You are on page 1of 11

Computers and Geotechnics 36 (2009) 725–735

Contents lists available at ScienceDirect

Computers and Geotechnics


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

Behavior of monopile foundations under cyclic lateral load


Martin Achmus *, Yu-Shu Kuo, Khalid Abdel-Rahman
Institute of Soil Mechanics, Foundation Engineering and Waterpower Engineering, Leibniz University of Hannover, Appelstr. 9A, Hannover 30167, Germany

a r t i c l e i n f o a b s t r a c t

Article history: This paper describes the development and application of design charts for monopile foundations of off-
Received 14 May 2008 shore wind turbines in sandy soil under long-term cyclic lateral load. It outlines a numerical model,
Received in revised form 5 November 2008 working with a numerical concept, which makes the calculation of accumulated displacements based
Accepted 7 December 2008
on cyclic triaxial test results possible, and it describes important factors affecting the deformation
Available online 31 January 2009
response of a monopile to cyclic lateral loads. The effects of pile length, diameter and loading state on
the accumulation rate of lateral deformation are presented and design charts are given, in which a nor-
Keywords:
malized ultimate lateral resistance of a pile is used. For monopiles with very large diameter, the suitabil-
Monopile
Offshore wind turbine
ity of the ‘‘zero-toe-kick” and ‘‘vertical tangent” design critera for determining the required embedded
Cyclic load length is discussed.
Lateral deformation Ó 2008 Elsevier Ltd. All rights reserved.
Sand

1. Introduction Although many methods have been proposed for estimating


lateral deformation of a pile in sand under cyclic lateral load, no
The monopile is one foundation option for offshore wind tur- generally accepted design procedure yet exists. The standard de-
bines planned in the German parts of the North Sea and the Baltic sign procedure for a laterally loaded offshore pile is the p–y curve
Sea. Most of the already existing offshore wind energy converters method, which is described in the guidelines of the American
are founded on monopiles. However, these turbines have rated Petroleum Institute (API) [1]. The p–y curves formulated for cyclic
power values of less than 3 MW and are located in relatively shal- loading conditions are based on field tests with fewer than 200 cy-
low water (in most cases less than about 10 m depth). For wind cles. Moreover, a major disadvantage is that the resulting deforma-
turbines of 5 MW rated power in water depths of 15–30 m a mono- tions, determined for instance with the LPILE computer program
pile foundation would require a diameter of 5–7.5 m to ensure ade- [2], are independent of the actual number of load cycles. This
quate bearing capacity under lateral load in this harsh means that a prognosis which takes site-specific soil and long-
environment. Monopiles of about 5 m diameter have already been term cyclic loading due to met-ocean conditions into account is
installed in the North Sea by driving, and it is believed that diam- not possible.
eters of up to 7.5 m can be executed if suitable driving equipment Using sophisticated material laws, in principle the behavior of a
is constructed. pile under cyclic loads can be simulated cycle by cycle. However,
Besides the design for the maximum static load, fatigue design the rate of displacement accumulation is usually very small, and
is a very important aspect for offshore structures. The effect of cyc- thus the accumulation of numerical errors could become large
lic loading of the soil has to be considered, since the number of compared with the displacements to be determined. Thus, such
loads due to wind and wave could exceed 108 over the lifetime numerical models with implicit calculation concepts are restricted
of the structure. Cyclic loading leads to an accumulation of the pile to the estimation of pile deflections with cycle numbers smaller
head deformation and rotation, which has to be limited to avoid than 50 [3–7].
exceeding the serviceability limit of a wind turbine. In recent pro- With explicit methods, semi-empirical approaches are used to
jects, usually a maximum permanent rotation of a monopile at describe the lateral deformation response of a pile with respect
mudline of 0.5° was required. However, the estimation of accumu- to the number of load cycles. The development of cyclic deforma-
lated displacements is one of the most difficult tasks for engineer- tions dependent on the number of cycles is directly described by
ing design. empirical equations [8–13], or the change of bedding resistance
in subgrade reaction methods with the number of cycles is de-
scribed [14,15]. In the methods of Long and Vanneste [15] and
* Corresponding author. Tel.: +49 511 7624155; fax: +49 511 7625105. Lin and Liao [13], the effects of soil density, pile installation meth-
E-mail address: achmus@igbe.uni-hannover.de (M. Achmus). od and the cyclic loading state are taken into account. However,

0266-352X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compgeo.2008.12.003
726 M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735

Table 1 ever, the disturbance is limited to a zone of relatively small thick-


Material parameters used for dense sand and medium dense sand. ness around the pile, whereas the soil in a much larger zone is
Parameter Dense sand Medium dense sand object to stress increase due to horizontal loading.
Unit buoyant weight c0
11 kN/m 3
11 kN/m3 In application, the finite element program ABAQUS [17] is used.
s
Oedometric stiffness parameter j 600 400 The model of a monopile foundation under monotonic lateral load
Oedometric stiffness parameter k 0.55 0.60 presented by Abdel-Rahman and Achmus [18] is used in this study
Poisson’s ratio m 0.25 0.25 as a basis for cyclic analysis. For the simulation of the soil behavior,
Internal friction angle u0 37.5° 35°
Dilation angle w 7.5° 5°
an elasto-plastic material with Mohr–Coulomb failure criterion is
Cohesion C 0.1 kN/m2 0.1 kN/m2 used. To account for the stress-dependence of the oedometric stiff-
ness modulus Es an increase with depth is assumed according to
the equation
these methods are based on small-scale laboratory pile test results  k
rm
or field tests with a limited number of cycles, and suitability for a Es ¼ j  rat  ð1Þ
rat
monopile with a large diameter is not clear.
The German guideline for soil exploration and testing with re- where rat is the atmospheric pressure, j and k are material con-
gard to offshore wind turbine foundations [16] suggests cyclic lab- stants and rm is the mean principal stress. Contact interaction be-
oratory tests to assess the foundation response under long-term tween the soil and pile elements is modelled by contact elements
cyclic loading. However, there is no approved method for taking with the contact friction angle d taken as two thirds of the soil fric-
into account the results of such tests in the design of piles for accu- tion angle u’. The soil parameters for monotonic loading used in the
mulated displacements. model are given in Table 1.
The purpose of this study is to apply the results of drained cyclic In Fig. 1 exemplary results of calculated load-deformation
triaxial tests on cohesionless soil in a numerical model to estimate curves of a monopile with embedded length L = 30 m, pile diameter
the progressive deformation of a monopile embedded in a sandy D = 7.5 m and pile wall thickness tp = 0.09 m in dense sand are gi-
seabed under long-term cyclic lateral load. In the following, the ven. With a variation of the moment arm h defined in Fig. 1, differ-
special numerical concept, the ‘‘degradation stiffness model”, is de- ent combinations of horizontal and moment loading can be
scribed, and, based on the results of a parametric study, design considered. In the numerical calculations of cyclic behavior, the
charts are presented which may be used as a tool for preliminary pipe section of the monopile is replaced by a solid section pile with
design. equivalent bending stiffness. Comparisons with the results of an
exact modelling of the pipe section confirmed the suitability of this
2. Degradation stiffness model simplification.
The degradation stiffness approach to account for cyclic loading
The degradation stiffness model presented in this study is a effects is elucidated in Fig. 2. In a cyclic triaxial test, an increase of
method based on a combination of a finite element simulation of the plastic axial strain can be observed. Assuming the elastic strain
the pile–soil interaction and an evaluation of drained cyclic triaxial to be negligible, the degradation rate of secant stiffness after first
tests. cycle Es1 and Nth cycle EsN can be presented by the plastic axial
In cyclic triaxial tests, the accumulation of plastic strains with strains after first cycle eacp;N¼1 and after Nth cycle eacp;N according
the number of cycles under different loading conditions can be ob- to the following equation:
served. This increase of plastic strain can be interpreted as a de- EsN eacp;N¼1
crease in soil secant stiffness. Assessing the stress conditions in ffi a ð2Þ
Es1 ecp;N
the distinct elements and introducing the stiffness degradation ob-
tained by comparison with the cyclic test results in the finite ele- The accumulation of plastic strains in a cyclic triaxial test can be
ment model yields the accumulated deformations of the pile–soil estimated from existing semi-empirical approaches of Huurman
system. This is the basic concept of this model. Of course the pile [19], Gotschol [20] and Werkmeister [21]. Here Huurman’s formula
installation by driving leads to a disturbance of the soil around [19] is considered, and with it the degradation of stiffness can be
the pile. This effect is omitted in the finite element model. How- described using two material parameters b1 and b2 as follows:

Fig. 1. Static load-deformation curves of a monopile.


M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735 727

Fig. 2. Degradation of secant modulus under cyclic loading in the pile–soil model (schematic).

EsN eacp;N¼1 b2 cidated in Fig. 3. To overcome this problem, a characteristic cyclic


¼ a ¼ Nb1 ðXÞ ð3Þ
Es1 ecp;N stress ratio Xc is defined here as

here N is the number of load cycles and X is the cyclic stress ratio X ð1Þ  X ð0Þ
defined by Huurman [19] for cohesionless material as follows: XC ¼ ð5Þ
1  X ð0Þ
r1;cyc here the index (1) means the cyclic stress ratio at loading phase and
X¼ ð4Þ
r1;sf the index (0) means at unloading phase (cf Fig. 4). At the initial (and
where r1,sf is the major principal stress at static failure state and unloading) phase, only the vertical load V due to the tower weight is
r1,cyc is the major principal stress for the cyclic stress state under considered, and the lateral load H is applied subsequently in the
consideration. The cyclic stress ratio is thus dependent on the con- loading phase. The characteristic cyclic stress ratio is derived from
fining pressure and on the cyclic stress level. the difference between the stress ratios in the loading and the
A problem to be dealt with is that the Eqs. (3) and (4) are valid unloading phase. Due to the denominator in Eq. (5) this value varies
for triaxial test conditions with isotropic initial stress conditions from 0 to 1. The accumulation of plastic strain and the degradation
and a constant confining pressure r3 during cyclic loading. In the of stiffness of the soil element can be obtained from Eq. (3) by
pile–soil system, the initial stress conditions (before application replacing X by Xc.
of the horizontal load) are anisotropic and the minor principal In the last step of the simulation (model C in Fig. 4), the defor-
stress in the elements as well as the direction of the principal stress mation response of the system is analyzed using the degradation
axes in general change with the application of the load. This is elu- stiffnesses EsN obtained from models A and B with respect to Eqs.

Fig. 3. Variation of principal stresses in the pile–soil system in initial phase and cyclic loading phase.
728 M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735

Fig. 4. Schematic sketch of the determination of degradation stiffness in the pile–soil system.

(3) and (5). Poisson’s ratio is assumed to remain constant in the


three discrete finite element models.
From cyclic triaxial test results documented in the literature,
typical regression parameters b1 and b2 were found for dense sand
to be b1 = 0.20, b2 = 5.76 and for medium dense sand b1 = 0.16,
b2 = 0.38 (Kuo [22]). These values were used in the parametric
study.
The monopiles with large diameter are driven in the field. The
installation of piles leads to a change of stress in the soil around
the pile as well as a change of the soil properties in the distorted
area close to the pile. When the monopile is subjected to a lateral
load, the area with stress variation in soil due to lateral loading is
much larger than the area with stress changing close to the pile
due to pile driving. Hence, the change of stress in soil and the soil
properties in the pile installation phase is omitted in the degrada-
tion stiffness model.
A cyclic triaxial test by Timmerman and Wu [23] and laboratory Fig. 6. Simulation of lateral pile deflection in a 1-g laboratory test [24] using the
pile tests by Achmus et al. [24] were simulated with the degrada- degradation stiffness model.
tion stiffness model. The results are shown in Figs. 5 and 6. The
permanent deformation responses of a soil element (Fig. 5) and
of a pile–soil system (Fig. 6) are well captured by the degradation ness in the upper half of the pile degrades significantly. The soil ele-
stiffness model. ments in the area subjected to active loading, i.e. in the opposite
In Fig. 7, the stiffness degradations resulting from the model for direction to which the pile is moving, also vary in stiffness as seen
two systems are compared. The stiffnesses of elements in the plane in Fig. 7. However, this stiffness variation in the opposite direction
of symmetry for one cycle (monotonic load) of a given load combi- of loading is negligible for a monopile subjected to one-way cyclic
nation and for 100 and 10,000 load cycles are given once for a pile loading and will not affect the accumulated displacement of the pile.
with a length of 40 m and once for a pile with a length of 20 m. The deflection lines of the two monopiles considered in Fig. 7
For the shorter monopile, the soil along the whole length of the pile are presented in Fig. 8. The monopile with an embedded length
is subject to stiffness decrease. For the longer pile, only the soil stiff- L = 20 m behaves like a short pile and exhibits a more or less rigid
response, whereas the monopile with L = 40 m behaves more flex-
ibly. In both cases the deflection increases with the number of cy-
cles and the zero deflection point gradually moves down with
cyclic loading. These results correspond with tests of Saglamer
[25]. Since the load for both piles is identical, the long pile has
much less deformation. However, not only the absolute displace-
ments, but also the relative rate of displacement accumulation is
much smaller for the longer pile. This is shown in Fig. 9, where
the relative increase of the pile displacement at seabed level is gi-
ven dependent on the number of load cycles in logarithmic scale.
The head displacement of the long pile is increased after 10,000 cy-
cles by a factor of less than 2, whereas the displacement of the
short pile is nearly 5 times larger than for monotonic loading.
The results are compared with the empirical approaches given in
[11] and [14]. Here the major advantage of the developed numer-
ical model becomes obvious. It seems logical that the rate of dis-
Fig. 5. Simulation of plastic strain response in a cyclic triaxial test [23] using the placement accumulation depends on the magnitude of loading,
degradation stiffness model. i.e. the ratio of cyclic load to, for instance, ultimate horizontal load.
M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735 729

Fig. 7. Variation of stiffness in two pile–soil systems dependent on the number of load cycles.

Fig. 8. Lateral deflection of piles according to the degradation stiffness model.

The numerical model is capable of accounting for that, whereas the degradation stiffness model is thus well able to simulate a mono-
results of existing empirical models are independent of the magni- pile under long-term cyclic lateral load. Although the model of
tude of loading and the system boundary conditions. course needs to be verified by further experimental evidence
The simulation results with the degradation stiffness model (large-scale tests, field measurements and observations), it seems
show a reasonable accumulation rate of lateral deformation at sea- to be appropriate for investigating monopile behavior under vari-
bed level. The effects of loading state and pile geometries are also ous boundary conditions. In the following, the model is applied
demonstrated reasonably by the degradation stiffness model. The to develop some reference charts for preliminary design.
730 M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735

For large head displacements, the obtained curve is approximately


a straight line, and the ultimate resistance is defined as the inverse
of the slope of this line.

3.2. Design charts for monotonic loading condition

Based on the simulation results for monotonic loading, the two


design charts given in Figs. 11 and 12 were developed.
In Fig. 11, the dimensionless lateral head displacement ys/L is
plotted versus a dimensionless load value. A similar presentation
of experimental test results has been given by Hettler [11], but
only for purely horizontal loading (h = 0). From the calculation re-
sults given in the left part of Fig. 11, the design chart given on the
right was developed. For the given regression lines the following
equations apply:
 
100ys
Medium dense sand : ln
L
" #
Hðh þ LÞ
¼ 1:168 ln þ 0:544 ð6Þ
Fig. 9. Comparison with existing methods on accumulated deformation of mono- c0s DL3
piles with D = 7.5 m, tp = 0.09 m, L = 20 and 40 m under H = 15 MN at h = 20 m.
  " #
100ys Hðh þ LÞ
Dense sand : ln ¼ 1:112 ln þ 0:151 ð7Þ
3. Parametric study and preliminary design charts L c0s DL3

3.1. Scope of analyses with the presentation in Fig. 11, different curves are valid for med-
ium dense and dense sand. If the normalized load H/Hu is consid-
Using the numerical model described above, a parametric study ered in the design chart, the deformation response can be
was carried out to identify and quantify the effects of geometric, estimated by only one curve, shown in Fig. 12.
subsoil and loading conditions on the behavior of monopiles. The For a specific case, the lateral head displacement ys of the mono-
following ranges of parameters were considered: pile under static loading can be determined with the chart in
Fig. 11. Subsequently, dependent on ys the normalized load level
 Pile geometry: embedded length L: 20–40 m, diameter D: 2.5– H/Hu – which is needed to assess the behavior under cyclic loading
7.5 m, pile wall thickness: tp: 0.09 m. – can be obtained from the chart in Fig. 12.
 Loading: one-way lateral load H: 5–30 MN, moment arm h: 0– It should be recognized that the results of the design charts are
40 m, number of load cycles N: 1–10,000. independent of the actual pile bending stiffness. In all calculations
 Soil types: dense sand and medium dense sand. a pile wall thickness of 0.09 m was assumed. This means that very
stiff piles, for small embedded lengths and diameters even almost
To simplify the practical design of monopiles, design charts rigid piles, were considered. Thus, the application of the charts is
based on the results of the parametric study are presented which restricted to such cases.
make a rough estimation of required pile dimensions for specific
loading conditions possible. 3.3. Monopile behavior under cyclic loading
In the design charts, a normalized load is used, which is defined
here as the ratio of the actual lateral load H and the ultimate lateral The increase of monopile displacements with the number of
resistance of the pile Hu. Many suggestions have been made for load cycles is illustrated in Figs. 8 and 9. To give a clear presenta-
determining the ultimate lateral resistance Hu of a single pile. tion of different effects on the pile performance under cyclic loads,
However, the results obtained by different methods vary widely. the increase of head displacement due to 100 load cycles ys100/ys1 is
To determine the ultimate lateral load consistent with the numer- chosen as a representative value. The effects of embedded pile
ical model, the Hu determined from the hyperbolic method sug- length, pile diameter, loading amplitude and load eccentricity on
gested by Manoliu et al. [26] is adopted as a reference value in this value are described below.
this study. With this method, ys/H is plotted versus ys (Fig. 10). Fig. 13 shows the effect of the embedded length for a monopile
with a diameter of 5 m and a load of H = 5 MN acting at h = 20 m
above sea bed level. A decrease of the embedded length leads to
an increase of the normalized load level. According to the numer-
ical results given in Fig. 13, this also leads to a larger accumulation
rate of lateral deformation ys100/ys1. A straight line can be assumed
to represent the dependence of accumulation rate on the normal-
ized load in the range of the parameters investigated. The dashed
line in Fig. 13 gives an estimation of the performance under higher
load levels. With approach to H/Hu = 1, of course an overlinear in-
crease of the accumulation rate is to be expected.
The example demonstrates that by decreasing the embedded
pile length L from 30 m to 20 m the accumulation rate increases
from about 1.8 to 2.5. Regarding the displacement under static
Fig. 10. Hyperbolic method to determine the ultimate lateral resistance of a single loading, Fig. 13 shows that the pile with the shorter length has a
pile. 2.3 times larger displacement. This means that after 100 load
M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735 731

Fig. 11. Dimensionless displacement of monopiles embedded in dense and medium dense sand under static lateral load.

Fig. 12. Lateral displacement of monopiles under static lateral load with regard to normalized load.

cycles the ratio of the absolute pile head displacements is even Fig. 1) of the load. The effect of a variation of the moment arm is
about 3.2. Thus, the pile performance under cyclic horizontal load- presented in Fig. 15, again in terms of the dependence of the accu-
ing is – as expected – very much dependent on the embedded pile mulation rate on the normalized load. The larger the moment arm
length. is, the smaller is the ultimate load and with that the larger is the
In Fig. 14 the effect of the pile diameter on the cyclic perfor- normalized load. The numerical results show that again the depen-
mance is shown for exemplary conditions (L = 20 m, H = 10 MN, dence can approximately be described by the straight line already
h = 4 m). The diameter of course also affects the ultimate load Hu used in Figs. 13 and 14.
and thus the normalized load H/Hu. The same line as in Fig. 13 is Finally, in Fig. 16 all the results for ys100/ys1 obtained for a
used to describe the dependence of accumulation rate and normal- monopile with a specific geometry (L = 30 m, D = 5 m) are pre-
ized load. sented together with the respective performance regarding the pile
Comparing piles with diameters of 5 m and 7.5 m, the pile with head rotation Us100/Us1. Considering the rotation, a similar perfor-
the larger diameter has only slightly better cyclic performance, since mance, i.e. an increase of accumulation rate with increasing nor-
the normalized load levels in these cases are not as different as in the malized load, is found. However, the accumulation rate is smaller
case of increasing the pile length by 50% considered in Fig. 13. Thus, than the rate for the displacement. The reason is that with increas-
to improve monopile performance the increase of the pile length is ing load level the point of rotation of the pile moves towards the
much more effective than the increase of pile diameter. pile tip, so that the pile rotation increase becomes smaller than
The loading conditions are represented by the amplitude of lat- the increase of the head displacement. This is of course only valid
eral load H and the moment arm h (with regard to seabed level, see if an almost rigid pile is considered.
732 M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735

(1) Define the tolerable lateral deformation of the wind tower at


seabed level yt.
(2) Determine the cyclic loading level H and h and number of
loading cycle N for design wind and wave load.
(3) Select a pile geometry, i.e. embedded length L and diameter
D. Then assess the lateral deformation of the monopile under
monotonic load from Fig. 11.
(4) From Fig. 12, determine the normalized load through the lat-
eral deformation obtained from step 3.
(5) Select the accumulation rate of lateral deformation from
Fig. 17 via normalized load obtained from step 4.
(6) The lateral deformation of the monopile for the chosen
geometry subjected to N cycles of the design load can be
determined from the accumulation rate obtained in step 5
multiplied by the lateral deformation under monotonic load
assessed in step 3.
(7) If the accumulated deformation exceeds the tolerant lateral
deformation, another pile geometry has to be selected.

An illustrative example of a monopile with L = 30 m and D = 5 m


subjected to a load of H = 15 MN at h = 30 m shall elucidate the use
Fig. 13. Effect of embedded length on the accumulation rate after 100 cycles for a of the design charts for estimating the lateral deformation after
monopile with D = 5 m, tp = 0.09 m, H = 5 MN and h = 20 m. 100 cycles. For a soil with unit buoyant weight of c0s ¼ 11 kN=m3 ,
the dimensionless parameter Hðh þ LÞ=c0s DL3 is 0.606. The static re-
sponse obtained from Fig. 11 is ys/L = 0.7%, i.e. the static displace-
ment is 21 cm. With ys/L = 0.7%, the normalized load H/Hu is
found by means of the chart in Fig. 12 to H/Hu = 0.24. The accumu-
lation rate after 100 cycles ys100/ys1 = 2.0 is obtained easily from
Fig. 17, so the lateral displacement at seabed level of the monopile
is 42 cm.
For the final design, numerical simulations should of course
be carried out with consideration of the specific site conditions.
This means the determination of a detailed soil model and the
assessment of the soil parameters in laboratory investigations
including cyclic triaxial tests to determine the cyclic parameters
b1 and b2 used in the proposed degradation stiffness model (see
Eq. (3)).

4. Design criteria concerning the required embedded pile length

The design guidelines for offshore wind turbines presented by


Germanischer Lloyd suggest as a design criterion for horizontally
loaded piles that the pile deflection line shall match the zero-toe-
kick or the vertical tangent condition to minimize the risk of
accumulated deformations under cyclic loading [27,28]. This
means that for the loading under consideration, the pile deflec-
Fig. 14. Effect of pile diameter on the accumulated rate after 100 cycles for a
monopile with L = 20 m, tp = 0.09 m, H = 10 MN and h = 4 m. tion line should exhibit two zero deflection points (zero-toe-kick)
or a least a vertical tangent at a certain depth below the point of
rotation. This requirement stems from the idea that a pile
3.4. Design charts for cyclic loading condition clamped in the soil in this way will be insensitive to cyclic load-
ing and limits the accumulation of pile deflection. For offshore
From the numerical results given in Figs. 13–16 it is evident piles of usual diameters up to 2 or 2.5 m, this requirement has
that the cyclic pile performance is mainly dependent on the nor- proved to be reachable. However, for very large diameters and
malized load level H/Hu. This can be used for the development of thus very stiff monopiles this requirement leads to very large
a design chart regarding cyclic behavior. embedded lengths.
In Fig. 17, all accumulation rates obtained in the parametric In Fig. 18, calculation results for monopiles with diameters of
study are depicted together. Although a large scatter of the results 7.5 m and 2.5 m are compared. The smaller monopile exhibits
is evident, for a rough estimation of the accumulation rate the pre- much larger deformation accumulation than the larger monopile
sented straight lines and the design chart given in the right part of and thus has, for the loading conditions considered, a worse cyclic
Fig. 17 may be used. load performance. However, this pile by far fulfills the vertical tan-
gent and even the zero-toe-kick criterion, whereas the large diam-
3.5. Summary of the design procedure eter pile is too short regarding these criteria, but shows very good
cyclic performance.
Using the design charts given in Figs. 11, 12 and 17, a prelimin- Wiemann [29] showed that for large diameter monopiles an in-
ary design of the required geometry of a monopile in sand soil can crease of the pile length to ensure rigid clamping has almost no ef-
be carried out. The procedure for this is summarized as follows: fect on the lateral head displacement under static loading. Thus,
M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735 733

Fig. 15. Effect of load eccentricity h/L on the accumulated rate after 100 cycles for a monopile with D = 7.5 m, tp = 0.09 m and H = 15 MN.

leads to much shorter required lengths, which are of course depen-


dent on the magnitude of loading.
Although the suggested design deformation is just a reference
value, these results make evident that adopting zero-toe-kick or
vertical tangent as design criterion for large diameter monopiles
is too strong and not appropriate.

5. Conclusions

The degradation stiffness model is presented – a numerical con-


cept which is able to account for cyclic lateral loading of piles in
the determination of pile deformations in a realistic manner. A ser-
ies of triaxial tests is needed to assess the parameters describing
cyclic behavior of the soil. A comparison of model results with
existing test results for piles in sand shows quite good correspon-
dence. The calibrations for degradation stiffness model with large-
scale model tests are planned in the further researches in the Insti-
tute of Soil Mechanics, Foundation Engineering and Waterpower
Engineering, Leibniz University Hannover.
The degradation stiffness model is used to carry out parametric
studies on the impact of pile geometry, loading conditions and rel-
ative density of sand soil on the cyclic pile performance. It emerges
Fig. 16. Accumulated rate of lateral deformation and rotation after 100 cycles for a
that the displacement accumulation rate is strongly dependent on
monopile with L = 30 m, D = 5.0 m and tp = 0.09 m. the loading level, i.e. the ratio of actual load to ultimate load. Since
the ultimate load of a horizontally loaded pile is very much depen-
dent on the embedded pile length, the accumulation rate for a gi-
the rigid clamping criterion is obviously not suitable for large ven load is mainly governed by the embedded length of a
diameter monopiles. monopile. Of course, the pile diameter also affects the ultimate
In fact, the required embedded length of a monopile should de- load, but not to such an extent.
pend on the tolerable pile deformation which is to be expected Based on the results of the parametric study, design charts are
over the lifetime of the supported structure. Thus, the displace- developed for both static and cyclic loading. With these charts the
ments ys1 or ysN should be limited. A possible design criterion could static displacement of a monopile in sand at seabed level, the
be that the accumulated deformation ys100 should be less than 0.3% load level and, dependent on the number of load cycles, also
of embedded pile length. For this assumed criterion, the accumula- the displacements due to cyclic loading can be determined. The
tion rate of lateral deformation ys100/ys1 should be less than 1.5 and rough values obtained can be used for the purpose of preliminary
the lateral displacement after the first cycle ys1 should be less than design.
0.2% of embedded pile length, for instance (see Figs. 12 and 17). The suitability of the zero-toe-kick or vertical tangent criterion
For monopiles in dense sand with a moment arm of loading of often used as a design requirement for horizontally loaded offshore
h = 10 m the pile lengths obtained with this criterion are compared piles is checked for the case of large diameter monopiles and is
in Fig. 19 with the pile lengths necessary to ensure zero-toe-kick. found to be inappropriate. Using a maximum tolerable deforma-
For zero-toe-kick, the required lengths for monopiles with tion is recommended instead. However, since the model described
D = 5 m and D = 7.5 m are 42–43 m and 53–54 m, respectively, al- here to determine the cyclic deformations needs more verification,
most independent of the lateral load H. Consideration of the sug- at the time being conservative approaches should be used in prac-
gested design criterion ys100/L 6 0.3% (as well as ys100/ys1 6 1.5) tical design.
734 M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735

Fig. 17. Accumulated rate of displacement against normalized load under cyclic loading.

Fig. 18. Comparison of clamping effect for piles with different diameters.

federal state of Lower Saxony, Germany. The numerical simula-


tions were carried out on a high performance computer provided
by the North German Alliance for the Advancement of High-Perfor-
mance Computing (HLRN), whose support is gratefully
acknowledged.

References

[1] American Petroleum Institute. Recommended practice for planning, designing


and constructing fixed offshore platforms – working stress design: API
recommended practice 2A-WSD (RP2A-WSD). 21st ed. Dallas: API; 2000.
[2] Reese LC, Wang ST, Isenhower WM, Arréllaga JA, Handrix J. LPILE plus 4.0 – a
program for the analysis of piles and drilled shafts under lateral loads. Austin:
Ensoft Inc.; 2000.
[3] Hutchinson TC, Chai YH, Boulanger RW. Simulation of fill-scale cyclic lateral
load tests on piles. J Geotech Geoenviron Eng Div (ASCE) 2005;131(9):1172–5.
[4] Kagawa T. Cyclic and loading-rate effects on pile responses. In: International
conference on numerical methods in offshore piling, numerical methods in
Fig. 19. Required embedded length determined using different criteria. offshore piling. Nantes; 1986. p. 417–32.
[5] Trochanis AM, Bielak J, Christiano P. Three-dimensional nonlinear study of
piles. J Geotech Eng Div (ASCE) 1991;117(3):429–47.
[6] Trochanis AM, Bielak J, Christiano P. Simplified model for analysis of one or two
Acknowledgments piles. J Geotech Eng Div (ASCE) 1991;117(3):448–66.
[7] Zhang F, Rimura M, Nakai T. Modellings of pile foundations subjected to cyclic
lateral loading. In: Proceedings of the seventh international symposium on
The results presented in this paper were obtained as part of the numerical models in geomechanics, NUMOG VII, Graz, Austria; 1999. p. 437–
FORWIND research group project funded by the Government of the 42.
M. Achmus et al. / Computers and Geotechnics 36 (2009) 725–735 735

[8] Dietrich Th. Experimental study of flexible piles in sand cyclically displaced at [20] Gotschol A. Veränderlich elastisches und plastisches Verhalten nichtbindiger
low frequency. In: International symposium on testing in situ of concrete Böden und Schotter unter zyklisch-dynamischer Beanspruchung. PhD thesis,
structures, Budapest; 1977. p. 233–45. Department of Civil Engineering, Institute of Geotechnics, University of Kassel;
[9] Gudehus G, Hettler A. Cyclic and monotonous model tests in sand. In: 2002.
Proceeding of the international conference on soil mechanics and foundation [21] Werkmeister S. Permanent deformation behaviour of unbound granular
engineering, Stockholm, vol. 3; 1981. p. 211–14. materials in pavement constructions. PhD thesis, Department of Civil
[10] Hettler A, Gudehus G. Estimation of shakedown displacement in sand bodies Engineering, Institute of Urban Engineering and Road Construction,
with the aid of model tests. In: International symposium on soils under cyclic Technical University of Dresden; 2004.
and transient loading, Swansea, vol. 1; 1980. p. 3–8. [22] Kuo YS. On the behavior of large-diameter piles under cyclic lateral load. PhD
[11] Hettler A. Verschiebungen starrer und elastischer Gründungskörper in Sand thesis, Department of Civil Engineering and Geodetic Science, Institute of Soil
bei monotoner und zyklischer Belastung. PhD thesis, Department of Civil Mechanics, Foundation Engineering and Waterpower Engineering, Leibniz
Engineering, Geo- and Environmental Sciences, Institute of Soil Mechanics and University of Hannover; 2008.
Rock Mechanics, University of Karlsruhe; 1981. [23] Timmerman DH, Wu TH. Behavior of dry sands under cyclic loading. J Soil
[12] Hettler A. Verschiebungen von lotrecht mittig belasteten Einzelfunda menten Mech Found Div (ASCE) 1969;95(4):1097–112.
und horizontal belasteten Pfählen in Sand unter Schwellast. Bauingenieur [24] Achmus M, Abdel-Rahman K, Kuo YS, Peralta P. Untersuchungen zum
1984;59:351–5. Tragverhalten von Monopilegründungen under zyklischer Belastung. In:
[13] Lin SS, Liao JC. Permanent strains of piles in sand due to cyclic lateral loads. J Pfahlsymposium 2007, Braunschweig, Germany; 2007. p. 95–114.
Geotech Geoenviron Eng Div (ASCE) 1999;125(9):798–802. [25] Saglamer A. Model study of lateral loaded single piles. In: Proceedings of the
[14] Little RL, Briaud JL. Full scale cyclic lateral load tests on six single piles in sand. seventh European conference on soil mechanics and foundation engineering,
Miscellaneous paper GL-88-27. Texas: Geotechnical Division, Texas A&M The measurement, selection and use of design parameters in geotechnical
University; 1988. engineering, Brighton, England, vol. 2; 1979. p. 115–20.
[15] Long JH, Vanneste G. Effect of cyclic lateral loads on piles in sand. J Geotech [26] Manoliu I, Dimitriu DV, Dobrescu GH. Load-deformation characteristics of
Eng Div (ASCE) 1994;120(1):33–42. drilled piers. In: Proceedings of 11th international conference on soil
[16] BSH. Standard Baugrunderkundung-Mindestanforderungen für die Gründung mechanics and foundation engineering, San Francisco, vol. 3; 1985. p. 1553–
von Offshore-Windenergieanlagen (WEA) und die Verlegung der 8.
stromabführenden Kabel. Hamburg and Rostock: Bundesamt für [27] Germanischer Lloyd Wind Energie GmbH. Guideline for the certification of
Seeschfffahrt und Hydrographie; 2003. offshore wind turbines. Hamburg, Germany; 2005.
[17] Hibbitt, Karlsson & Sorensen Inc. ABAQUS standard (Version 6.5). Pawtucket; [28] Faber T, Klose M. Experiences with certification of offshore wind farms. In:
2005. Proceedings of the sixteenth international offshore and polar engineering
[18] Abdel-Rahman K, Achmus M. Finite element modelling of horizontally loaded conference, San Francisco, California; 2005. p. 375–82.
monopile foundations for offshore wind energy converters in Germany. In: [29] Wiemann J. Bemessungsverfahren für horizontal belastete Pfähle,
Proceedings of the international symposium on frontiers in offshore Untersuchungen zur Anwendbarkeit der p–y Methode, PhD thesis,
geotechnics, ISFOG, Perth, Australia; 2005. p. 391–96. Department of Civil Engineering, Institute of Foundation Engineering and
[19] Huurman M. Development of traffic induced permanent strain in concrete Soil Mechanics, University of Duisburg-Essen; 2007.
block pavements. Heron 1996;41(1):29–52.

You might also like