Japan 02 [PDF]

Zitiervorschau

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/305657928

Assessment of SPT-based methods of pile bearing capacity- Analysis of a database Conference Paper · April 2002

CITATION

READS

1

338

1 author: Ali BOUAFIA Saad Dahlab University 102 PUBLICATIONS 144 CITATIONS SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Analysis of full-scale loading of piles under lateral loads View project

calcul fiabiliste et stochastique des ouvrages View project

All content following this page was uploaded by Ali BOUAFIA on 27 July 2016. The user has requested enhancement of the downloaded file.

Assessment of SPT-based methods of pile bearing capacity-Analysis of a database A.Bouafia

Department of civil engineering University of Blida, Blida, Algeria

A.Derbala

Sromek mechanical Foundations Dubai UAE ABSTRACT: It is nowadays recognised the in-situ tests–based methods are the most appropriate to predict pile foundations bearing capacity. In foundation design codes as well as in the literature are prescribed several SPT-based design methods, most of them being empirical. The paper presents a database consisting of 46 axial pile loading tests carried out in 27 sites in the United Arab Emirates with comprehensive geotechnical data. The piles are bored in slightly silty sandy soils. An evaluation of some currently used methods for the calculation of pile bearing capacity in sandy soils on the basis of SPT test was made. Comparison of predicted values to the ones experimentally derived from pile tests has led to rank these approaches with respect to their predictive capability for the bearing capacity of bored piles in sand. 1 INTRODUCTION The in-situ tests are nowadays widely used in geotechnical projects to characterise soil materials. They are faster and cheaper than those carried out in laboratory and do not necessitate any sampling. Standard Penetration Test (SPT) is appropriate to estimate the resistance and density of sandy media and is largely practised in the Arabian gulf region within the scope of foundation projects. It should be however mentioned that the diversity of empirical interpretations of SPT and the dependence of this latter on the test procedure and the device features are source of uncertainties and discrepancies between methods aimed to evaluate bearing capacity and settlement of foundations. These uncertainties in geotechnical pile design, notably within the scope of big sized projects, often lead to carry out static pile loading tests in order to experimentally determine pile bearing capacity and settlements as well. Although the SPT-based pile design literature is wealthy one needs to be aware of local geotechnical conditions from which several empirical formulas were derived. Caution is then necessary when using any SPT-based method. Several studies on pile tests databases were undertaken to empirically derive methods of pile design (Meyerhof 1956, Robert,1997) or to assess existing methods (Briaud and Tucker 1988, Bustamante et Al 1991). This paper is aimed to present the results of an evaluation study of bored piles behaviour in relation

with local geotechnical conditions in the U.A.E. The quality of prediction of ten commonly used design methods is assessed through a database including 27 silty sandy sites and 46 static pile loading tests. 2 FEATURES OF THE DATABASE 2.1 Pile loading test Piles studied are usually cast in-situ bored with either casing method (for large diameter piles) or continuous flying auger method (for medium diameter piles in stable soils). Slurry products like bentonite are sometimes used for maintaining the borehole of pile. Piles subjected to axial loading test are usually non-instrumented by strain gauges or sliding extensometers. They are simply connected at the top with four dial gauges, with a usual sensitivity of 0.01 mm for the settlement reading. The load is usually applied in increments by a hydraulic jack and pump assembly fitted with pressure gauge, against weighted platform. The testing programme consists of two cycles of loading. One up to the working load (design load) and the other one up to 1.5 to 2.0 times the working load. Each load increment is usually maintained until the rate of settlement is less than 0.25 mm/hour. Pile diameter B ranges between 0.45 and 1.10 m. The slenderness ratio D/B (Length/Diameter) varies from 10 to 36.7. Pile concrete compressive strength usually ranges between 20 and 40 MPa.

2.3 Soil material Most of the region of Arabian Gulf was subject in past geological eras to extensive carbonate sedimentation. Deposits of the UAE are mostly Pleistocene or recent in age. The superficial deposits mainly consist of sand dunes, loess, marine sand and silts. Typical soil material encountered in the sites studied is fine grained to medium grained, slightly silty to silty sand with traces of gravel, gypsum, shell fragments and cemented sand pieces. Ground water level was recorded from 0.0 to 6.50 m below soil surface. In every site were drilled two to three SPT boreholes with recovery of disturbed samples for usual identification laboratory tests. Blow counts (N value) at piles base varies between 42 to 100 blows/30 cm, which correspond to dense to very dense deposits. 3 INTERPRETATION OF PILES LOADING TESTS Typical load-top settlement Q-v0 curve is shown in figure 1 with a usual hyperbolic shape. The ultimate load Ql theoretically corresponds to an infinite pile top settlement and therefore to a horizontal asymptote of Q-v0 for large settlements. It is conventionally defined in many standards as the load corresponding to a settlement of 10 % of B. According to usual recommendations in the pile foundations literature, load-settlement curve shape usually fits well a hyperbolic function like: Q = v0 1 + v0 α Ql

(1)

where α is the initial slope of Q-v0 curve. The procedure of hyperbolic fitting by least squares was applied to the piles analysed. In all cases the correlation coefficient R was found greater than 0.85, which indicates good quality of the fitting procedure of experimental data. It is to be noted that Ql derived from the hyperbolic formulation is in excellent accordance with the con-

Experimental data Hyperbolic fitting

2500

2000

1500

1000

500

0 0,0

0,5

1,0

1,5

2,0

2,5

Cumulative pile settlement v0 (mm)

Figure 1.Typical Q-v0 curve

30000

Q (hyper. fitt.) kN

SPT test is the most commonly used in-situ test in the UAE. Typical SPT hammer has a mass of 63.5 kg falling form a height of 760 mm. The drive shoe has a length of 75 mm and internal diameter of 35 mm. Corrections recommended to take into account the ground water table or the vertical overburden pressure effects on N values are applied if they are prescribed by the methods tested here.

Axial load Q (kN)

2.2 The SPT test features

Q l = Q ( B / 10 ) 25000

20000

15000

10000

5000

0 0

5000

10000

15000

Q (B/10)

20000

25000

kN

Figure 2. Comparison between limit vertical load and conventional value

ventional value of ( Bouafia, 2001).

Ql, as illustrated by figure 2

4 METHODS OF COMPUTATION OF PILE BEARING CAPACITY The vertical limit load Ql on a pile, usually called pile bearing capacity, is the sum of the limit load Qp at pile base, or end bearing capacity, and the limit load Qs due to skin friction stresses mobilised along the soil/pile interface. Thus for circular uniform shaped piles : Ql =Q p +Qs = ql π.B 2 +π.B∫ qs.dz 0 4 D

(2)

The SPT-based methods of pile bearing capacity computation in sandy soils may be divided into two main categories. The first one contains empirical methods derived from back-analysis of databases of pile loading tests in correlation with N value. The end bearing capacity is assumed to be proportional to a representative Ne value around the pile base: (3)

The limit skin friction stress at a given depth is proportional to the N value at this depth: qs = ns.N

(4)

Ks and ns have the unit of stress and are respectively called tip resistance factor and skin friction factor. Methods 1-9 summarised in table 1belong to this category whereas method 10 belong to the following category. The second category comprises semi-empirical methods based on theoretical formulas of the end bearing capacity and limit skin friction along the pile adjusted by experimental observations of pile behaviour. Hansen-Burland’s method is a typical theoretical method which may be empirically adjusted as it will be further detailed. The end bearing capacity is expressed as follows: Nq = Nqs .sq .dq

(5)

Nqs corresponds to an ideal strip footing and theoretically depends only on ϕ. sq is a shape factor and dq is a depth factor which depends on ϕ and the slenderness ratio. For infinite depths, Hansen’s Nq factor reaches an asymptotic value. The term Ko.tgδ was found slightly varying for N values between 10 and 50. This fact shows that the choice of ϕ-N correlation has little effect on the values of skin friction stress. In some methods the vertical effective overburden pressure σv0’ along the pile is required. The computation procedure is as follows: - The friction angle ϕ is estimated according to Peck-Hanson’s ϕ-N value chart (Peck at al, 1973) which may be fitted for N less than 55 by the following linear relation : ϕ(°) = 27.560 + 0.274xN

γs

1+ m tgϕ

(7)

where m is an empirical factor ranging between 0.40 and 0.60 for silica sands. The analysis made by the author on 59 sandy samples subject to direct shear tests from different sites has shown the above relation fits the best the experimental data with m= 0.50 (Bouafia, 2000).

γw γs

)(1− )

(8)

The specific unit weight γs was taken equal to 26.5 kN/m3. - The effective overburden pressure is the computed as : σv’= ∑ γi’.Zi

(9)

Table 1 summarises main features of the methods to be assessed in this paper. In most of the cases, location of piles is not specified with respect to boreholes. In order to take into consideration the non homogeneity of soils, the calculation of pile bearing capacity was done for each borehole in the site, which necessitates 107 cases to be computed by each method. A computer program was made for accomplishing this task. Primary theoretical predictions of HansenBurland’s method were adjusted by an empirical factor λ in such a way as the values of bearing capacity Ql correspond within a reasonable tolerance to those derived from pile loading tests. Adjustment factor λ was found depending on the representative N value at pile base, as shown in figure 3. Moreover, it was noticed this factor does not depend on the slenderness ratio D/B. The Variation of λ-N value, as illustrated by figure 3, seems to be reasonably linear with : λ=0.0512xNe

(10)

5

Hansen-Burland λ=Ql/Qlexp.

4

3

(6)

- The void ratio e is correlated with ϕ, the internal friction angle of sand, by Caquot-Kerisel’s empirical correlation: e.tgϕ =m

γ'=(

λ

ql = Ks.N

- The submerged unit weight may be computed by:

2

1

0 0

10

20

30

40

50

60

Ne Figure 3. Variation of the adjustment factor λ versus Ns

70

Table 1. Summary of features of the different methods

1 2 3

4

5

References Ks (kPa) Bazaraa & Kurkur (1986) 135 if B≤0.5 m 270xB else ( B in m) 400 in sand Decourt (1982) 250 in residual silty sand 98.4 in sand Lopes & Laprovitera 87.0 in silty sand (1988) Meyerhof (1976) CFEM (1985)

ns (kPa) Ne & zone of influence Pile installation. 0.67 if B≤0.5 m Average within [D3.75xB, D+B] 1.34xB else bored

(B in m)

qs=10x(N/3+1) (in kPa )

1.62 in sand Average within [D- bored 1.94 in silty sand B, D+B]

120 400

1 2

100

1

-Average within [D-8B, D+3B] -correction due to depth effect

bored driven bored

Shioi et Al (1982)

6

Aoki et Velloso (1975)

7

PHRI Standard (1980)

8

Reese et Al (1989)

9

Robert (1997)

300 286 in sand 228 in silty sand

400 60 if B=0.52-1.27 76/B if B>1.27 m (B in m) 115

2 driven 2.00 in sand Average of the 3 N val2.28 in silty sand ues closest to pile base bored - average of N1 and N2 -correction due to ground water effect driven 2

190

10

bored

Hansen-Burland (1973)

3.3

bored

1.90

bored

1.90

driven

-correction due to depth effect ’ -Average within ql= Nq. σ v (D) Nq = Hansen’s bearing capacity factor [D-B/2, D+2B] Nq =f(ϕ ) ϕ derived from ϕ-N chart qs = K0.σv’.tgδ Burland’s β formula - correction due K0 = (1-sinϕ).(OCR)1/2 (OCR=1) intermediate roughness of pile shaft : to ground water effect δ= 0.75xϕ

5 ASSESSEMENT OF PREDICTIONS Table 2 summarises some statistics related to each method tested. µ is defined as the ratio of predicted baring capacity to the one experimentally derived. SD and COV respectively mean standard deviation and coefficient of variation (ratio average/SD) of the ratio µ. The underprediction rate is defined as the percentage of cases underpredicted. From table 2, since the coefficient of variation is almost same for all the methods, it can be concluded that these methods are characterised by the same level of dispersion of predictions with respect to the mean values of µ. Figure 4 shows that Decourt’s method in all the cases has overpredicted the bearing capacity.

According to table 1, this fact may be explained by relatively high values of factors Ks and ns. The same explanation is possible for Aoki and Vellosos’s method which has overpredicted almost all the cases with an average value of 1.91 for µ. Meyerohf’s method and CFEM(1985) have underpredicted most of cases with an average value of 0.70 for the ratio µ. The pessimistic prediction which may justify the large use of this method was already highlighted by many authors working on similar databases ( Turnbull and Kaufmann 1956, Mansur & Focht 1960). According to table 2, the factor ns of Meyerhof’s method for bored piles is the smallest one. Moreover, N values used in this method are to be reduced by the so-called depth effect.

Table 2. Results of assessment of the methods

Method No.

Average of µ

Min

Max

SD. COV % Underprediction rate %

1 2 3 4 5 6 8 9 10

0.97 3.13 0.95 0.70 0.82 1.91 1.11 0.98 0.93

0.35 1.29 0.37 0.31 0.36 0.86 0.43 0.44 0.32

1.92 6.55 2.40 2.03 1.86 4.27 3.08 2.56 1.54

0.31 0.99 0.32 0.25 0.27 0.60 0.40 0.34 0.30

20000

31.8 31.6 33.2 36.2 31.6 31.3 36.0 35.1 32.3

58.0 0.0 68.2 94.4 83.2 7.5 50.5 66.3 67.3

Aoki-Velloso

25000

Decourt-Quaresma 40000

Q pred. (kN)

15000

.= red p Q

10000

20000

xp. Qe

42.0 100.0 31.7 5.6 16.8 92.5 49.5 33.7 32.7

50000

30000

Meyerhof, CFEM

Overprediction rate %

30000

15000 20000

10000 10000

5000 5000

0

0

0

0

5000

10000

15000

20000

0

5000

10000

15000

0

5000

10000

15000

20000

20000

Qexp. Qex (kN) p. (kN)

20000

20000

20000

Reese-O'Neill

Hansen-Burland adjusted

Shioi-Fukui 15000

15000

15000

10000

10000

10000

5000

5000

5000

0

0

5000

10000

15000

20000

0

0

5000

10000

15000

20000

20000

0

0

5000

10000

15000

20000

15000

20000

20000

Robert

20000

Lopes-Laprovitera

Bazarra-Kurkur

15000

15000

Q pred. (kN)

15000 10000

10000

10000 5000

5000

0

5000

0

5000

10000

15000

20000

0

0

0

5000

10000

15000

20000

0

5000

10000

Qexp. (kN)

Figure 4. Comparison of predicted and experimental values of bearing capacity of bored piles for each method

Number of cases

20 0 0,0 20 0 0,0 20 0 0,0 20 0 0,0 20 0 0,0 20 0 0,0 20 0 0,0 20 0 0,0

Aoki-Velloso

0,5

1,0

1,5

2,0

2,5

3,0

Meyerhof

0,5

1,0

1,5

2,0

2,5

3,0

Shioi-Fukui

0,5

1,0

1,5

2,0

2,5

3,0

Reese -O'Neill

0,5

1,0

1,5

2,0

2,5

3,0

Hansen-Burland

0,5

1,0

1,5

2,0

2,5

3,0

Lopes-Laprovitera

0,5

1,0

1,5

2,0

2,5

3,0

Bazarra-Kurkur

0,5

1,0

1,5

2,0

2,5

3,0

Robert

0,5

1,0

1,5

2,0

2,5

3,0

Figure 5. Histograms of µ counts for different methods tested

As illustrated by figure 4, methods of Robert, Lopes & Laprovitera, Bazarra & Kurkur and Hansen & Burland seem giving the best prediction with a range of µ between 0.93 and 0.98. They can be considered as rather pessimistic since almost two third of cases were underpredicted by these methods. The prediction of the semi-empirical HansenBurland’s method seems better than many other purely empirical methods, which encourages to improve the quality of adjustment of this approach. According to figure 5, histograms of µ counts may be fitted by a Gaussian distribution. If one defines the criterion of ranking of these methods as the frequency counts of µ between 0.8 and 1.0, Robert, Lopes & Laprovitera, Bazarra & Kurkur and Hansen & Burland are respectively characterised by the frequencies of 30.8 %, 28.9%, 24.3% and 21.5%. 6 CONCLUSIONS On the basis of a database comprising 46 static pile loading tests carried out in 27 silty sand an evaluation study of nine SPT-based methods of

pile bearing capacity calculation was undertaken. The quality of prediction of these methods is assessed by direct comparison of the predicted limit vertical load carried by the pile to the one experimentally derived from static pile loading test. Decourt and Aoki-Velloso’s methods were found optimistic in most of the cases studied whereas Meyerhof’s method is rather pessimistic. The semi-empirical Hansen-Burland’s method allows very good prediction in comparison with the other empirical approaches. Methods of Robert, Lopes & Laprovitera, Bazarra & Kurkur and Hansen & Burland give an average value of the ratio µ (ratio load predicted to experimental load) ranging between 0.93 and 0.98. According the ranking criterion defined as the frequency count of µ ranging between 0.8 and 1.0, methods of Robert, Lopes & Laprovitera, Bazarra & Kurkur and Hansen & Burland are respectively characterised by the frequencies of 30.8 %, 28.9%, 24.3% and 21.5%. However, al the methods tested express some dispersion of prediction with a coefficient of variation of about 30% with respect to the mean value.

7 REFERENCES Aoki,N & Veloso,D.1975.An approximate method to estimate the bearing capacity of piles. Proceedings of the 5th Pan-American Conference on soil Mechanics and Foundation engineering,Vol 1, pp : 367-376, Buenos-Aires. Bazarra, A.R & Kurkur,M.M.1986. N-values used to predict settlements of piles in Egypt. In: Use of In-situ tests in geotechnical engineering, ASCE Geotech. Special Publication, Clemence ed., Vol. 6, pp : 462-474. Bouafia,A. 2000. Some comments on e-ϕ correlation for sandy soils. Internal publication (in French), Department of Civ.Engg. Univ. of Blida, 6 pages. Bouafia, A.2001. Pile foundations bearing capacity –The UAE experience, World of Engineering, Journal of the UAE society of Engineers, 7 pages. Briaud,J.L, L.M.Tucker.1988. Measured and predicted axial response of 98 piles. Journal of Geotech. Engg., Vol.114,No.9, September 1988. Burland,J.B.1973. Shaft friction piles in clay-A simple fundamental approach, Ground Engineering, vol. 6, N°3, PP. 30-42. Bustamante,M et Al.1991. Evaluation dequelques méthods de calcul des pieux forés (in French), RFG, French Geotech. Journal No.54, pp :39-52, january 1991. CGS.1985. Canadian Foundation Engineering manual CFEM, 2nd edition, Canadian Geotechnical Society, C/o Bitech publishers Ltd.,Vancouver, BC. Decourt,L.1982. Prediction of the bearing capacity of piles based exclusively on N-Value of the SPT. Proceedings of 2nd European Symposium on penetration testing, Vol 1, pp :29-34, Amsterdam. Hansen, J.B.1970. A revised and extended formula for Bearing Capacity, Danish Geotechnical Institute report No. 28, Copenhagen, 21 pages.

View publication stats

Lopes,R.F, Laprovitera,H.1988. On the prediction of the bearing capacity of bored piles from dynamic penetration tests. Proceedings of Deep foundations on bored and auger piles BAP’88, Van Impe (ed), pp: 537-540. Meyerhof,G.G.1956. Penetration tests and bearing capacity of piles in cohesionless soils. Proceedings of ASCE, Journal of SMFE, Vol.82,No. SM.1, January 1956. Meyerhof,G.G.1976. Bearing capacity and settlement of pile foundations. Journal of Geotech. Engg. ASCE, Vol.102, No.3, pp :1-19. Peck,R.B et Al.1973. Foundation engineering. 2nd edition, John Wiley & sons editors, 514 pages. Reese,L.C & O’Neill,M.W. 1989. New design method for drilled shafts from common soil and rock tests. Proceedings of Congress foundation engineering-Current principles and practices,ASCE, Vol 2, pp :1026-1039. Robert,Y,1997. A few comments on pile design. Can. Geotech. J. Vol.34. pp : 560-567. Shioi,Y & Fukui,J.1982. Application of N-value to design of foundations in Japan, Proceeding of the 2nd ESOPT, Vol. 1, pp 159-164. Tornbull,W.J & Kaufmann,R.I .1956.Discussion on the paper :Penetration tests and bearing capacity of piles in cohesionless soils by :Meyerhof, proceedings of ASCE, Journal of SMFE, Vol.82, No. SM.1, January 1956.