11th International Stress Wave Conference Rotterdam, The Netherlands September 20-23, 2022

An assessment of pile driveability analyses for monopile foundations

Ioannis Kourelis i), Stavroula Kontoe ii), Roisin Buckley iii) and Andrew Galbraith iv)

i) Geotechnical Engineer, ADT Omega, 25 Avlidos Str., Athens, Greece.

ii) Reader in Soil Dynamics, Department of Civil and Environmental Engineering, Imperial College, Exhibition Rd, London, UK.

iii) Lecturer in Geotechnical Engineering, School of Engineering, University of Glasgow, Oakfield Avenue, Glasgow, UK.

iv) Director, Geowynd, 6 Grosvenor Street, 9-13, London, UK.

ABSTRACT

Several methodologies to predict the static soil resistance to driving (SRD) available in the literature have found wide
use in the offshore industry over the last decades. These range from simple methods that require few soil strength
parameters to more advanced semi-empirical methods that correlate the driving resistance to cone penetration test
measurements. These methods were primarily developed based on driving records for piles less than 2.5m in diameter
i.e. much smaller than the monopiles currently used in the offshore wind industry today. The aim of this study is to
evaluate the accuracy of some of the most widely used SRD prediction methods when employed for driveability
analysis of large diameter monopile foundations, by comparing the predicted SRD profiles with the driving records of
6.5m diameter monopiles installed in the Danish region of the North Sea.

Keywords: driveability analyses, SRD prediction methods, monopiles, pile run

1 INTRODUCTION takes place.

In this study, four SRD prediction methods [Stevens,
Pile driveability is a major component of the pile
et al. (1982), Toolan and Fox (1977), Alm and Hamre,
design process for the offshore industry, as the selection
(2001) & DNV skirt foundation method (1992)] are
of a suitable driving system based on preliminary
assessed for their accuracy in predicting self-weight pile
predictions can significantly reduce the pile installation
penetrations. Three of the SRD approaches are then
cost. This is becoming increasingly important as the
employed in driveability analyses of monopiles. Both
typical pile diameter is rising to accommodate the heavy
pile runs and continuous driving cases are examined, and
loads from the increasingly larger wind turbine
the predicted blow-counts are compared to the measured
generators.
response, aiming to give guidance for the use of the
Pile driveability analyses evaluate the ability of the
considered SRD methods for large diameter piles and to
pile to be installed with an acceptable risk of early
suggest appropriate modifications in their formulations.
refusal, to reach the designed penetration depth. In cases
of early pile refusal, a larger hammer needs to be
2 THE SITE
sourced, causing delays and additional costs. Moreover,
in cases of low driving resistances, unexpected pile runs The windfarm examined in this study is situated in
could damage the crane, the steel wires and even cause the Danish sector of the North Sea, comprising 49 wind
the loss of the pile and hammer into the sea if no turbines, all founded on 6.5m diameter open-ended steel
precaution measures are in place. Attention is also paid monopiles, with foundation depths between 25m and
to the number of hammer blows, which must remain 32.3m. The facility sits in a water depth between 11m
within a reasonable range, to avoid overstressing the and 19m.
steel and causing fatigue damage to the pile. Accurate Foundation conditions at the site comprise post-
assessment of the soil resistance during driving (SRD) is glacial sediments of shallow marine-coastal origin,
key to such predictions, ensuring that a safe installation which can be simplified from a geotechnical point of

1
 11th International Stress Wave Conference Rotterdam, The Netherlands September 20-23, 2022

view to two main geotechnical units. The first unit with a skirt foundation method proposed in the Danish
represents the recent shallow marine depositional code of practice (1992), briefly presented in Table 1, are
environment and is formed by very dense sands of employed in this study for the driveability analyses and
varying thickness, while the second unit reflects the tidal the prediction of self-weight pile penetrations.
lagoon dominated depositional environment and
comprises predominantly sandy silts and clays with local 4 PILE DRIVEABILITY ANALYSIS
sand lenses.
During pile driving, each hammer blow generates an
elastic wave which propagates away from the point of
3 SRD PREDICTION METHODS
impact at the pile head, decreasing in amplitude as it
During pile driving there are two components travels downwards. Pile driving can be idealised as a
constituting the total driving resistance: the dynamic wave propagating within a 1-D unconstrained elastic
resistance, which is an outcome of inertial and viscous rod. For this study, the software ALLWAVE.PDP
rate effects and the static resistance (Randolph, 2000). (Allnamics), which adopts the Smith model (1960), was
The latter is the sum of the shaft friction and the end used to solve the 1-D analysis problem. The pile is
bearing resistances applied to the pile wall surface and discretised in segments, while the interface condition
the tip area respectively. Various methods of evaluating and far-field response is represented by a linear
the SRD exist in the literature and have found wide use weightless spring and a plastic slider with a viscous
in the industry. Steven’s et al. (1982), Toolan and Fox dashpot. The interaction between the pile and the soil is
(1977) and Alm and Hamre (2001) are some examples controlled by the linear springs, until the threshold of the
of SRD prediction methods developed through back plastic slider is reached. The viscous enhancement of the
analysis, using records from pile installations in various soil caused by rapid strains along the shear zone, is
soil conditions. These three standard approaches, along represented by the viscous dashpots, which add to the

Table 1. Summary of main principles of SRD prediction methods.

Method formulations
Method Origin of the method Method basic principles Type of foundation soil
Cohesionless soils Cohessive soils

No CPT based method.

Back analyses of pile shaft friction:
Shaft and tip resistance is shaft friction: fs=Ks∙p'0∙tanδ
Stevens et al installation records in the fs=Fp∙α∙su
determined on the basis of end bearing:
(1982) Arabic Gulf end bearing:
the API code (1981) static q=p'o∙Nq
(pile diameter 0.91m-1.06m) q=9∙su
formulations.

Back analyses of pile shaft friction: shaft friction:

Toolan and Fox installation records of open- fs=1/300*qt fs=su res
CPT based method
(1977) ended steel piles in the North end bearing: end bearing:
Sea area q=qt (weighted average) q=qt (weighted average)

Initial shaft friction: Initial shaft friction:

fsi=(Ks∙σ'v0∙tanδ)∙0.5 fsi=fs
Back analyses of pile Residual shaft friction: Residual shaft friction:
CPT based method fsres=0.2∙fsi
Alm and Hamre installation records of open fsres=0.004∙qt∙(1-0.0025∙qt/p'o)
Friction Degradation Tip resistance:
(2001) ended piles in the North Sea Tip resistance:
Mechanism
(pile diameter 1.8m-2.7m) qtip=0.15∙qt∙(qt/p'o)0.2 qtip=0.6∙qt
Shaft friction: Shaft friction:
fres+(fsi-fres)∙ek∙(d-p) fres+(fsi-fres)∙ek∙(d-p)

shaft friction:
DNV skirt Guidelines described in the
fs=kf∙qt
foundation method DNV classification notes for CPT based method
end bearing:
(1992) skirt design.
fs=kp∙qt
Legend: Ks:coefficient of lateral earth pressures, p' o:effective overburden stress, δ:interface friction angle, F p: dimensionless OCR factor
proposed by Semple & Gemeinhardt (1981), α: shaft friction factor, s u:undrained strength, qt:cone tip resistance, k: friction degradation factor,
d: depth of layer, p: depth of the pile tip.

2
 11th International Stress Wave Conference Rotterdam, The Netherlands September 20-23, 2022

model a resistance which is proportional to the loading recorded and was compared to pile weight force along
rate. The total soil resistance applied to each element is the zones where zero blow-count was indentified. The
the sum of the static resistance, modeled by the springs,
and the dynamic resistance modeled by the dashpots.
In addition to the information regarding the
characteristics of the hammer, the pile geometry, the
CPT profile for each location and the SRD prediction
methodology, the software requires a set of parameters
(quake, damping, alpha factor, yield factor) which
accompany Smith’s solution. Quake values define the
elastic threshold of the spring. When the soil-pile relative
displacement exceeds the quake value plastic behaviour
is triggered. The damping parameter relates the shaft
friction applied to the pile during driving with the
velocity of the pile particles. The relationship can be
linear or exponential, which is controlled by the alpha
factor. The yield factor defines the relationship between
the plastic resistance force of a spring in compression
and tension. Finally, the hammer delivered energy per
blow for 0.25m penetration increments is required so that
the analyses results in terms of blow-count can be
compared to the measured pile driving response.
Furthermore, in the case of the Alm and Hamre SRD
methodology, a modified friction degradation
mechanism is also examined. The original model
suggests that the residual shaft friction in sands can drop
as low as 20% of the initial value. However, this
degradation concept yielded SRD results which typically
lie below the measured response. Therefore, in addition Fig. 1. SRD predictions – pile run case 1.
to the original method, a case was examined where the
residual shaft friction was raised to 50% of the initial
value.

5 ANALYSIS RESULTS AND DISCUSSION

5.1 Pile run predictions
The SRD prediction methodologies were employed
to establish the soil resistance profiles at the wind
turbine generators (WTG) locations where pile runs
occured. Pile run initiates when the weight force of the
pile exceeds the soil resistance. To assess the accuracy
of each method in predicting pile runs, the total predicted
soil resistance force is compared to the pile and hammer
weight force (Fig. 1, Fig. 2 & Fig. 3).
The Alm and Hamre and DNV methodologies yield
similar SRD trends, which typically fall below or very
close to the weight of the pile in zones where the pile run
is well defined and continuous (Fig. 1 & Fig. 2).
Furthermore, in locations where pile running is not
continuous and clear, but is disrupted by very low blow-
counts (Fig. 3), both methods seem to yield very accurate
predictions, since the calculated SRD force lies between
the pile force and the combined pile and hammer force,
reflecting precicely the pile driving sequence.
In order to evaluate the methods examined above in
a larger scale, the calculation of the SRD force was
undertaken in 17 WTG locations where pile runs were Fig. 2. SRD predictions – pile run case 2.

3
 11th International Stress Wave Conference Rotterdam, The Netherlands September 20-23, 2022

Fig. 4 provides an overview of how the predicted SRD

force compares with the pile weight force according to
each method. The DNV method produces overall the
most accurate results, since the predicted SRD force is
on average 24.7% larger than the pile weight force along
the identified pile run zones. This means that while the
method provides accurate estimates of the pile run
locations, there are still zones along which the SRD was
overpredicted. Toolan and Fox yields the least accurate
results overall, as it overpredicts the SRD force by
167.8% on average along the pile run zones. The SRD
computed with the Alm and Hamre method exceeds the
pile weight force by 51.5% on average. Even though the
method considers friction degradation, in some cases the
SRD is significantly overpredicted. Furthermore, from
Fig. 4, it might seem that Stevens method predicts the
SRD with adequate accuracy in some cases. This is
because the method yields very low SRD profiles at
shallow depths that fall below the pile weight, however,
this method is not suitable for predicting the extent of the
pile runs when these reach greater depths.
5.2 Driveability Analysis
The driveability predictions computed with the
Toolan and Fox, Steven’s et al. and Alm and Hamre
methodologies, as well as the modified Alm and Hamre
Fig. 3. SRD predictions – pile run case 3. model, are compared to the measured response at three
wind turbine locations in Fig. 5, Fig. 6 & Fig. 7.
predicted SRD force should drop below the pile self In general, all methods consistently overpredicted the
weight force along the recorded pile-run zones. driving resistance over the first approximately 7m of
Conversly, should the predicted SRD force be larger than penetration. This is attributed to the input data to the
the pile weight force along a recorded pile run zone, the analyses, and more specifically the CPT measurements.
excess SRD force is measured (difference between the The measured cone tip resistance swiftly rises over the
SRD force and pile and hammer weight force) and is first metres of penetration, which leads to a high
compared to the pile weight force. It should be computed driving resistance, which is not observed in
emphasised that the weight of the hammer was not the pile driving records. Puech and Forey (2002) have
included in the computations described above, as in described an upwards concave parabolic increase in cone
modern offshore pile installation practice the hammer is tip resistance during the first metres of penetration in
typically detached from the pile when pile run initiates. medium dense to very dense sands, similar to the ground

350
(Predicted SRD Force - Pile Weight

167,8
Force)/ Pile Weight Force [%]

300
250
51,5
200 46,4
150 24,7
100
50
0
Alm and Hamre Toolan and Fox Stevens (low estimate) DNV (best estimate)
Pile run case 1 Pile run case 2 Pile run case 3 Pile run case 4
Pile run case 5 Pile run case 6 Pile run case 7 Pile run case 8
Pile run case 9 Pile run case 10 Pile run case 11 Pile run case 12
Pile run case 13 Pile run case 14 Pile run case 15 Pile run case 16
Pile run case 17 Average value
Fig. 4. Comparison of SRD prediction methods with respect to the average difference between the calculated SRD force and the pile
weight force along recorded pile run zones.

4
 11th International Stress Wave Conference Rotterdam, The Netherlands September 20-23, 2022

conditions encountered at the examined windfarm site. underlying softer silt zone, resulting in higher results.
They suggest that this rapid rise can be attributed to
upward movement of the soil which surrounds the cone Blowcount [blows/0.25m]
rods. On the other hand, the Toolan and Fox, Stevens et 0 20 40 60 80 100
al. and Alm and Hamre (without modifications) methods 0
typically underpredicted the driving resistance at greater
depths.
The Alm and Hamre model produces the lowest 5
blow-counts, underpredicting the measured response by
approximately a factor of 2. Furthermore, the deviation sand
between the Alm and Hamre output and the driving 10
records became more significant with increasing depth.

Depth [m below seabed]

This suggests that the effect of the friction degradation
concept, embedded in the Alm and Hamre method, 15 silt
whereby the friction resistance applied by the overlying
layers degrades as the pile penetrates deeper into the
20 sand
ground, might be overestimated. On the other hand, the
Alm and Hamre method seems to yield better estimates
at locations where low blow counts were measured. Fig. clay
7 shows that this method provided predictions similar to 25
the recordings for the weaker layer of silt which sits sand
between 10m and 20m below seabed. Consequently, the
method can be particularly useful for identifying weaker 30
zones where pile run might occur. Measured
The Toolan and Fox and Stevens methods usually Toolan and Fox
35
produced similar blow-counts which typically lie higher Stevens
than the Alm and Hamre predictions, as these methods Alm and Hamre
do not consider a friction degradation mechanism. In Alm and Hamre modified
40
many cases, the methods gave relatively accurate
estimations of the pile driving response, even Fig. 5. Driveability predictions – indicative location 1.
outperforming the Alm and Hamre approach, which is
considered a more advanced method. However, both Blowcount [blows/0.25m]
Toolan and Fox and Stevens et al. approach predicted 0 20 40 60 80 100
blow-counts which did not reflect the low driving 0
resistance along weaker silt and clay zones. As a result,
low driveability zones, susceptible to pile-runs, could be
overlooked, if pile driveability analyses do not consider 5
a wider range of SRD prediction methodologies.
The modified Alm and Hamre method yields higher 10
results compared to the original method, as the lower
Depth [m below seabed]

limit of the residual resistance has been raised to 50%

from 20%, which was the value adopted by the original 15 sand
model. The measured response in most cases was
accurately reflected by the modified Alm & Hamre,
especially for sites characterised by predominantly Post- 20
Glacial sand deposits (Fig. 5 & Fig. 6). However, the
method seems to overpredict the driving resistance along 25
low driveability zones (Fig. 7). Consequently, while the
method gives accurate results in “normal” driving
conditions, it is not suitable for predicting pile runs. It 30
should be emphasized however that the accuracy of the Measured
predicted blow-counts along the soft zones could be Toolan and Fox
affected by the over-predicted resistance in the overlying 35 Stevens
layers, especially when a reduced friction degradation Alm and Hamre
effect is adopted. As it can be seen in Fig. 7, the higher Alm and Hamre modified
40
predicted driving resistance in the upper sand layer,
contributes to the predicted blow-count within the Fig. 6. Driveability predictions – indicative location 2.

5
 11th International Stress Wave Conference Rotterdam, The Netherlands September 20-23, 2022

Blowcount [blows/0.25m] attributed to very high CPT records. The modified Alm
0 20 40 60 80 100 and Hamre model generally provided predictions closer
0 to the recordings, with the average deviation between the
recorded and predicted values lying in 7.9 blows per
0.25m of penetration. On the other hand, the least
5 sand accurate results were produced by the original Alm and
Hamre method (average difference 12.7 blows/0.25m),
while the Toolan and Fox and Stevens predictions were
10 both off by 10.1 blows per 0.25m on average.
Depth [m below seabed]

6 CONCLUSIONS
15 silt
Some of the most widely used SRD methods were
assessed at a North Sea site which comprises Post-
20 Glacial sands interlayered with silts and clays. The
methods were first assessed regarding their accuracy in
predicting pile-runs. Subsequently the original methods,
25 sand as well as a modified version of one, were employed in
driveability analyses and a comparison with the recorded
response was undertaken.
30 The results indicated that the DNV approach could
Measured predict the pile run zones quite accurately in most cases,
Toolan and Fox while the Alm and Hamre method produced similar, yet
35 Stevens slightly overestimated results. On the other hand, the
Alm and Hamre Toolan and Fox method was shown to significantly
Alm and Hamre modified overpredict the SRD and hence to be unsuitable in
40
identifying potential pile run zones. Stevens method was
Fig. 7. Driveability predictions – indicative location 3. not suitable at accurately predicting pile run zones either,
due to the method’s formulas being dependent on
In order to compare the methods examined above in confining pressures and not in-situ measurements.
a larger scale, driveability analyses were carried out in The results from the driveability analyses showed
10 WTG locations in total. Fig. 8 provides an overview however that when the Alm and Hamre method is
of how all four methods performed in comparison to the employed within a driveability software, the blow-count
measured blow-counts. The trends show the average was underpredicted, particularly at greater depths. It is
absolute difference between the recorded and the tentatively concluded that the reasoning behind the
predicted blows for each penetration increment (0.25m), underestimations is the method’s friction degradation
providing an indicator of how accurately each method concept not reflecting the pile dimensions. Therefore, a
estimates the recorded blows. It should be noted that the modified version of the original Alm and Hamre method,
upper 7m of penetration have been neglected from the with a reduced friction degradation effect was examined
calculation, since all methods yielded unrealistic results, and found to yield more accurate results in most cases

30
AVERAGE DIFFERENCE
Average absolute difference between
recorded & predicted blows/0.25m

25 Alm and Hamre: 12.7

Toolan and Fox: 10.1
Stevens: 10.1
20 Alm and Hamre modified: 7.9
12,7
15 10,1 10,1
7,9
10

0
Alm and Hamre Toolan and Fox Stevens Alm and Hamre modified
Pile 1 Pile 2 Pile 3 Pile 4 Pile 5 Pile 6 Pile 7 Pile 8 Pile 9 Pile 10

Fig. 8. Comparison of driving resistance prediction methods with respect to the average difference between measured and recorded
blow-count.

6
 11th International Stress Wave Conference Rotterdam, The Netherlands September 20-23, 2022

analysed. 3) DNV, 1992. Det Norske Veritas, Foundations, Classification

To conclude, the results of this study highlight the Notes 30.4.
need of employing multiple methods when assessing 4) Puech, A. & Foray, P., May 2002. Refined Model for
driveability conditions, as some of the existing methods Interpreting Shallow Penetration CPTs in Sands. Houston,
might yield unrealistic results for monopiles Texas, Offshore Technology Conference.
Modifications of existing methods or new methods
based on driving records of monopiles are needed which 5) Randolph, M., 2000. Pile-soil interaction for dynamic and
should probably include a friction degradation term static loading. In Proceedings of the 6th International
which is function of the pile diameter to wall thickness Conference on Application of Stress Wave Theory to Piles.
pp. 3-11.
ratio.
6) Semple, R. & Gemeinhardt, J., 1981. Stress History
ACKNOWLEDGEMENTS Approach to Analysis of Soil Resistance to Pile Driving.
Houston, Texas: Offshore Technology Conference.
The authors wish to express their gratitude to Vattenfall
for supplying the pile installation records. 7) Smith, E., 1960. Pile driving analysis by the wave equation.
Journal of the Soil Mechanics & Foundations Division,
REFERENCES ASCE, vol.86 (SM4), pp. 35-61.

1) Alm, T. & Hamre, L., 2001. Soil model for driveability 8) Stevens, R., Wiltsie, E. & Turton, T., 1982. Evaluating
predictions based on CPT measurements. In Proc. 15th Int. drivability for hard clay, very dense sand, and rock. In
Conf. on Soil Mech & Geotech Engineering (pp. 1297-1302). Offshore Technology Conference. Offshore Technology
Conference.
2) API, 2000. Recommended Practice for Planning, Designing
and Constructing Fixed Offshore Platforms - Working Stress 9) Toolan, F. & Fox, D., 1977. Geotechnical Planning of Piled
Design. 21st ed. Washington, DC: American Petroleum Foundations. In: Proceedings of the Institution of Civil
Institute. Engineers. Institution of Civil Engineers, pp. 221-244.