Reese (1972)

18 M O'Neill and Reese (1972)

11.4 M Jelenek et al. (1977)

10 M Jelenek eta!. (1977)

5.6 M Jelenek eta!. (1977)

23.1 Mb Yukang and Qianghua (1985)

24.6 Mb Yukang and Qianghua (1985)

33.3 s Finno (1989)

33.3 s Finno (1989)

to drilled shafts by reducing the side· resistance calculated for a driven pile of the same geometry by 25 percent (Schmertmann, 1978).

The value of end-bearing resistance is determined in both sands and clays by utilizing the Dutch method which averages cone tip resistance between 8B above the shaft tip and 0.7B to 4B below the shaft tip. If the mechanical penetrometer is used in clays, the computed average qc value is multiplied by 0.6. Nottingham also recommends using limiting end-bearing capacities of 140 ton/ft2 in sands and 93 tonfft2 in very silty sands.

The side resistance is calculated in sands by using the expression

(2)

where ks correction factor for sand (Fig. 2a) z depth at which side resistance is calculated D pile length B pile diameter fc local side friction measured by a cone device As pile surface area

If the cone side friction fc is not available, then fc is assumed to be equal to 0.007qc in sands.

The side resistance in clays is calculated using equation (3) and the design curve in Fig. 2b.

(3)

where kc is a reduction factor for clays (Fig. 2b ).

114

ks f., (kg!cm2) 0

0

~ Q II 10 0

'.tl .. 1:> ...

I 20 kc c. 2:> <I)

"0 <I)

30 :-.:= c.

3!>

40 0 (a) (b)

Fig. 2 Nottingham's Side-Friction Correction Factors for (a) Sand and (b) Clay

LPC Method

The LPC method is based on 197 full-scale load tests, of which 55 are drilled shafts. The diameter of the shafts range from 0.42 to 1.50 m (1.4 to 4.9 ft). Tests were conducted at 48 sites comprising soils of wide variety that include clay, silt, sand, gravel, peat, marl and weathered rock.

The equivalent cone resistance at the shaft tip is found by averaging qc along 1.5B above and 1.5B below the shaft tip. The unit tip resistance of the shaft qeb is then calculated as kqc , where k is a factor that depends on soil type and pile installation procedure. The value of k is 0.375 for drilled shafts in clay and 0.15 for shafts in sand. The unit side resistance f5 along the shaft is determined for each soil layer as a function of cone tip resistance, pile category and shaft dimensions. The pile category is determined based on excavation technique and drilling details. The procedure is described in detail by Bustamante and Gianeselli (1982).

Poulos and Davis Method

The Poulos and Davis method predicts axial capacity of a single pile in sand by using an idealized distribution of effective vertical stress (a'v) with depth adjacent to a pile. The value of a'v is taken as the overburden pressure to some critical depth Zc. beyond which a' v remains constant. The end-bearing and side capacities are respectively equal to:

where

(4)

z•D

Q, - I nBi. <K. tan;.J b.z (5)

Ap = cross-sectional area of the pile a'vp = effective vertical stress at the shaft tip Nq = bearing capacity factor and a function of q,' below the

, shaft base (see Fig. 3) · a v = effective vertical stress along the shaft (Iimi ted to the

critical overburden stress a' vc for z > Zc) ~ = coefficient of lateral pressure q, s = angle of friction between shaft and sand .1.\.z = increment of depth

The value of q,' for sand is related to the cone resistance using Fig. 4 (Durgunoglu and Mitchell, 1975). The Figure can be used to estimate a reasonable lower bound for the angle of shearing resistance of sand (Robertson and Campanella, 1983).

Third International Conference on Case Histories in Geotechnical Engineering Missouri University of Science and Technology http://ICCHGE1984-2013.mst.edu

1000

Nq 100

v 10

25

/ v

30 35

cj> '.

/

/

40

For drilled shafts cj>', = cj>' -3

/

45

cj> ', : angle of friction between shaft and sand cj>' : angle of shearing resistance of sand

Fig. 3 Relationship between Nq and cj>', (after Berezantzev et al., 1961)

In concept, the method involves predicting the drained strength of sand ( cj> ') and then estimating shaft capacity. Values of zdB and Ks tan q, '5 are plotted against values of q,' in Fig. 5. It is recommended by Poulos and Davis to redu~ q,' for drilled shafts by 3 degrees before using Figs. 3 and Sa.

RESULTS

The calculated and measured capacities are shown in Thble 2 and Thble 3 for shafts in clay and sand, respectively. 'IOta! calculated capacities are plotted versus measured capacities on log-log plots for each predictive method to illustrate general trends (Figs. 6-8).

For shafts in clay, the LPC method predicts capacity with a calculated to measured ratio (QciOm) between 0.84 and 1.4. The Nottingham procedure results in QciOm ratios from 0.55 to 267.

In sand, the Nottingham method overpredicts total capacity for all data points, sometimes excessively as in load test no. 95 (Qc!Om = 6.2). The ratio of calculated to measured capacities for the Nottingham method ranges from 1.1 to 6.2 while the LPC method predicts capacity generally within a factor of two. The Poulos and Davis method predicted capacities with QciOm ratios of 0.6 to about 3. It is also noted that all three methods predicted the highest capacity load test (load test no. 713) with relatively good accuracy.

cone tip resistance (MN/ m2)

.,..""' , .. e

~ Ill ll "' ~ ~ .,

~ ~

Fig. 4 Relationship between cp' of Sand and Cone Resistance (after Durgunoglu and Mitchell, 1975)

115

zJB

20 1·6

15 1·2

10 0.8

5 0·4

028 0

33 38 43

q,', cj>

(a) (b)

For drilled shafts cj> '1 = cj>' - 3 cj> ', : angle of friction between shaft and sand cj>' : angle of shearing resistance of sand

Fig. 5 Values of ~/Band kstan cj>', (Polous and Davis, 1980)

Thble 2 Ratios of Predicted to Measured Capacities for Load Tests in Qay

Measured Predicted CaQaci!Y Load Thst

Capacity Measured Capacity no.

(tons) LPC Nottingham

17 144 1.33 1.14

20 316 1.15 0.99

33 164 1.41 1.48

35 535 1.22 1.31

37 713 132 1.23

158 865 0.84 2.67

159 922 1.17 2.62

652 160 1.05 0.55

653 165 1.10 0.59

Thble 3 Ratios of Predicted to Measured Capacities for Load Tests in Sand

Load Thst Measured Predicted CaJ2aci!Y no . Capacity Measured Capacity

(tons) LPC Nottingham Poulos & Davis

89 295 0.86 2.73 1.82

91 358 1.38 4.23 2.77

93 278 1.25 3.87 1.90

95 300 1.88 6.16 3.24

302 162 1.06 1.77 0.59

404 200 1.26 2.26 1.46

406 302 0.34 2.19 1.63

407 270 038 2.45 1.82

628 212 0.33 1.25 1.34

713 1640 0.85 1.10 1.57

Third International Conference on Case Histories in Geotechnical Engineering Missouri University of Science and Technology http://ICCHGE1984-2013.mst.edu

to•

1 a ......... a

~ 10 I

·~ <11 a

~ '0

~

l 10 I

measured capacity Qm (tons)

Fig. 6 Thtal Capacity Predictions Using the Nottingham Method

ERROR ANALYSIS

Comparisons between predictive methods are based upon a simple statistical approach using a mean and a standard deviation. The measure of accuracy for a predictive method is defined as the ratio of calculated to measured capacity, Qc/Qm. The value of QcfQm varies from zero to one if the method predicts a capacity less than measured, equals to one if the predicted and measured values are equal, and is greater than one for predicted values greater than measured. The frequency distribution is approximated well by a log normal distribution which can be used to reduce the positively skewed histogram of Qc/Qm. Statistics for a log normal distribution are similar to a normal distribution, except operations are performed on the logarithm of the

~ g 8 10 1

t ~

I ...

10 I

measured capacity Qm (tons)

Fig. 7 Thtal Capacity Prediction Using the LPC Method

116

ratio QcfQm. The average value of Qc/Om for a log normal distribution, (QcfQm)avg , is determined as follows:

(6)

where n is the number of tests used in the analysis. The standard deviation cr is found using the expression:

(7)

Poulos and Davis

1 g 8 10 I a a

·t ~ 1 l 10 I

measured capacity Om (tons)

Fig. 8 Thtal Capacity Prediction Using the Poulos and Davis Method

Values of (QcfQm)avg and CJ were determined for each of the predictive methods and are given in 'Th.ble 4. The value of (QcfQm)avg is a measure of the bias of the method and therefore indicates how, on the average, predicted capacity agrees with measured capacity. For instance, if (QcfQm)avg equals 0.8, the method generally underpredicts capacity by 20 percent. The standard deviation numerically represents the dispersion associated with the ability to predict capacity. The method having the lower standard deviation will predict capacity with greater certainty.

'Th.ble 4 Factors of Safety for the Predictive Methods at a Reliability of 1:1000

Predictive Soil no. (QciOnJ.v. f1 Factor of Safety Method 'l}'pe of Required for a

Thsts 99.9 % Reliability

sand 10 25 0.23 128 Nottingham

clay 9 1.2 0.24 6.7

!.PC sand 10 0.8 0.27 5.7

clay 9 1.2 0.07 1.9

Poulos & Davis sand 10 1.7 0.20 6.8

Third International Conference on Case Histories in Geotechnical Engineering Missouri University of Science and Technology http://ICCHGE1984-2013.mst.edu

Assuming a log normal distribution and using the average, (QciOm)avg, and standard deviation o, the factor of safety can be associated with a prescribed reliability. Given in 'Thble 4 are factors of safety required for a reliability of 99.9 percent.

For a reliability of 99.9 percent, the LPC method predicts capacities for shafts in clay with the smallest factor of safety (1.9). The LPC method demonstrated an ability to predict capacities in clay superior to the other methods evaluated in this study. The performance of the LPC method was not as impressive for sand profiles.

The Nottingham method predicted measured values poorly, and therefore requires unreasonable factors of safety as shown in Th.ble 4. In part, unreasonably large factors are required because the method consistently overpredicts capacity. For shafts in sand, the Nottingham method overpredicts capacity by 2.5 times ((QcfQm)avg). However, the Nottingham method exhibits a smaller dispersion (o) than the LPC for sand. The Nottingham method exhibited a tendency to overpredict axial capacity of shafts in clay, and requires a factor of safety equal to 6.7 for a reliability of 99.9 percent.

The Poulos and Davis method was only applied to granular soils, and the method performed best of all three methods investigated. While a factor of safety of 6.8 is required for a reliability of 99.9 percent, the large factor is due primarily to the tendency of the Poulos and Davis method to overpredict capacity ((QcfOm)avg = 1.7). The dispersion (o) for the Poulos and Davis method is smallest of all the methods investigated herein.

CONCLUSIONS

Nine load tests of large drilled shafts in clay and ten load tests in sand are used to compare the predictive methods: the Nottingham method, the LPC method, and the Poulos and Davis method. The LPC method predicts capacities in clay with superior precision and accuracy; however, no method was clearly superior for shafts embedded in sand. The Poulos and Davis method performed the best for predicting the capacity of shafts in sand by exhibiting the smallest value for dispersion (o), however because the method overpredicts capacity (Qd'Om)avg = 1.7), a relatively large factor of safety of 6.8 is required to ensure a reliability of 99.9 percent.

The statistical values reported herein are undoubtedly influenced by the small number of load tests investigated. Additional tests are being collected to improve the confidence associated with these methods. However, the results suggest that results of static cone penetrometer tests may provide information useful in the efficient design of drilled shafts, particularly in clay. The advantage of using cone penetration tests for shafts in sand is less obvious. limited results indicate that the use of direct relationship between cone resistance and pile capacity, such as employed by the LPC and Nottingham methods, do not provide results as reliable as methods that use the cone penetration test to identify soil properties , which are then used to predict shaft capacities.

REFERENCES

Berezantzev, V. et al. (1961) "Bearing Capacity and Deformation of Piled Foundations," Proc. of the 5th ICSMFE, Paris, pp. 11-15.

Burch, S. et al. (1988) ''Design Guidelines for Drilled Shaft Foundations: VoLt: An Evaluation of Design Methods for Drilled Shafts in Cohesionless Soils," Dept. of Civil Eng., University of Florida, Gainsesville.

Bustamante, M. and Gianeselli, L. (1982) "Pile Bearing Capacity Prediction by Means of Static Penetrometer CYI;" Proc. 2nd European Symposium on Penetration 'Thsting, Amsterdam.

Caputo, V. and Viggiani, C. (1988) "Some experiences with bored and Auger Piles in Naples Area," Proc. Deep Foundations on Bored and Auger Piles, Van Impe, pp. 273-281.

117

Durgunoglu, H. and Mitchell, J. (1975) "Static Penetration Resistance of Soils," Proc. Conf. on In-situ Measurement of Soil Properties, ASCE, Raleigh, Vol.l, pp. 151-188.

Finno, R. (1989) "Subsurface Conditions and Pile Installation Data: 1989 Foundation Engineering Congress Thst Section," Proc. Predicted and Observed Axial Behavior of Piles, ASCE, No.23, 385 p.

Franke, E. and Garbrecht, D. (1977) '"lest-Loading on 8 Large Bored Piles in Sand," Proceeding 9th ICSMFE, Vol.l, Thkyo, pp. 529-532.

Kruizinga, J. and Nelissen, H. (1985) "Behavior of Bored and Auger Piles in Normally Consolidated Soils," Proceedings 11th ICSMFE, Vol.3, San Francisco, pp. 1417-1420.

Jelenek, R. et al. (1977) "Load Thsts on 5 Large-Diameter Bored Piles in Clay," Proc. 9th ICSMFE, Vol.l, Thkyo, pp. 571-576.

Maertens, J. (1985) "Comparative Thsts on Bored and Driven Piles at Kallo," Belgian Geotechnical Volume Published for the 1985 Golden Jubilee of the International society for Soil Mech. and Found. Eng., San Francisco, pp. 31-38.

Martins, F. and Martins, J. (1989) "CPT and Pile Thsts in Granitic Residual Soils," Proceedings 12th ICSMFE, Vol.l, Rio de Janeiro, Brazil, pp. 529-531.

Nottingham, L (1975) "Use of Quasi-Static Friction Cone Penetrometer Data to Predict Load Capacity of Displacement Piles," PhD Dissertation, University of Florida, Gainsville.

O'Neill, M. and Reese, L. (1972) "Behavior of Bored Piles in Beaumont Clay," J. Soil Mechanics and Foundation Division, ASCE, 98(2), pp. 195-213.

Poulos, H. and Davis, E. (1980) Pile Foundations Analysis and Desj~. John Wiley & Sons, New York.

Robertson, P. and Campanella, R. (1983) ''Interpretation of Cone Penetration 'Thsts: Parts 1 and 2," Canadian Geotech. J. Vol.20 pp. 718-745.

Schffiertmann, J. (1978) "Guidelines for Cone Penetration Test: Performance and Design," U.S. Dept. of 'ftansportation FHWA-TS-78-209, Washington, D.C.

Yakang, H. and Qianghua, C. (1985) "Load 'ftansfer Behavior of Bored Piles," Selected Papers from the Chineese Journal of Geotechnical Engineering, ASCE, pp. 101-112.

Third International Conference on Case Histories in Geotechnical Engineering Missouri University of Science and Technology http://ICCHGE1984-2013.mst.edu