Escolar Documentos
Profissional Documentos
Cultura Documentos
GEOMECHANICS GROUP
BY
B.M. LEHANE
J.A. SCHNEIDER
X. Xu
5 CONCLUSIONS ........................................................................................... 17
ACKNOWLEDGEMENTS ................................................................................... 17
REFERENCES .................................................................................................... 18
APPENDIX A ....................................................................................................... 20
APPENDIX B ....................................................................................................... 33
1) All three CPT based design methods considered (Fugro-04, ICP-05 & NGI-04) had
significantly better predictive performance than the existing API recommendations,
which were seen to lead large under-predictions in dense sands and become
progressively non-conservative as the pile length (L) or aspect ratio (L/D) increased.
2) Despite the CPT based methods having a broadly similar predictive performance
against the UWA database of load tests, their formulations relating the pile end bearing
with the CPT end resistance (qc) are notably different. Formulations for shaft friction
also differ although all assume a near-proportional relationship between local shaft
friction (f) and qc and allow for the degradation of f with distance above the pile tip (h)
due to friction fatigue.
3) The ICP-05 method indicated the lowest coefficient of variation (COV) for calculated to
measured capacities (Qc/Qm) of 0.32, when an equal weighting is given to each pile test
in the database. However, for various categories within the database the position is less
clear. For example, NGI-04 predictions appear best for open-ended piles in
compression while the COV of Qc/Qm given by Fugro-04 is the same as that of ICP-05
for open-ended piles in tension.
4) When account was taken of the relative reliability of the pile test data, the design
methods listed below for each category of pile lead to the lowest probability of failure
(i.e. defined here as the relative probability that the calculated capacity does not exceed
the measured capacity by more than a factor of 2)
The fact that API-00 gives the lowest probability of failure for closed-ended piles is
partly because the method generally under-predicts the capacity of these piles1.
1
However, while the same average level of under-prediction also applies to API-00 predictions for closed-ended piles in
tension, the estimated probability of failure is larger than the three alternative design methods.
6) The shapes of the (weighted) probability density functions (PDFs) for Qc/Qm provided
for each category of pile indicate that the Qc/Qm variances for the ICP-05 and Fugro-04
methods are much lower than those of API-00 and marginally lower (on average) than
NGI-04.
This examination of the three CPT based methods coupled with a review of their various
deficiencies and a careful examination of the new extended database of static load tests
prompted the authors to propose the UWA-05 method presented here. This method is believed
to represent a significant improvement on Fugro-04, ICP-05 and NGI-04 methods. Particular
comparisons are made in this report with the ICP-05 method, which was provisionally selected
to form the basis of new API recommendations at the May 6th 2005 API-ISO committee meeting.
It will be shown in this report that the UWA-05 method:
has superior predictive performance to ICP-05 for the existing database of pile load
tests;
adopts more widely acceptable formulations for end bearing than those proposed by
ICP-05;
allows directly for the inclusion of the effects of partial plugging (and hence increased
level of soil displacement) on both the shaft and base capacity of pipe piles;
provides a more conservative extrapolation than ICP-05 from the existing database of
test piles to full scale offshore piles;
removes the bias shown by ICP-05 to over-predict tension pile capacity for large
diameter slender piles.
Supplementary data in support of the main text is provided in Appendix A, while Appendix B
provides detailed responses to a number of issues raised by the API- piling subcommittee on
review of the initial draft of this document issued on May 16th 2005.
Closed-ended piles
The strong direct relationship between the end bearing resistance of a closed-ended driven
pile and the CPT end resistance qc has been recognised for many years and arises because
of the similarity between their modes of penetration.
Given the difference in size between a pile and a CPT penetrometer, a correlation between
qb0.1 and qc requires use of an appropriate averaging technique to deduce an average qc =
q c . Xu & Lehane (2005) show that, for many stratigraphies encountered in practice, q c may
be taken as the average qc value taken in the zone 1.5 pile diameters (D) above and below
the pile tip.
Xu & Lehane (2005), however, also show that when qc varies significantly in the vicinity of
the pile tip (i.e. within a number of diameters), the Dutch averaging technique (Van Mierlo &
Koppejan 1952 and Schmertmann 1978) provides the most consistent relationship for end
bearing and should be employed e.g. see Figure 1.
The values of qb0.1 for driven piles is less than q c because the displacement of 0.1D is
insufficient to mobilise the ultimate value (of q c ).
The findings of Randolph (2003), White & Bolton (2005), and others, are consistent with the
UWA-05 proposal to adopt a constant ratio of qb0.1/ q c .
The UWA-05 method employs a constant qb0.1/ q c ratio of 0.6, which is independent of pile
diameter, pile length and sand relative density i.e.
2
Q b = q b 0. 1 D where q b 0.1 / q c = 0.6 (1)
4
The available database of base capacity measurements shown on Figure 1 confirms the
suitability of equation (1).
1.2 Baghdad
Ogeechee River Drammen
1.0
Hoogzand
Hsin Ta
0.8
qb0.1 / qc
Hunter's Point
0.6 Indiana
Kallo
0.4
Ogeechee
0.2 Sermide
Salt Lake
0.0
0 0.2 0.4 0.6 0.8 1
Pile diameter (m)
________________________________________________________________________
Figure 1: Database of qb0.1 values for closed-ended piles compiled by Xu & Lehane (2005)
(noting that qc is derived using the Dutch averaging technique)
Open-ended piles
The relationship between qb0.1 and q c may be expected to be less consistent than that for
closed-ended piles in view of the differences between the mode of penetration of a cone and
that of a coring or partially plugging pile during installation.
Salgado et al. (2002), Lehane & Gavin (2001, 2004), and others, have shown that a
relatively consistent relationship between qb0.1 for a pipe pile and the CPT qc value becomes
apparent when the effects of sand displacement close to the tip during pile driving are
accounted for. This installation effect is best described by the incremental filling ratio (IFR)
measured over the final stages of installation- and is referred to here as the final filling ratio
(FFR). As the FFR approaches zero, the value of qb0.1 approaches that of a closed-ended
pile.
The displacement induced to the sand in the vicinity of the base is most conveniently
expressed in the terms of the effective area ratio Arb*, defined below. This ratio depends on
the piles D/t ratio and the FFR value and varies from unity for a pile installed in a fully
plugged mode to about 0.08 for a pile installed in coring mode with a D/t ratio of 50.
Lehane & Randolph (2002), and others, have shown that if the length of the soil plug is
greater than between 3 and 5 internal pile diameters (3 to 5Di), the plug will not fail under
static loading, regardless of the pile diameter.
Experimental data indicate that the resistance that can develop on the tip annulus at a base
movement of 0.1D varies between about 0.6 and 1.0 times the CPT qc value (e.g. Bruno
1999, Salgado et al. 2002, Lehane & Gavin 2001, Paik et al 2003).
Recommended values of qb/ q c at 10% pile tip displacement for bored piles range from 0.15
to 0.23 (Bustamante & Gianeselli 1982, Jamiolkowski & Lancellotta 1988, Ghionna et al.
1993, Franke 1989). These ratios are not dependent on the pile diameter.
The value of q c should be evaluated in the same way as that employed for closed-ended
piles, but employing an effective diameter (D*) related to the effective area ratio, Arb* i.e. D*
= D Arb*0.5.
There are relatively few documented case histories, which report the incremental or final
filling ratios. In the absence of FFR measurements, a reasonable estimate of the likely FFR
may be obtained from the trend line through existing measurements shown on Figure 2; an
equation for this trend line is provided in equation 2d.
1.2
0.6
0.4
D (m)
0.2
IFRmean = min1, i
0.2
1.5
0
0 0.4 0.8 1.2 1.6 2
___________________________________________________________________________
Figure 2: Database of incremental filling ratios averaged over the last 20D of penetration; note
that these IFRmean values are virtually identical to the FFR values measured at the case history
sites considered in Figure 3.
D i2
where A = 1 FFR 2
*
rb
D (2c)
D (m ) 0.2
FFR = min 1, i
1.5m (2d)
0.60
0.50
0.40
qb0.1/qc
Dunkirk
0.30
Euripides
Hoogzand
0.20 Indiana
Ras Tanajib
0.10 Shanghai
Tokyo
0.00
0.0 0.2 0.4 0.6 0.8 1.0
Effective area ratio, Arb*
____________________________________________________________________
Figure 3: Database of qb0.1 values for open-ended piles in sand
2
Strong indirect evidence in support of this observation is also apparent in Lehane, Schneider & Xu (2005), which
shows that the Fugro-04, ICP-05 and NGI-04 progressively under-predict the shaft capacity of jacked piles as the pile
length increases.
32
Employed for
database evaluation
30
Interface Friction Angel, cv
26
24
22
20
0.01 0.1 1 10
Median Grain Size, D50 (mm)
Q s = D f dz
(3a)
where
f = local shear stress at failure along the shaft of a pile, and
f
f = ' rf tan cv = (' rc + ' rd ) tan cv
fc (3b)
with
cv = constant volume interface friction angle; see Figure 4
'rf = radial effective stress at failure
'rc = radial effective stress after installation and equalization
'rd = increase in radial stress due to loading stress path (dilation)
f / fc = 1 for compression loading and 0.75 in tension
The radial effective stress after installation and equalisation is given as:
0 .5
( )
' rc = 0.03 q c A rs
* 0 .3 h
max D ,2 where
(3c)
D 2
A rs = 1 IFR mean i 2
*
D
(3d)
where IFRmean is the average IFR measured over the final 20 diameters of
penetrations Where an IFR profile is not measured, IFRmean may be estimated
from:
D i (m) 0.2
IFR mean = max 1,
1.5m (3e)
The change (increase) in radial stress during pile loading, which may be
assumed to be zero for a full scale offshore pile, is given as:
r
and q c1 N = (q c p atm ) / ( ' vo p atm )
0 .7 0 .5
' rd = 4G with G = q c 185 q c1 N
D (3f)
where G = operational shear modulus (assumed equal to the in-situ very small
strain value), r = interface dilation (assumed =0.02mm for the database) and
patm is a reference stress=100 kPa.
3
Professor Jardine expressed reliability concerns for the CLAROM piles due to multiple restriking of the piles, re-
testing, their relatively small diameter, as well as the time period allowed between installation and load testing.
Additionally, the single qc profile available for the CLAROM tests did not allow for the same confidence levels in the qc
profiles at the test pile locations as the five CPTs at the Dunkirk ZdH site.
Table 1: Assessment of the performance of UWA-05 and that of API-00, ICP-05, Fugro-04, NGI-
04 for the database of load tests on driven piles in siliceous sand
Qc [ICP-05]/Qm
1.5 1.5
113 112
0 0
0 20 40 60 80 100 0 20 40 60 80 100
L/D (-) L/D (-)
2.5 2.5
UWA-05 ICP-05
average=1.02 average=1.02
cov=0.27 cov=0.28
2 2
Qc [UWA-05]/Qm
203
Qc [ICP-05]/Qm
1.5 1.5
205
0 0
0 20 40 60 80 100 0 20 40 60 80 100
L/D (-) L/D (-)
2.5 2.5
UWA-05 ICP-05
average=0.98 average=0.89
cov=0.19 cov=0.25
2 2
Qc [UWA-05]/Qm
Qc [ICP-05]/Qm
1.5 1.5
306 307
313
310 306
318 309
311 314 308
307 309
1 317
312 308 1 317
311 300
301 314 310
318
302300 316
315 313 315
312 316
303
301
303
0.5 0.5 302
0 0
0 20 40 60 80 100 0 20 40 60 80 100
L/D (-) L/D (-)
Qc [ICP-05]/Qm
1.5 1.5
414
405
407 412
408 416 411
415
1 410 409
416 1 409
417
408
411 410 406
417 407
406 413
401 405
412 400 400
401
413 414
0.5 0.5 415
0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
L/D (-) L/D (-)
One of the factors giving rise to the superior performance of the UWA-05 method for pipe piles
is the inclusion of the effective ratio terms in the expressions for base and capacity of open-
ended piles. This is not surprising given the acknowledged importance of soil displacement on
capacity and the fact that many of the database piles showed evidence of partial plugging.
However, given that the incremental filling ratio (IFR) is not commonly measured in practice, the
sensitivity of the predictive performance summarised in Table 1 to the IFR parameter employed
was re-examined and a summary of this exercise is provided in Table 2.
It is clear from Table 2 that the estimation of IFR using the empirical equations 2d & 3e (plotted
on Figure 2), rather than direct use of the measured IFRs to deduce Ar* values has only a
minimal impact on the COV values for Qc/Qm. It may also be inferred from Figure 2 that the
assumption in UWA-05 of a fully coring pile (i.e. IFR=1) for the database piles (most of which
had diameters less than 800mm) will lead, on average, to a 20% under-prediction of capacity.
This is in keeping with observed levels of partial plugging of database piles and suggests that
other design methods, such as ICP-05, which provide a best fit to the existing database of load
tests, but do not include an appropriate soil displacement term (such as the effective area ratio),
will over-predict the capacity of full scale offshore piles.
2
Q comp = Q b + Q s = q b 0.1 D + D f dz
4 (4a)
Q tens = 0.75 D f dz
(4b)
0 .5
0 .3 h
f = 0.03 q c A r max D ,2 tan cv
(4d)
Di 2
where A r = 1
D 2
(4e)
There is currently no load test data available for piles with diameters typically used in offshore
practice (i.e. in excess of 1.8m) and reliance is placed in design for appropriate extrapolation
from the existing database of load tests. The extrapolation to offshore piles in sand deposits
with uniform relative densities of 40% and 80% is presented on Figures 9 and 10 for various
piles with a slenderness ratio (L/D) of 35 and 75; the extrapolation is represented as the ratio of
the capacity predicted by UWA-05 to the capacity predicted by ICP-05. It is seen that both
methods provide broadly similar capacities within the pile diameter range typical of the database.
However, as the diameter increases beyond this range, the UWA-05 predictions for tension
capacity fall below those of ICP-05 and are about 70% of the ICP-05 predictions for a 2m
diameter pile. The ratio of the predictions for compression capacities4 given by UWA-05 to those
of ICP-05 are a little higher (due to the very conservative assumption of ICP-05 for end bearing),
but still fall to about 0.9 for typical offshore piles. Extrapolation of the ICP-05 method to offshore
piles is evidently non-conservative compared to UWA-05; this is primarily because ICP-05
under-estimates the effects of partial plugging of the piles in the database.
4
The jumps in these plots for compression capacities arise due to the discontinuity predicted for end bearing by ICP-
05 at Di =0.02 (Dr-30) which represents its presumed transition between a pile failing statically in a plugged and coring
mode. Note that the jump may appear as a transition because of the pile length interval selected for calculation. No
such discontinuity arises for UWA-05 which does not consider that failure in a coring mode is possible if the plug length
is greater than 5D.
Diameter (m)
0 0.5 1
1.6
1.5 D/t = 20; qc1N = 70
1.4 D/t = 50; qc1N = 70
Qt (UWA-05)/Qt (ICP-05)
1.1
1
0.9 Typical
database
0.8 piles Compression
0.7 L/D = 70
IFR = f (Di)
0.6
0 10 20 30 40 50 60 70 80 90 100
Length (m)
Figure 9: Ratio of UWA-05 to ICP-05 predictions in sands with relative densities of 40% and
80% ( qc1N=70 and 220 respectively) for pipe piles with L/D=70 in tension and compression
Diameter (m)
0 0.5 1 1.5 2 2.5
1.6
1.5 D/t = 20; qc1N = 70
1.4 D/t = 50; qc1N = 70
Qt (UWA-05)/Qt (ICP-05)
Figure 10: Ratio of UWA-05 to ICP-05 predictions in sands with relative densities of 40% and
80% ( qc1N=70 and 220 respectively) for pipe piles with L/D=35 in tension and compression
ACKNOWLEDGEMENTS
The authors are grateful to the members of the API-ISO committee for their feedback on the
UWA evaluation study, which ultimately led to the design method presented here. In particular,
the authors would like to acknowledge the numerous constructive comments raised in
discussions with the following three members of the API piling sub-committee: Mr. Harry Kolk
(Fugro Engineers B.V.), Prof. Richard Jardine (Imperial College) and Prof. Mark Randolph
(Centre for Offshore Foundations, UWA).
APPENDIX A
131 Tickfaw River 0.1 1.00 0.25 0 0.05 0 0.27 0.69 1.00
Table A4: Variances and weighting factors for closed-ended piles in tension
ID Name qc Nt t R f g-i Q CWF PWF
200 Baghdad 0.1 1.33 0.25 0.15 0 0 0.31 1.00 1.00
201 Drammen 0.1 1.00 0.25 0.15 0 0 0.31 1.00 1.00
202 Drammen 0.1 1.00 0.25 0.15 0 0 0.31 1.00 1.00
203 Drammen 0.1 1.00 0.25 0.15 0 0 0.31 1.00 1.00
204 Hoogzand 0.1 1.00 0.25 0.15 0 0 0.31 1.00 1.00
205 Hsin Ta 0.1 1.11 0.15 0 0.08 0 0.20 0.27 1.00
206 Lock&Dam 26 0.1 1.14 0.15 0 0 0 0.18 1.00 0.33
207 Lock&Dam 26 0.1 1.11 0.15 0 0 0 0.18 1.00 0.33
208 Lock&Dam 26 0.1 1.11 0.15 0 0 0 0.18 1.00 0.33
209 Ogeechee 0.1 0.87 0.15 0.15 0 0 0.23 1.00 1.00
Table A6: Variances and weighting factors for open-ended piles in compression
ID Name qc Nt t R f g-i Q CWF PWF
300 85 SFOBB 0.1 1.04 0.15 0 0.05 0 0.19 0.81 1.00
301 Drammen E18 0.1 0.95 0.15 0 0 0 0.18 1.00 1.00
302 Drammen E18 0.1 0.95 0.15 0 0.05 0 0.19 1.00 1.00
303 Drammen E18 0.1 0.95 0.15 0.15 0.05 0 0.24 1.00 1.00
306 EURIPIDES 0.1 0.98 0.15 0 0.1 0 0.21 1.00 1.00
307 EURIPIDES 0.1 0.95 0.15 0.15 0.05 0 0.24 1.00 1.00
308 EURIPIDES 0.1 1.00 0.15 0 0 0 0.18 1.00 0.50
309 EURIPIDES 0.1 0.98 0.15 0 0.05 0 0.19 1.00 0.50
310 Hoogzand 0.1 1.06 0.15 0 0 0 0.18 1.00 1.00
311 Hoogzand 0.1 1.03 0.15 0 0 0 0.18 1.00 1.00
312 Hound Poind 0.1 1.03 0.15 0 0 0 0.18 0.88 1.00
313 I-880 0.1 1.02 0.15 0 0.05 0 0.19 0.80 1.00
314 Pigeon Creek 0.1 0.96 0.15 0 0 0 0.18 0.97 1.00
315 Shanghai 0.1 1.04 0.15 0 0 0 0.18 0.84 0.50
316 Shanghai 0.1 1.05 0.15 0 0 0 0.18 0.84 0.50
317 Trans-Tokyo B 0.1 1.07 0.15 0 0 0 0.18 0.94 1.00
318 Dunkirk ZdH 0.1 1.08 0.2 0 0.05 0 0.23 1.00 1.00
Table A8: Variances and weighting factors for open-ended piles in compression
ID Name qc Nt t R f g-i Q CWF PWF
56 Los
400 Coyotes 0.1 0.87 0.15 0 0 0 0.18 0.39 1.00
401 85 SFOBB 0.1 1.10 0.15 0 0 0 0.18 0.76 1.00
405 EURIPIDES 0.1 0.96 0.15 0.15 0.75 0 0.79 1.00 1.00
406 EURIPIDES 0.1 0.87 0.15 0 0.15 0 0.23 1.00 1.00
407 EURIPIDES 0.1 1.01 0.15 0.15 0.1 0 0.25 1.00 0.50
408 EURIPIDES 0.1 0.96 0.15 0.15 0.1 0 0.25 1.00 0.50
409 Hoogzand 0.1 1.14 0.15 0.15 0 0 0.23 1.00 1.00
410 Hoogzand 0.1 1.07 0.15 0.15 0 0 0.23 1.00 1.00
411 Hound Poind 0.1 1.01 0.15 0 0 0 0.18 0.92 1.00
412 Hound Poind 0.1 0.90 0.15 0 0 0 0.18 0.94 1.00
413 I-880 0.1 1.11 0.15 0 0 0 0.18 0.71 0.50
414 I-880 0.1 1.05 0.15 0.15 0 0 0.23 0.74 1.00
415 I-880 0.1 1.08 0.15 0 0 0 0.18 0.71 0.50
416 Dunkirk ZdH 0.1 1.21 0.2 0.15 0 0 0.27 1.00 1.00
q b,plug / q c = 0.65 (0.6 0.15) FFR = 0.6 0.45 FFR (i.e. as shown on Figure A1)
1 A*rb 1 A *rb
FFR = =
(Di D)2 1 A r
1 A *rb
q b,plug / q c = 0.6 0.45
1 Ar
q b = q b,an A r + q b,plug (1 A r )
1 A *rb
= 0.6 q c A r + 0.6 0.45 q c (1 A r )
1 A r
1 A *rb
q b / q c = 0.6 A r + 0.6 (1 A r ) 0.45 (1 A r )
1 Ar
= 0.6 0.45 (1 A *rb )
= 0.15 + 0.45 A *rb
Therefore,
0.60
0.50
0.40
qb,plug/qc
0.30 Dunkirk
Euripides
Hoogzand
0.20
Indiana
Ras Tanajib
0.10 Shanghai
Tokyo
0.00
0.0 0.2 0.4 0.6 0.8 1.0
FFR
Figure A1: Ratio of qplug to the averaged CPT qc value versus FFR
The determination of q c using the Dutch technique (illustrated above) for offshore conditions is
hampered by the fact that continuous CPT profiles are often not available (e.g. when seabed
CPT methods are not employed). It is proposed that a conservative estimate of the Dutch
q c value may be estimated as:
q c = (qca + qcb)/2
where qca is the minimum value of qc over the depth interval extending from the pile tip to a
depth of between 0.7D* to 4D* below the tip (where D* = D Arb*0.5) and qcb is the average qc
value from the tip to a height of 8D* above the pile tip.
1.8 1.8
Probability Density Function
Figure A2. Weighted probability density functions for UWA-05 and ICP-05
120
Qc [UWA-05]/Qm
1.5
113
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Diameter (m)
2.5
UWA-05
average=0.98
121
cov=0.37
2
122
120
Qc [UWA-05]/Qm
1.5
113
0
0 5 10 15 20 25 30 35 40
L (m)
2.5
UWA-05
average=0.98
121
cov=0.37
2
122
120
Qc [UWA-05]/Qm
1.5
113
0
0 0.2 0.4 0.6 0.8 1 1.2
Averaged Dr,shaft (-)
1.5
205
203 207
201206
1 204 208
202
209
200
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Diameter (m)
2.5
UWA-05
average=1.02
cov=0.27
2
Qc [UWA-05]/Qm
1.5
205
207 203
201 206
1 204 208
202
209
200
0.5
0
0 5 10 15 20 25 30 35 40
L (m)
2.5
UWA-05
average=1.02
cov=0.27
2
Qc [UWA-05]/Qm
1.5
205
203
207
201 206
1 208 204
202
209
200
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
Averaged Dr,shaft (-)
1.5
306
310 318 313 309
311
307
1 308 312
317
314 301
300 316
302 315
303
0.5
0
0 0.3 0.6 0.9 1.2 1.5 1.8 2.1
Diameter (m)
2.5
UWA-05
average=0.98
cov=0.19
2
Qc [UWA-05]/Qm
1.5
306
313
310318 309
311
307
1 312
317 308
314 301
302
300 316
315
303
0.5
0
0 20 40 60 80 100
L (m)
2.5
UWA-05
average=0.98
cov=0.19
2
Qc [UWA-05]/Qm
1.5
306
318 309 310
313
311
307
1 317
312 308
301 314
302 316 315 300
303
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
Averaged Dr,shaft (-)
1.5
414
407
408
415
1 410
409
416
411
417 406
401 405
400 412
413
0.5
0
0 0.3 0.6 0.9 1.2 1.5 1.8 2.1
Diameter (m)
2.5
UWA-05
average=0.91
cov=0.23
2
Qc [UWA-05]/Qm
1.5
414
407
408
1 409 415
410
416
411
417 406
401 405
400 412
413
0.5
0
0 20 40 60 80 100
L (m)
2.5
UWA-05
average=0.91
cov=0.23
2
Qc [UWA-05]/Qm
1.5
414
407
408
415
1 416
410
409
411
417 406
400 405 401
412
413
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2
Averaged Dr,shaft (-)
I am very interested in Table 6.2 which lists the Pf for the various methods. You used a safety
factor of 1.5 and 2.0, as per API RP2A WSD. These are total safety factors and therefore are
also capturing uncertainties in the loads. In API RP2A LRFD, it is recommended to use a
resistance factor of 0.8 for operating conditions, and 0.7 for design conditions. These factors
are in terms associated with proper load factors.
It therefore seems to me that what we really would like to have is the probability of failure for the
various methods for safety factors of 1/0.8= 1.25 and 1/0.7= 1.43. Then we would properly
assess the risks associated with various methods, for the resistance part of the foundation
design.
Question: how much work would it be to update or redo Table 6.2 with safety factors of 1.25 and
1.43? If it is not too much work I would love to have the results so I can pass those on to
SC2. I would like to be able to say that the new method has Pf comparable with API-00. Only
then I feel we have the proper justification for using the updated method without changing our
safety factors.
We have provided a response to this enquiry by providing a comparison between existing API
recommendations and the proposed UWA-05 method of (a) the weighted probability density
functions on Figure B1 and (b) the relative probability of failure (Pf) plotted as a function of the
factor of safety on Figure B2. The reader is referred to Lehane, Schneider & Xu (2005) for a full
description of the procedures involved in the derivation of these figures.
Please note that:
(i) Eqns 1 & 3 and Eqns 2 & 3 are the general UWA recommendations for driven piles,
while Eqn. (4) is the simplified proposal for offshore piles where dilation and partial
plugging effects are assumed to be negligible.
(ii) However, because of time constraints, the probability density functions and Pf values for
the UWA-05 method labelled as Eqn. 4 on Figure B2 still include the contribution of
dilation to shaft capacity; this will be rectified in due course. As a consequence, the
actual Pf values for all categories of piles determined using Eqn. 4 will be lower than
those plotted.
(iii) The Pf values are not actual probabilities of failure and merely provide a means for
comparison of the predictive performance of the API-00 and UWA-05 methods for the
database piles (most of which are smaller than full scale offshore piles).
(iv) The Pf values obtained for capacity predictions using Eqn. 4 (i.e. the UWA-05
recommendation for offshore piles) are equal 5 or lower to the equivalent values
obtained by API-00 for all categories of piles. Given the comment in (ii) and estimating
that dilation effects contribute to an average of about 10% of the shaft capacity of the
database piles, it is concluded that Eqn. 4 of UWA-05 leads to substantially improved Pf
values over current recommendations.
5
Arising for closed-ended piles in compression for which API-00 significantly under-predicts the capacity of the
database piles. However the bias with respect to pile length and sand relative density for this method are such that API-
00 will tend to have higher Pf values for long piles as well as in loose sands.
Figure B1. Weighted probability density functions for API-00 and UWA-05
1.E-02 1.E-02
API-00 API-00
1.E-03 1.E-03
UWA-05: UWA-05:
Eqns 1& 3 Eqns 1&3
1.E-04 1.E-04
1 1.5 2 2.5 1 1.5 2 2.5
Factor of Safety Factor of Safety
1.E+00 1.E+00
API-00
OET
Relative Probability of Failure
1.E-02 1.E-02
OEC
1.E-03 1.E-03 API-00
UWA-05: Eqns 2&3
UWA-05: Eqn 4
1.E-04 1.E-04
1 1.5 2 2.5 1 1.5 2 2.5
Factor of Safety Factor of Safety
(a) Please confirm that the recommended method for offshore piles (equations 5) include an
upper limit on tan delta of 0.55 and determine qc at tip using the Dutch method.
Apologies for our ambiguity which we have now rectified in a new Figure 4 of the main text
(included in this file). The database analyses used measured cv values and values that were
employed in ICP-05 predictions (to ensure compatibility between the comparison of UWA-05
and ICP-05 predictions). For tests outside of the ICP-05 database, when no interface data were
available, the database analyses employed the trend line shown on Figure 4. As seen on Figure
4, this trend line differs very slightly from the recommendations of UWA-05 (which are in line
with those agreed at the API-ISO meeting in early May 2005).
The UWA-05 method proposes use of the Dutch method to calculate qc average, but accepts
averaging within 1.5D of the tip in situations where qc values in the vicinity of the tip do not vary
significantly; see note also in Appendix A which deals with the Dutch method.
Please also note that equation (5) is now equation (4) there was previously no equation (4).
(b) Are figures 9 and 10 made for the simplified UWA method or the full UWA method (i.e incl.
Dilatancy and IFR)?
Figures 9 and 10 do include dilatancy and partial plugging effects i.e. they use equations (2)
and (3) which reduce to equation (4) for typical offshore piles.
(c) Are you sure that curves in Fig. 9 and 10 are correct? I would in particular double-check the
compression curves for D/t=50 and qc1N=70. (by the way I forgot definition for qc1N; can you
provide this please?)
Yes we are sure that the curves are correct and have had them checked independently. The
jumps arise due to the discontinuity predicted for end bearing by ICP-05 at Di =0.02 (Dr-30)
which represents its presumed transition between a pile failing statically in a plugged and coring
mode. Note that the jump may appear as a transition because of the pile length interval
selected for calculation. No such discontinuity arises for UWA-05 which does not consider that
failure in a coring mode is possible if the plug length is greater than 5D.
The equation for qc1N is provided in equation 3f.
(d) I fully agree with the concern on ICP potentially being unconservative for typical offshore
piles. Therefore I find the trends in Figs 7 & 8 just as important as the statistical data (for same
reason might be good to have Figs A5 & A6 also for ICP & eventually other methods)
These plots are provided in the UWA evaluation study report.
(e) It would be good to get a "feel" for how the new method(s) would compare with the
traditional API method for typical GoM pile/soils (48 inch OD, t=1.25 inch?, I will check by
separate mail). Would it be possible to have pile capacity curves made at short notice for UWA,
ICP and traditional API (and eventually other methods) for site C given in OTC 1986 paper 5104
("Results of recent CPT in GoM" by Focht et al.)?
We include in Figures B4 & B5, predictions (for compression loading) made for 72 and 96
piles with a wall thickness of 1.75 currently being considered for a site in the Gulf of Mexico.
The design CPT profile for this project is provided in Figure B3.
We also include (on Figure B6 & B7) plots of ratios of predictions made using UWA-05 with
those given by API-00 for idealized sand profiles (Dr=40% qc1N=70 and Dr=80% qc1N=220).
10
20
Depth (m)
30
40
50
60
72" pile
10
Pile Length (m)
20
30 API-00
Fugro-05
ICP-05
40
NGI-04
UWA-05
50
Qend, C (MN)
0 5 10 15 20 25 30
0
10
20
Pile length (m)
30
40 API-00
Fugro-05
ICP-05
50 NGI-04
UWA-05
60
Qshaft, C (MN)
0 10 20 30 40 50
0
API-00
Fugro-05
10 ICP-05
NGI-04
UWA-05
Pile length (m)
20
30
40
50
0 20 40 60 80 100 120
0
96" pile
10
Pile length (m)
20 API-00
Fugro-05
ICP-05
30
NGI-04
UWA-05
40
50
Qend, C (MN)
0 10 20 30 40 50
0
10
Pile length (m)
20
30
API-00
Fugro-05
40 ICP-05
NGI-04
UWA-05
50
Qshaft, C (MN)
0 10 20 30 40 50 60
0
10
Pile length (m)
20
30 API-00
Fugro-05
ICP-05
40 NGI-04
UWA-05
50
1.50
1.00
0.50
L/D=70 Compression
0.00
0 20 40 60 80 100 120
Length (m)
D/t=20; qc1N=70
2.00 D/t=50; qc1N=70
D/t=20; qc1N=220
D/t=50; qc1N=220
1.80
Q end [UWA-05] /Qend [API-00]
1.60
1.40
1.20
1.00
0.80
L/D=70 Compression
0.60
0 20 40 60 80 100 120
Length (m)
2.50
D/t=20; qc1N=70
D/t=50; qc1N=70
2.00 D/t=20; qc1N=220
Qshaft [UWA-05]/Qshaft [API-00]
D/t=50; qc1N=220
1.50
1.00
0.50
L/D=70 Compression
0.00
0 20 40 60 80 100 120
Length (m)
Figure B6: Ratio of UWA-05 to API-00 predictions for piles with L/D=70 in compression
1.5
1.0
0.5
L/D=35, compression
0.0
0 20 40 60 80 100 120
Length (m)
3.0
D/t=20; qc1N=70
D/t=50; qc1N=220
2.0
1.5
1.0
0.5
L/D=35, compression
0.0
0 20 40 60 80 100 120
Length (m)
3.0
D/t=20; qc1N=70
D/t=20; qc1N=220
D/t=50; qc1N=220
2.0
1.5
1.0
0.5
L/D=35, compression
0.0
0 20 40 60 80 100 120
Length (m)
Figure B7: Ratio of UWA-05 to API-00 predictions for piles with L/D=35 in compression
Figure B8: Borehole and CPT data at Site C from GoM given in Focht et. al. (1986)
qc (MPa)
0 10 20 30 40 50 60 70
30
40
50
depth (m)
60
70
80
90
100
40
50
API-00
Tip depth (m)
Fugro-05
60 ICP-05
NGI-04
70 UWA-05, Eqn 4
80
90
100
Qend, C (MN)
0 10 20 30 40 50 60
30
40 API-00
Fugro-05
50 ICP-05
NGI-04
Tip depth (m)
UWA-05, Eqn 4
60
70
80
90
100
Qshaft, C (MN)
0 10 20 30 40 50 60
30
40
API-00
Fugro-05
50
ICP-05
Tip depth (m)
NGI-04
60 UWA-05, Eqn 4
70
80
90
100
Figure B10: Predictions for typical 48 pipe pile installed at Site C in GoM
40
50 API-2000 UWA
API-1984 (Gradation Based)
Depth (m)
70
80
90
100
Figure B11: Predictions using API reported by Focht et al. (1986) compared with UWA
predictions derived using API-00