Geomechanics for Energy and the Environment 6 (2016) 70–80

Contents lists available at ScienceDirect

Geomechanics for Energy and the Environment

journal homepage: www.elsevier.com/locate/gete

A parametric study of different analytical design methods todetermine the axial bearing capacity of monopilesJohannes Labenski ∗, Christian MoormannInstitute for Geotechnical Engineering (IGS), University of Stuttgart, Pfaffenwaldring 35, 70569 Stuttgart, Germany

h i g h l i g h t s

• A sensitivity analysis is carried out to show the influence of different parameters.• The sensitivity analysis shows a variation between the capacities of 1.7 to 16.4.• A comparison with five different field tests shows the performance of the methods.• For some methods, there is good agreement between predicted and measured capacities.• Other methods show a strong deviation to the measured capacity.

a r t i c l e i n f o

Article history:Received 31 August 2015Received in revised form 4 March 2016Accepted 5 March 2016Available online 12 April 2016

Keywords:MonopilesAxial bearing capacityAnalytical design methodsCPT based design methods

a b s t r a c t

In Europe many offshore wind power parks have already been realized and more areplanned. Oftenmonopiles are used as foundation forwind power plants. The determinationof the axial pile capacity of monopiles is among the important issues for offshore energyprojects. Despite progress made so far in this field, there are still big challenges faced byresearchers. Among them is to provide guidance to assist engineers in the selection ofan appropriate design method for the soil condition and pile configuration of interest.First, different state of the art methods (API, Fugro-05, ICP-05, NGI-05 and UWA-05)together with two new methods (HKU-12 and a German approach) to determine the axialbearing capacity based on the cone resistance of the soil are presented. Afterwards thereis a summary of different approaches to determine the average cone resistance value.Furthermore a sensitivity analysis is carried out to show the influence of the pile diameterand the cone resistance on the predicted axial bearing capacity of the different designmethods. In the last part, published field tests are used to compare the presented designmethods and to identify their potential for reliable predictions for engineering practice. Inaddition to existing publications, these comparisons contain the two new design methodsto determine the axial bearing capacity.

© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Monopiles (open steel pipe piles in general) are becom-ing increasingly important in constructing offshore struc-tures, especially as foundation for offshore wind powerplants. In order to replace fossil and nuclear energy with

∗ Corresponding author.E-mail addresses: johannes.labenski@igs.uni-stuttgart.de

(J. Labenski), christian.moormann@igs.uni-stuttgart.de (C. Moormann).

http://dx.doi.org/10.1016/j.gete.2016.03.0012352-3808/© 2016 Elsevier Ltd. All rights reserved.

renewable energy sources, more and more wind powerparks are being constructed. The determination of the ax-ial pile capacity ofmonopiles is one of the important issuesof offshore energy projects. Despite the progress made sofar in this field, researchers are still faced with many chal-lenges. Among them is the provision of guidance to assistengineers in the selection of an appropriate designmethodbased on the soil condition and pile configuration of inter-est. Hence, this paper discusses an important issue relatedto geomechanics in energy and environment.

J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80 71

Fig. 1. Axial bearing behaviour of an open steel pipe pile: (a) fully coring, (b) partially plugged, (c) fully plugged.

The axial bearing capacity of open steel pipe piles ingranular soils depends, among others, on the installationmethod, the pile diameter, the embedment length of thepile, the interface friction angle between the pile andthe soil, and the relative density of the soil. During theinstallation, a plug may be formed inside the pile. A plugcan be visualized as a spatial bracing of soil between theinner surface of the pile, cf. Lüking.1 Such a plug influencesthe axial bearing capacity significantly.

In Fig. 1, the principle axial bearing behaviour of anopen steel pipe pile is shown. In the case of a pile with noplugging effect, the axial bearing capacity is achieved by aninternal and external frictional resistance and a resistancemobilized by the tip pressure of the annulus (a). In thecase of a partially plugged pile (b), there is an additionalresistancemobilized by the plug inside the pile, though theinternal frictional resistance is lowered compared to thepile without a plug. In the case of a fully plugged pile (c), anopen pile behaves like a closed pile, with a high resistancemobilized by the plug and no internal frictional resistance.Observations by Jardine et al.2 show, that an open steelpipe pile is likely to form a plug when its diameter is 1.5 mand below. This means that in the case of large diameteropen steel pipe piles, no plugging effect is expectedand the bearing behaviour presented in Fig. 1(a) can beanticipated.

A parameter to define the degree of plugging is theIncremental Filling Ratio (IFR) introduced by Brucy et al.3It is defined as follows

IFR =1h1L. (1)

In real applications, for example in the installation ofmonopiles, it is hard to determine the IFR because theheight of the plug as well as the penetration depth of thepile need to be continuously monitored. For this reasonPaik et al.,4 introduced the Plug Length Ratio (PLR). Itsdefinition is similar to the IFR except that the PLR is notan incremental value and is only measured once at the endof the installation process.

PLR =hL. (2)

In scientific papers, different analytical approachesexist to determine the axial bearing capacity, e.g. Randolphet al.5,6 and Paik et al.4 Nowadays the API7 and the DINEN ISO 19902:2014-18 operate as technical standards

for the design of monopiles. Both are referring to CPTbased design methods. These CPT based methods shouldlead to a more realistic and economical axial design ofthe piles because the axial bearing capacity is directlylinked to the cone resistance of a CPT. The referredmethods are the Fugro-05,9 ICP-05,2 NGI-0510 and UWA-05.11 More recent methods to determine the axial bearingcapacity are the HKU-12 method12,13 and an approachrecommended by EA-Pfähle.14 What all these methodshave in common is that they are only valid for impact-driven piles. Lammertz15 proposed a method to determinethe axial bearing capacity of vibratory-driven piles.

The objective of this paper is to perform a comparisonamong the different analytical axial design methods formonopiles. A sensitivity analysis with regard to the pilediameter and the cone resistance shows the quality of themethods in a parametric study. A comparison with theresults from different field tests shows the quality of themethods under real conditions. The result of this papercould serve as a guide to help the engineer select a methodthat suits his individual case.

2. Analytical design methods

All the respectively predicted capacities and resistancesare ultimate values assuming a settlement of the pile of 10%of the diameter, i.e. s

D = 0.1. To calculate the total axialbearing capacity,Qtot , of open steel pipe piles, the followinggeneral equation is used for almost all of the presentedmethods

Qtot = Qb + Qs = 0.25 πD2 qb + πDτs (z) dz. (3)

Fig. 2(a) and (b) can be used to determine the qualitativedistribution of Qb and Qs before the relative settlement ofthe pile reached 0.1D. A summary of the design equationsto determine qb and τs of the CPT based design methodscan be found in Tables 3 and 4.

2.1. API

The API 2A-WSD7 is a technical standard that providesguidance in the design of offshore structures. Among otherthings, a recommendation on the determination of theaxial bearing capacity is given. This recommendation canalso be found in DIN EN ISO 19902:2014-01.8 The values of

72 J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80

(a) Q–z curve. (b) t–z curve.

Fig. 2. Q–z and t–z curves according to API and ISO.

Table 1Recommended values according to API7 respectively DIN EN ISO 19902:2014-01.8

Relative density Dr [%] Soil classification Nq qmax [MPa] β τmax [kPa]

Very loose 0–15 Sand Not applicableLoose 15–35 SandLoose 15–35 Sand–silt

CPT based methods recommendedMedium dense 35–65 SiltDense 65–85 Silt

Medium dense 35–65 Sand–silt 12 3 0.29 67

Medium dense 35–65 Sand 20 5 0.37 81Dense 65–85 Sand–silt

Dense 65–85 Sand 40 10 0.46 96Very dense 85–100 Sand–silt

Very dense 85–100 Sand 50 12 0.56 115

qb and τs can be determined by the following equations

qb = Nq σ′

v0 (z) ≤ qmax (4)

τs (z) = β σ ′

v0 (z) ≤ τmax. (5)

The values can be determined according to Table 1.qb and τs are limited due to the fact that the end bearing

and shaft friction do not linearly increase with depth andoverburden pressure, especially for long piles.

2.2. ICP-05

The ICP-05 method is a CPT based method proposedby Jardine et al.2 and it intends to predict the axial pilecapacity which may be mobilized in slow maintainedload tests on previously unfailed piles conducted 10 daysafter installation. Jardine et al.16 however, stated thatthe axial pile capacity predicted by the ICP-05 methodis more or less the measured capacity after 100 days ofinstallation.

This method states equations to determine qb and τs.It is based on field tests of closed and open ended pileswith pile diameters of up to 0.76 m driven into sand andclay. The method distinguishes between a fully pluggedand unplugged pile. In the case of an unplugged pile theunit end bearing resistance is directly linked to the averagecone resistance and the internal shaft friction is neglected.The external unit shaft friction is determined by using theCoulomb failure criterion.

The radial effective stress after installation and equal-ization, σ ′

rc , is based on the correlation between thelocal cone resistance, the normalized effective verticalstress and the normalized distance above the tip of the pile.The change in radial stress due to dilation, σ ′

rd, is based ona cylindrical cavity expansion analogy.

2.3. UWA-05

The UWA-05 method is a CPT based method proposedby Lehane et al.11 which is based on findings by Xu et al.17

Lehane et al.18 published an updated version of the UWA-05 method with an extrapolation to large diameters. TheUWA-05 method states equations for qb and τs. It alsodivides the unit end bearing resistance into the bearingresistance of the annulus, qb, an, and the plug, qb, plug . TheUWA-05 method is based on a database of 13 full-scalestatic load tests done by driving open ended piles insiliceous sand at seven different sites. For each site, thecone resistance profiles of a CPT are available. The diameterof most piles varied from 0.324 to 0.914 m. One pile hada diameter of 2.0 m. The embedment lengths of the pilesvaried from 5.3 to 79.1 m. The IFR was partially measuredduring the installation of the piles. If the IFR could notbe measured, the PLR was measured at the end of theinstallation process.

J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80 73

Fig. 3. Constant volume interface friction angle according to Jardine et al.2 and Lehane et al.11

2.4. HKU-12

The HKU-12 method is a CPT based method proposedby Yu and Yang.12,13 This method uses the same approachas the UWA-05 method, dividing the unit end bearingresistance into a resistance of the annulus and the pluginside the pile.

The proposed equation for the resistance of the annulusis based on centrifuge tests performed by De Nicola andRandolph.19 The centrifuge study investigates the effectof plugging inside open steel pipe piles driven into sand.The piles had prototype embedment lengths of 15–18 m,diameter of 1.6 m and wall thickness of 0.055 m. The testswere performed in loose, medium dense and very densesand. With these results, Yu and Yang derived an equationwhich is directly dependent on the slenderness of the pile.There is no direct dependency on the relative density of thesoil what is reasoned with the direct correlation with thecone resistance of the CPT.

The proposed equation for the resistance of the plug isbased on tests by De Nicola and Randolph,19 Lehane andGavin,20 and Lee et al.21 Lehane and Gavin performed smallscale tests with open and closed ended steel pipe pilesjacked in loose sand. The piles had diameters of 40 mmand 114 mm respectively and their embedment lengthsvaried from 1 to 1.55 m. Lee et al. performed small scaletests with open steel pipe piles driven into loose, mediumdense and very dense sand. The piles had a diameter of42.7 mm and embedment lengths of 0.25–0.76 m. Thederived equation is based on an investigation of the plugresistance dependence on the PLR.

To calculate the external shaft friction, the HKU-12method assumes a relative depth in the soil equal to ξL,where 0 ≤ ξ ≤ 1 and ξ = 1 describes the co-ordinateat the pile toe, whereas ξ = 0 describes the co-ordinate atthe ground level. Utilizing this approach, the formula of theshaft resistance Qs of Eq. (3) has to be modified as follows

Qs = πDL 1

0τs (ξ) dξ . (6)

2.5. Fugro-05

The Fugro-05method, proposed by Kolk et al.,9 is basedon a database of 45 pile load tests at ten different sites. Thepiles, open and closed ended, were all installed in siliceoussand with diameters varying from 0.356 to 0.763 m andpile slenderness, L/D, varying from 5 to 100.

2.6. NGI-05

The NGI-05 method, proposed by Clausen et al.,10 isbased on a database of 85 pile tests in sand at 35 differentsites. The piles, open and closed ended, were driven intoloose to very dense sand with embedment lengths of up to90m. Contrary to the other CPT based designmethods, thismethod utilizes the relative density, Dr , instead of the pilediameter to determine the unit end bearing resistance.

The unit end bearing capacity distinguishes betweenplugged and coring behaviour, although there is no specificcriterion for distinction. The unit end bearing resistanceis taken as the smallest of the coring and the pluggedresistance.

2.7. German approach—EA-Pfähle

TheGermanapproach, proposedby Lüking andBecker,22is based on the EA-Pfähle14 and EAU23 recommendationsand is documented in Moormann and Kempfert.24 Themethod tries to capture the strengths of both technical rec-ommendations and combines them into one method. BothEA-Pfähle (‘‘Recommendations on Piling’’), and EAU (‘‘Rec-ommendations onwaterfront structures, harbours andwa-terways’’) were elaborated by the German GeotechnicalSociety (DGGT) and are widely used in Germany and partlyin other European countries.

The proposed method is applicable for open steelpiles with various diameters. Basically, the method distin-guishes among three cases:

1. Fully plugged with D ≤ 0.5 m.2. Full coring with D ≥ 1.5 m.3. Partially plugged with 0.5 m < D < 1.5 m.

74 J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80

Table 2Recommended values for the EA-Pfähle approach according to Lüking and Becker.22 .

qc [MPa] s [mm] qb, an [kPa] qb, plug [kPa] τs [kPa] τis [kPa]Case 1 Case 2

7.5s∗sg 15–25 15–20 5–100.035D 3900–7500 1200–33000.100D 7500–9000 2250–4000 25–35 20–30 10–15

15s∗sg 35–50 30–45 10–200.035D 7900–11500 2100–40000.100D 15000–18000 4000–6250 50–70 40–60 20–30

25s∗sg 40–70 35–60 15–250.035D 10300–16300 2500–47500.100D 20000–25000 4750–7250 60–90 50–80 22–40

s∗sg [mm] = 0.05

j ηs τs,j(s∗sg ) As,j [MPa] ≤ 10 mm

The EA-Pfähle approach is based on a database of 113pile load tests of open steel piles with diameters varyingfrom 0.32 to 1.42 m. The piles were mainly driven intomedium dense to dense sand with embedment lengths of7 to 32 m.

Case 1 assumes a total bearing resistance resulting fromthe frictional resistance of the outer shaft, the resistanceof the annulus and an additional resistance of the plug. Incase 2, the additional resistance of the plug is dropped andan additional inner shaft friction is utilized. However, it isassumed that the inner shaft friction is not acting at theupper 20% of the embedded part of the pile. Furthermore,the inner shaft friction is reduced to 50% of the outer shaftfriction. This reduction is justified by the uncertainties inthe determination of the internal shaft friction as well asfield measurements by Ko and Sangseom.25 In both casesthe resistance of the annulus is directly linked to the coneresistance of the CPT of the soil.

For case 1, the fully plugged pile, the equation of thetotal axial capacity isQtot, case 1(s) = ηb, plug qb, plug Aplug + qb, an Aan

+

j

ηs τs,j As,j (7)

where ηb, plug = 2.52 e−1.85D and ηs = 1.53 e−1.85D bothwith D in [m].

For case 2, the full coring pile, the equation of the totalaxial capacity is

Qtot, case 2(s) = qb, an Aan +

j

ηs τs,j As,j

+

j

τis,j Ais,j (8)

where the index, i, is linked to the inner part of the pile.For case 3, the partially plugged pile, the equation of the

total axial capacity isQtot, case 3(s) = ψ Qtot, case 1 + χ Qtot, case 2 (9)where ψ and χ are calculation factors which can bedetermined according to Fig. 4.

Themissing values of Eqs. (7) and (8) can be determinedaccording to Table 2. For each parameter there are twovalues. The lower bound describes the 10% quantile,whereas the upper bound describes the 50% quantile.Furthermore, the proposed method is only valid for a coneresistance of qc ≤ 25 MPa.

Fig. 4. Calculation factors ψ and χ according to Lüking and Becker.22

3. Determination of qc, avg

The determination of qc, avg plays a significant role in allthe CPT based designmethods. DIN EN ISO 19902:2014-01suggests taking the average 1.5D above and below the piletip

qc, avg,1.5D =

L+1.5DL−1.5D qc(z) dz

3D

. (10)

Another method is the Dutch method by Schmert-mann.26 The Dutchmethod assumes a zone of 0.7D to 4.0Dbelow and 8D above the pile tip. This method takes thevalue 1 as an average below the pile tip of all values of qc(z)until min qc(z). Afterwards, the average of the value 1 andmin qc(z) is taken as qc1. The average above the pile tip qc2is taken as the average of all minima in the range of 8D.The average considered is then obtained by the followingequation

qc, avg, Dutch =qc1 + qc2

2. (11)

Amore recent approachwas proposed byYu andYang12within the HKU-12 method. Assuming hd is the depth of

J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80 75

Table 3Summary of the design equations for the unit endbearing resistance of theCPT basedmethods for open ended steel piles impact driven into siliceoussand.

Method Unit end bearing resistance

Fugro-05 qb = 8.5 pa

qc, avgpa

0.5A0.25r

HKU-12qb = qb, an Ar + qb, plug (1 − Ar )

qb, an = qc, avg · min

1.063 − 0.045

LD

0.46

qb, plug = 1.063 qc, avg exp (−1.933 PLR)

ICP-05qb = qb, plugged if

d < 0.02 (Dr − 0.30)d

DCPT< 0.083

qcpa

qb, coring = Ar qc, avg

qb, plugged = qc, avg · max

0.5 − 0.25 log

D

DCPT

0.15Ar

NGI-05

qb = minqb, coring , qb, plugged

qb, coring = qb, an Ar + qb, plug (1 − Ar )

qb, an = qc, tipqb, plug = 12 τs, avg L

π d

qb, plugged =0.7 qc, tip1+3 D2

r

UWA-05 qb =0.15 + 0.45 A∗

r

qc, avg

Remark: qc, avg has to be evaluated with D∗= D A∗

0.5r

General

pa = 100 kPa

Ar = 1 − dD

2A∗r = 1 − FFR

dD

2Remark: FFR = IFR averaged over the last 3D of pilepenetration

Dr (z) = 0.4 ln

qc (z)22(σ ′

v0(z) pa)0.5

Remark: Dr values greater 1.0 are possible and shouldbe used

the pile in the end-bearing layer, the ranges above, namedA, and below, named B, the pile tip can be determinedaccording to Table 5. At first the geometric mean of bothzones has to be determined according to the equation

MA or MB = n√qc1 qc2 · · · qci · · · qcn. (12)

To get the average cone resistance, the followingequation is used

qc, avg, HKU =

0.5 (MA + MB) if MA ≤ MBMB ifMA > MB.

(13)

4. Sensitivity analysis

To gain a first impression about the relative perfor-mances of themethods presented in Section 2, a sensitivitystudy was carried out. An open steel pipe pile driven intoa sand layer is assumed. The pile has a wall thickness of40mm and an embedment length of 20m. For the purposeof this study, qc and D have been varied. Three differentqc profiles are assumed: 7.5, 15 and 25 MPa. For the sake

Table 4Summary of the design equations for the unit shaft friction of the CPTbased methods for open ended steel piles impact driven into siliceoussand.

Method Unit shaft friction

Fugro-05 τs(z) = 0.08 qc(z)σ ′v0(z)pa

0.05 L−zR∗

−0.9for L−z

R∗ ≥ 4

τs(z) = 0.08 qc(z)σ ′v0(z)pa

0.05 L−z4 R∗

4−0.9 for

L−zR∗ < 4

HKU-12τs(ξ) =

σ ′rc(ξ)+1σ ′

rd(ξ)tan δcv

σ ′rc(ξ) = ρ0.3 qc(ξ) min

0.03

LD

−0.5

(1 − ξ)−0.5

0.021

1σ ′r (ξ) = 4 G (ξ) 1r

(D2−PLR d2)0.5

ICP-05τs(z) =

σ ′rc(z)+1σ ′

rd(z)tan δcv

σ ′rc(z) = 0.029 qc(z)

σ ′v0(z)pa

0.13 max

L−zR∗ , 8

−0.38

1σ ′

rd(z) = 4 G(z) 1rD

NGI-05

τs(z) = max

zLpa FDr Fload Ftip Fmat Fsig

0.1 σ ′

v0

FDr = 2.1 (Dr − 0.1)0.17

Fload = 1.3 for compression

Ftip = 1.0 for an open ended pile

Fmat = 1.0 for steel

Fsig =

σ ′v0pa

−0.25

UWA-05τs(z) =

σ ′rc(z)+1σ ′

rd(z)tan δcv

σ ′rc(z) = 0.03 qc(z)

A∗r,s(z)

0.3 max

L−zD , 2

−0.5

1σ ′

rd(z) = 4 G(z) 1rD

General

R∗=

R2

− r20.5

δcv can be determined according to Fig. 3

1r = 0.00002 m

ρ = 1 − PLR dD

2G(z) =

qc(z)0.0203 + 0.00125 η(z)− 1.21 10−5 η(z)2

−1

η(z) =qc (z)

(σ ′v0(z) pa)

0.5

A∗r,s(z) = 1 − IFR

dD

2of simplicity and to preclude potential error sources, qc isdistributed as a constant over the whole height of the sandlayer. The external diameter is varied from 0.5 to 5 m, al-thoughmostmethodswere verified against field testswithpile diameters≤1.5m. In Figs. 5–7, the total calculated ax-ial capacity, Qtot , with regard to the external pile diameteris shown.

With increasing qc , the predicted capacity increases.Furthermore, with increasing qc , the gradient of the curveof the NGI-05 method increases up to a diameter of2 m. Afterwards, the gradient slightly decreases and thenremains constant. Comparing the gradient of the curvesof the other methods, it was noted that the gradient ofthe curves of the Fugro-05, HKU-12 and UWA-05 methodsincreases with the pile diameter.

From the sensitivity analysis, it can be seen that withincreasing pile diameter, the predicted capacities vary byfactors of 1.84 (qc = 7.5 MPa), 1.75 (qc = 15 MPa), and

76 J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80

Table 5Zones above and below the pile tip according to Yu and Yang.12 .

Case Soil condition Range above pile tipA B

hd < 8D Extreme variation in qc 8D 1DOther situations 1.5D 1.5D

hd ≥ 8D Embedded in sand of Low compressibility 2D 4.5DHigh compressibility 1D 2.5D

Fig. 5. Total predicted axial bearing capacity with an assumed qc of7.5 MPa.

Fig. 6. Total predicted axial bearing capacity with an assumed qc of15 MPa.

1.70 (qc = 25 MPa) for D = 0.5 m up to factors of 16.40,12.56, and 13.10 for D = 5.0 m. For a D of 1.5 m, thefactors of variation are 6.40, 5.09, and 5.47. This strongvariation with increasing diameter indicates already thelimits of the different design methods. None of them were

Fig. 7. Total predicted axial bearing capacity with an assumed qc of25 MPa.

validated against pile load tests with a pile diameter largerthan 2.0 m.

In comparison to offshore conditions, comparable lowqc values were chosen for this sensitivity analysis. Allof the presented methods in Section 2 were developedfor different applications. Some of them were developedfor application in the offshore environment, revwhilesome revothers were developed for other applications butare used in the offshore environment. To qualitativelycompare the presented design methods, the range of qcvalues was chosen in a way that all methods could be usedin their validated range of site conditions.

5. Comparison with published field test results

To show the revperformances of the proposed methodsunder real conditions, they were compared against theresults of published field tests. These well documentedfield tests, containing the measurement of the totalaxial capacity of a driven pile in non-cohesive soil, onlyinvestigated piles with diameters less than 2 m. Thus,the maximum investigated pile diameter used was 2 m.In order to perform this comparison, four pile load testsdocumented in Yang et al.27 were used. Instead of asimplification with a constant cone resistance profile as inSection 4, the real measured cone resistance profiles of thedifferent sites were utilized (cf. Fig. 8). The four field testsare the following:

J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80 77

Table 6Average qc at the pile tip (MPa).

qc, avg,1.5D qc, avg, Dutch qc, avg, HKU

Mobile Bay 10.0 9.7 5.6EURIPIDES 49.0 50.7 52.0Hound Point 21.5 16.8 21.2Tokyo Bay 28.5 9.3 14.8

(a) Mobile Bay, USA: A pile with an external diameterof 0.324 m and an embedment length of 15.2 m.There was no set-up time before the pile load testdocumented. The soil consists of silty and fine sand.The qc profile linearly increases up to a maximum of40 MPa, cf. Fig. 8(a).

(b) EURIPIDES, Netherlands: A pile with an externaldiameter of 0.763 m and an embedment length of38.7 m. The set-up time before the pile load test was 2days. The first 20mof the soil consists of fine sandwithpartial clay inclusions and a soft clay layer, followedby the bearing layer. The bearing layer consists ofmedium to very dense silty sand. The qc profile linearlyincreases in the bearing layer up to a maximum of60 MPa, cf. Fig. 8(b).

(c) Hound Point, Scotland: A pile with an externaldiameter of 1.22 m and an embedment length of 26 m.The set-up time before the pile load test was 21 days.The soil consists of a 17m thick clay layer followed by asandy gravel layer which acts as the bearing layer. Theqc profile shows large variations, oscillating from 5 to40 MPa, cf. Fig. 8(c).

(d) Tokyo Bay, Japan: A pile with an external diameter of2.0 m and an embedment length of 30.6 m. The set-up time before the pile load test was 52 days. The soilconsists of a 4m thick loose alluvial sand layer followedby a dense sand containing thin layers of sandy clay.The qc profile oscillates between 5 and 60 MPa, cf.Fig. 8(d).

The average cone resistance at the pile tip calculatedwith the methods presented in Section 3 is shown inTable 6. To ensure a better comparability of the designmethods, qc, avg,1.5D was utilized for the calculationspresented in this section. The predicted capacities of theEA-Pfähle approach were calculated by applying the 50%value. Furthermore, the design values given in Table 2were linearly extrapolated up to the cone resistance ofthe investigated sites, even though the design method isnot validated against such conditions. The same was donewith the other design methods; for example, the UWA-05method was validated against one pile load test with a pilediameter of 2.0 m, whereas the ICP-05 method was onlyvalidated against a maximum pile diameter of 0.76 m.

In Fig. 9, the normalized predicted capacities arepresented, i.e. the predicted capacity, Qtot , is divided bythe measured capacity, Qm, at a settlement of 0.1D. InFig. 10, the proportion of the end bearing, Qb, to the totalcapacity, Qtot , is shown. In Fig. 11, the proportion of theshaft resistance, Qs, to the total capacity, Qtot , is shown.From the figures, the following points can be extracted:

Fig. 8. Cone resistance profiles of (a) Mobile Bay. (b) EURIPIDES. (c)Hound Point and (d) Tokyo Bay according to Yang et al.27

Fig. 9. Predicted axial capacities in relation to the measured axialcapacity at a settlement of 0.1 D.

78 J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80

Fig. 10. Predicted end bearing in relation to the total predicted capacity.

Fig. 11. Predicted shaft resistance in relation to the total predictedcapacity.

(a) With increasing diameter, the range of total predictedaxial capacities increases. For the Mobile Bay and theEURIPIDES site, the range is rather small in comparisonto the Hound Point and Tokyo Bay site. The onlyexception is the result of the EA-Pfähle approach at theMobile Bay site. This observation is in accordance withthe findings in Section 4.

(b) The division of the total resistance into end bearingresistance and shaft friction shows a significantvariation among the methods. As in (a), a dependencyon the diameter visible. For the Mobile Bay andEURIPIDES sites, the ratio Qb/Qtot varies from 0.25 to0.85 and 0.12 to 0.75 respectively. At the Hound Pointsite, most of the methods predicted a ratio of 0.75. Theexceptions are the ICP-05 method and the EA-Pfähleapproach with a ratio of 0.4. The Tokyo Bay site showsa similar result. Most methods show a ratio of 0.5,while the ICP-05 predicts a ratio of 0.2 and the API aratio of 0.7. Even though the scatter at the Mobile Bayand EURIPIDES sites is large, a trend of a decreasingportion of the end bearing resistance onto the totalcapacity with increasing diameter is visible. This is inaccordancewith the explanation given in Section 1 thatwith increasing pile diameter, the probability of a plugbeen formed decreases and thusmore andmore load iscarried by the shaft resistance of the pile.

(c) It is evident that the predicted capacities have a certainscatter. The Mobile Bay field test gives the lowest

Table 7Values of Qtot/Qm for the field tests presented in Fig. 9.

MobileBay

EURIPIDES HoundPoint

Tokyo Bay

API 0.55 0.89 1.13 1.25Fugro-05 0.70 1.17 1.55 1.32EA-Pfähle 1.56 0.72 0.27 0.32HKU-12 0.65 0.94 0.96 0.94ICP-05 0.62 1.00 0.65 0.85NGI-05 0.85 0.71 1.31 0.82UWA-05 0.71 1.14 1.09 0.93Min 0.55 0.71 0.27 0.32Max 1.56 1.17 1.55 1.32Mean 0.80 0.94 0.99 0.92Standarddeviation

0.32 0.17 0.39 0.30

Median 0.70 0.94 1.09 0.93

Table 8Mean values, standard deviation and COV of Qtot/Qm for the investigateddesign methods.

Mean Standard deviation COV

API 0.95 0.27 0.28Fugro-05 1.18 0.31 0.26EA-Pfähle 0.72 0.52 0.72HKU-12 0.87 0.13 0.15ICP-05 0.78 0.15 0.19NGI-05 0.92 0.23 0.25UWA-05 0.97 0.17 0.18

scatter, whereas the Hound Point field test shows abandwidth from 0.27 to 1.55. This may be because ofthe fact that the true axial bearing capacity of drivenpiles increases over time. This effect is the set-up effect,cf. Gavin et al.28 This set-up effect comes along withtwo potential error sources. As a first source, the timeof the pile load test can be identified. It will make adifference if this test is carried out immediately after,a few days after or even 100 or more days after pileinstallation. The second source is partially combinedwith the first source. All of the design methods arebased on pile load tests of different field tests. Thismeans that all of these methods contain an errorrelative to the set-up effect. Jardine et al.16 stated thatthe predicted capacity of the ICP-05 method is moreor less the capacity of the piles after 100 days. Thiscould be one reason for the high range among all of thedifferent predicted capacities.

(d) By further taking into account the set-up effect, itshould be noted that the lowest scatter of the predictedcapacities was observed at the EURIPIDES site and theyare all relatively close to the measured capacity. Thisis astonishing as the pile load test was carried outonly two days after pile installation and thus, the axialpile capacity measured after 100 days should be muchlarger. It should be further noted that with increasingtime of the pile load test after the pile installation, thepredicted capacities show an increasing scatter.

Comparing the individual values of Qtot/Qm, shown inTable 7, and the mean values together with the standarddeviation andCOV, shown in Table 8, of the different designmethods against each other, gives rise to the followingnoteworthy points:

J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80 79

(a) In the mean, the UWA-05, API and NGI-05 methodsshowed the best agreement with the measured pilecapacities. Theworst agreementwas achievedwith theEA-Pfähle approach.

(b) By taking into account the COV of the different designmethods, the new HKU-12 method showed the lowestscatter followed by the UWA-05 and ICP-05 methods.The largest scatter was again achieved with theEA-Pfähle approach.

(c) As in Section 4, the Fugro-05 method tends to predictthe highest capacities while the EA-Pfähle approachtends to predict the lowest capacities.

Schneider et al.29 and Gavin et al.30 conducted similarinvestigations. Among other methods, they compared theFugro-05, ICP-05, NGI-05 and UWA-05 methods againsteach other and demonstrated their relative performance.As shown in Table 8, they also determined a mean valueand COV. The results of Lehane et al. are in good agreementwith the results obtained here. The values of the meanand COV are comparable. The results of Gavin et al. differfrom the results in Table 8, especially regarding the COVwhere they obtained larger values, e.g. 0.41 for the NGI-05method. This may be due to the fact that Gavin et al. wereinvestigating the tension shaft resistance of open steel pipepiles to assess the relative performance of different designmethods.

As indicated in Section 3, the determination of qc, avgplays a significant role in the calculation of the axialcapacity. Table 6 gives an overview and impression ofpossible differences. It can be seen that the values showdifferences of up to 50% for different sites, especially ifthe measured qc profile oscillates strongly with a largescatter. The different methods to determine qc, avg showdifferent results. For the four different sites used in thiscomparison, qc, avg,1.5D often shows the highest value.Besides the Fugro-05 method, the end bearing resistanceof all of the CPT based methods depends linearly on qc, avg .The end bearing resistance calculated with the Fugro-05method has a dependency of

qc, avg

0.5. As indicated at thebeginning of this section, qc, avg,1.5D was initially applied toall of the design methods. By taking the different values ofqc, avg into account when interpreting the calculated pilecapacities, the following points are noted:

(a) At the Hound Point site, the calculated pile capacitieswould decrease by utilizing qc, avg, Dutch or qc, avg, HKUand thus would lead to worse results.

(b) At the EURIPIDES site, the influence on the calculatedpile capacities would be rather small. The values ofqc, avg vary only in a range of 49–52 MPa. Furthermore,the ratio of Qb/Qtot of most design methods is less than50%.

(c) At the Hound Point site, the influence on the calculatedpile capacities could be high: On one hand, thevariation between the different values of qc, avg is notthat high; on the other hand, the ratio of Qb/Qtot formost of the methods is 75%. Furthermore, there is alarge scatter in the total predicted capacities, i.e. theFugro-05, NGI-05 and UWA-05 methods would givebetter results when utilizing qc, avg, Dutch. Nevertheless,the ICP-05 and HKU-12 methods would predict worseresults by utilizing a lower value of qc, avg .

(d) At the Tokyo Bay site, the Fugro-05 method wouldbenefit fromusing a lower value of qc, avg . However, theother CPT based designmethods show good agreementby the utilization of qc, avg,1.5D.

6. Conclusion and outlook

This paper presents a study of the state of the artmethods used to determine the axial bearing capacity ofopen steel pipe piles. In addition to existing publicationscomparing axial pile design methods against field tests,e.g. Schneider et al.29 and Gavin et al.,30 this studycontains two new design methods: the HKU-12 methodand the EA-Pfähle approach. In the sensitivity analysis, thedependency of the predicted bearing capacity on the coneresistance qc and the diameter of the pile was shown. TheFugro-05, HKU-12 and UWA-05 methods demonstrated ahigher dependency on the pile diameter than the otherdesign methods. For these three methods, the predictedbearing capacity increases in a nonlinear manner withincreasing pile diameter.

The comparison with different field tests shows thequality of the design methods under field conditions. TheCPT based methods showed that they can give a realisticprediction of the real bearing capacity. The new HKU-12method especially, predicts results that are in goodagreement to the measured pile resistance and a low COV.Moreover, the UWA-05 and HKU-12methods showed verygood overall agreement. The German approach, EA-Pfähle,showed a less suitable overall agreement with the worstmean and COV values.

Furthermore, the dependency of qc, avg on the total axialbearing capacity could be demonstrated. It was observedthat the determined value of qc, avg can vary a lot fordifferent qc profiles. Also, it was shown that fairly goodagreement can be achieved when utilizing qc, avg,1.5D.

This paper has shown, under the study conditionsassumed, that the difference range of the predicted resultsis large. While one design method can predict just 25%,another method predicts almost 150% of the real capacity.The range of the predicted capacities increases with thepile diameter. This indicates that the design methodswhich we presented work less efficiently, if they areused outside their calibrated environment. To choose anappropriate design method for a specific project, careshould be taken to ensure that the method has alreadybeen validated against projects with comparable siteconditions. However, especially in the case of monopiles,there is a lack of published high quality field tests and therange of predicted capacities may be high. To overcomethe lack of documented field tests dealing with thebearing capacity of monopiles, numerical simulations incombination with the experience from design practice ofcomparable projects can be utilized.

To improve the current design methods for open steelpipe piles with a focus on larger diameters, the lack ofpublished field tests for large diameter piles has to beovercome. This can be done by partially replacing the fieldtests with calibrated and validated numerical simulations.In future research, a focus will be put on the numericalsimulation of the installation process of open steel pipe

80 J. Labenski, C. Moormann / Geomechanics for Energy and the Environment 6 (2016) 70–80

piles with regard to different installation methods. Withthese simulations, comprehensive model- and field testsand the experience from design practice of comparableprojects, the axial and lateral bearing behaviour can beinvestigated and monopiles can be designed in a moreeconomical way.

References

[1] Lüking J. Tragverhalten von offenen Verdrängungspfählen unterBerücksichtigung der Pfropfenbildung in nichtbindigen Böden. Heft23, Universität Kassel; 2010.

[2] Jardine R, Chow F, Overy R, Standing J. ICP designmethods for drivenpiles in sands and clays. In: ICP Design Methods for Driven Piles inSands and Clays. London: Thomas Telford Publishing; 2005:1–14[Chapter 1,2].

[3] Brucy F, Meunier J, Nauroy J-F. Behaviour of a pile plug in sandysoils during and after driving. In: Offshore Technology Conference.1991:145–154.

[4] Paik K, Salgado R, Lee J, Kim B. Behavior of open- and closed-endedpiles driven into sands. J Geotech Geoenviron Eng . 2003;129(4):296–306.

[5] RandolphMF, Leong EC, Houlsby G. One-dimensional analysis of soilplugs in pipe piles. Géotechnique. 1991;41(4):587–598.

[6] Randolph MF, Dolwin J, Beck R. Design of driven piles in sand.Geotechnique. 1994;44(3):427–448.

[7] American Petroleum Institute. Recommended practice for planning,designing and constructing fixed offshore platforms—working stressdesign. 2007.

[8] DIN EN ISO 19902:2014-1. Petroleum and natural gas industries—fixed steel offshore structures. 2014.

[9] Kolk HJ, Baaijens AE, Senders M. Design criteria for pipe pilesin silica sands. In: Frontiers in Offshore Geotechnics, ISFOG 2005—Proceedings of the 1st International Symposium on Frontiers in OffshoreGeotechnics. 2005:711–716.

[10] Clausen CJF, Aas PM, Karlsrud K. Bearing capacity of driven piles insand, the NGI approach. In: Frontiers in Offshore Geotechnics, ISFOG2005—Proceedings of the 1st International Symposium on Frontiers inOffshore Geotechnics. 2005:677–682.

[11] Lehane BM, Schneider JA, Xu X. The UWA-05 method for predic-tion of axial capacity of driven piles in sand. In: Frontiers in OffshoreGeotechnics, ISFOG 2005—Proceedings of the 1st International Sympo-sium on Frontiers in Offshore Geotechnics. 2005:683–690.

[12] Yu F, Yang J. Base capacity of open-ended steel pipe piles in sand.J Geotech Geoenviron Eng . 2012;138:1116–1128.

[13] Yu F, Yang J. Improved evaluation of interface friction on steel pipepile in sand. J Perform Constr Facil. 2012;26(April):170–179.

[14] Members of the AK 2.1 piling committee of the German geotechnicalsociety V. Recommendations on Piling (EA Pfähle). Wiley; 2013.

[15] Lammertz P. Ermittlung der Tragfähigkeit vibrierter Stahlrohrpfählein nichtbindigem Boden. Heft 35, Universität Duisburg–Essen, 2008.

[16] Jardine RJ, Thomsen NV, Mygind M, Liingaard MA, Thilsted CL. Axialcapacity design practice for North European wind-turbine projects.In: Meyer V, ed. Frontiers in Offshore Geotechnics III. Oslo: CRC Press;2015:1428.

[17] Xu X, Lehane BM, Schneider JA. Evaluation of end-bearing capacityof open-ended piles driven in sand from CPT data. In: Frontiers inOffshore Geotechnics, ISFOG 2005—Proceedings of the 1st InternationalSymposium on Frontiers in Offshore Geotechnics. 2005:725–731.

[18] Lehane BM, Schneider JA, Xu X. The design of displacement piles insiliceous sands using the CPT. Aust Geomech J . 2008;43(2):21–40.

[19] De Nicola A, Randolph MF. The plugging behaviour of driven andjacked piles in sand. Géotechnique. 1997;47(4):841–856.

[20] Lehane BM, Gavin KG. Base resistance of jacked pipe piles in sand.J Geotech Geoenviron Eng . 2001;127(6):473–480.

[21] Lee J, Salgado R, Paik K. Estimation of load capacity of pipe piles insand based on cone penetration test results. J Geotech GeoenvironEng . 2003;129(5):391–403.

[22] Lüking J, Becker P. Harmonisierung der Berechnungsverfahren deraxialen Tragfähigkeit für offene Profile nach EA-Pfähle und EAU.Bautechnik. 2015;92:161–176.

[23] EAU. Recommendations of the Committee for Waterfront StructuresHarbours and Waterways. 9th ed. Wiley; 2015.

[24] Moormann C, Kempfert H-G. Jahresbericht 2014 des Arbeit-skreises ‘‘Pfähle’’ der DeutschenGesellschaft für Geotechnik (DGGT).Bautechnik. 2014;91(12):922–932.

[25] Ko J, Sangseom J. Analysis of plugging effect for open-ended pilesin sandy soil. In: 11th International Conference on Piling and DeepFoundations. Stockholm. 2014.

[26] Schmertmann JH. Guidelines for cone penetration test - performanceand design. Tech. rep.. Washington, DC: US Department of Trans-portation; 1978.

[27] Yang ZX, Jardine RJ, Guo WB, Chow F. A new and openly accessibledatabase of tests on piles driven in sands. Géotech Lett . 2015;5(1):12–20.

[28] Gavin K, Jardine R, Karlsrud K, Lehane B. The effects of pile aging onthe shaft capacity of offshore piles in sand. In: Meyer V, ed. Frontiersin Offshore Geotechnics III. Oslo: CRC Press; 2015:1428.

[29] Schneider JA, Xu X, Lehane BM. Database assessment of CPT-baseddesign methods for axial capacity of driven piles in siliceous sands.J Geotech Geoenviron Eng . 2008;134(September):1227–1244.

[30] Gavin K, Igoe D, Doherty P. Piles for offshore wind turbines: a state-of-the-art review. Proc ICE—Geotech Eng . 2011;164:245–256.