GeoFlorida 2010: Advances in Analysis, Modeling & Design

(GSP 199) © 2010 ASCE 1857

Finite Element Analysis of an Offshore Wind Turbine Monopile

E. N. Hearn1, E.I.T and L. Edgers2, PhD, PE

1
Tufts University, Department of Civil and Environmental Engineering, 200 College
Downloaded from ascelibrary.org by TEMPLE UNIVERSITY on 04/30/13. Copyright ASCE. For personal use only; all rights reserved.

Avenue, Medford, MA 02155; PH (781) 454-6952; email: [email protected]

2
Tufts University, Department of Civil and Environmental Engineering, 200 College
Avenue, Medford, MA 02155; PH (617) 627-5875; email: [email protected]

Abstract

Large diameter monopile foundations for offshore wind turbines are subject to
large horizontal loads and overturning moments. These foundations have diameters
ranging from four to seven meters with the potential to become larger with the
development of larger wind turbines. They are often analyzed by means of the p-y
method to characterize the pile-soil interaction. While the method is theoretically
rigorous, the input p-y curves, for example those recommended by the American
Petroleum Institute (API), are based upon very limited field data.
This paper describes some analyses of a large diameter monopile in dense
sand. The pile characteristics are representative of the foundations for a 3 to 5 MW
wind turbine. The soil characteristics are representative of dense sandy soils that may
be encountered at wind farm sites in the southern North Sea and offshore the
Northeast United States. The pile was modeled by the p-y method and also by 3D
finite element analysis (FEA). The paper describes the details of the FEA model
development, important in minimizing numerical inaccuracies in 3D analyses.
Equivalent p-y curves, back-calculated from the 3D FEA, may be useful for p-y
analyses, which are simpler than 3D FEA. The results suggest that the API method
over predicts soil resistance and under predicts pile deflection for large diameter
monopiles subjected to lateral load and in stiff soils.

Background

As interests regarding clean and sustainable energy have expanded, harvesting

wind power has become an appealing option for future energy demand. Consistently
steady winds typical of offshore environments provide a prime location for wind
farms in some parts of the world. Due to unpredictable and challenging weather
conditions and the resulting extreme loads on the wind turbines, research on reliable
and economically feasible support structures for ocean environments is necessary to
move forward with this clean energy option.
Current methods for the analysis and design of monopile foundations for the
support of offshore wind turbines use the p-y method of analysis (Reese and Van
Impe, 2001). This method treats the monopile as a beam-column subject to both axial
and lateral loads. It characterizes the response of soil to lateral loads by means of p-y
curves, which relate the lateral soil resistance per unit length (p) to the lateral

GeoFlorida 2010
 GeoFlorida 2010: Advances in Analysis, Modeling & Design
(GSP 199) © 2010 ASCE 1858

deflection of the pile (y). These curves are empirically derived from a limited
number of field tests (Reese et al., 1974, Reese and Welch, 1975). Thus while the p-
y method is theoretically rigorous, the input p-y curves are based upon very limited
field data, particularly on large diameter piles.

Objectives and Overview of Analyses

This paper describes some analyses of a large diameter monopile in dense

Downloaded from ascelibrary.org by TEMPLE UNIVERSITY on 04/30/13. Copyright ASCE. For personal use only; all rights reserved.

sand. The pile and load characteristics are representative of the foundations for a 3 to
5 MW wind turbine. The soil characteristics are representative of dense sandy soils
that may be encountered at wind farm sites in the southern North Sea and offshore the
Northeast United States. The pile was modeled by the p-y method, using p-y curves
recommended by API (2000) and the computer program LPile (Ensoft, 2006) and
also by 3D finite element analysis (FEA). The paper describes the details of the FEA
model development, important in minimizing numerical inaccuracies in 3D analyses.
It presents equivalent p-y curves that were back calculated from the 3D FEA and
describes some parametric studies on the effects of soil stiffness and pile diameter on
p-y curves. All loads used for the following analyses are equivalent static loads. The
effects of repeated and dynamic loading, including pore pressure development and
dissipation, are beyond the scope of this paper.

p-y Curves for Sand

API (2000) recommends the use of p-y curves for sands which are based on
full-scale field tests reported by Reese et. al. (1974) at a field site located at Mustang
Island in Texas. The test piles had a 61 cm diameter, and extended 21 m into the
subsurface. These p-y curves are estimated first by determining over a range of
depths the ultimate soil resistance as the smaller of the two calculated values:
pus = (C1 x + C2 b)γ ' x Equation 1

pud = C3bγ ' x , Equation 2

where Pus and Pud are the ultimate resistance corresponding to shallow and deep
failure mechanisms respectively. The coefficients C1, C2, and C3 are found using
Figure 1, γ’ is the effective soil weight, x is the depth below the seafloor, and b is the
average pile diameter. At each depth, the p-y curve is then estimated by the equation:
⎛ kx ⎞
p = Apu tanh⎜ y⎟ . Equation 3
⎝ Apu ⎠
The factor A accounts for cyclic and static loading recommended by API, k is the
initial modulus of subgrade reaction determined by Es = kx where Es is the soil
modulus of elasticity and x is the depth below seafloor. y is the lateral deflection.
API (2000) suggests k-values for use in dense, medium, and loose sands below the
water table of 34 MN/m3, 16.3 MN/m3, and 5.4 MN/m3, respectively.

GeoFlorida 2010
 GeoFlorida 2010: Advances in Analysis, Modeling & Design
(GSP 199) © 2010 ASCE 1859
Downloaded from ascelibrary.org by TEMPLE UNIVERSITY on 04/30/13. Copyright ASCE. For personal use only; all rights reserved.

Figure 1: Coefficients for ultimate soil resistance as a function of friction angle (API,
2000)

Some studies have shown that p-y curves determined by this method are
inaccurate for large diameter piles such as those for the support of offshore wind
turbines. Abdel-Rahman and Achmus (2005) observed that the p-y method
significantly underestimates the deflection of large diameter piles compared with the
finite element method when large diameter piles are analyzed. Lesny and Weimann
(2005) observed that the standard p-y method overestimates the pile-soil-stiffness of
large diameter monopiles.

Finite Element Model

A 3D finite element model of a large diameter monopile in dense sand was

developed using the program Plaxis 3D Foundation (Plaxis, 2008). The monopile has
a diameter of 5 meters and a wall thickness of 55 mm, extending 26 meters into the
seafloor. The monopile steel properties were modeled as linear elastic with E =
2.10E+08 kN/m2, and Poisson’s Ratio of 0.2. The material characteristics used for the
dense sand are presented in Table 1. The modulus of elasticity of the sand is assumed
to follow the equation Es = kx, increasing linearly with depth at a rate corresponding
to the API (2000) recommended k value.

Table 1: Soil Model Parameters for Baseline Case Analysis

Material Model Mohr-Coulomb
Drained
Parameter Symbol Value
Saturated Density γsat 20 kN/m3
Friction Angle φ 37.5°
Poisson's Ratio ν 0.25
Dilation Angle ψ 7.5°
Interface Strength Rinter 0.65

GeoFlorida 2010
 GeoFlorida 2010: Advances in Analysis, Modeling & Design
(GSP 199) © 2010 ASCE 1860

Plaxis 3D Foundation formulates a set of piecewise continuous functions over

a set of stress points and nodes across the three dimensional model. To develop a
finite element mesh with sufficient fineness and appropriate geometry, a number of
initial studies were conducted. These studies considered the use of symmetry in the
model, the possibilities of terminating the monopile at the seafloor, and mesh
fineness. The model used for the analyses in this paper is shown in Figure 2 with the
finite element mesh and monopile highlighted. It consists of a fine mesh with ≈
109,000 nodes and isoparametric interface elements with an interface factor of one
Downloaded from ascelibrary.org by TEMPLE UNIVERSITY on 04/30/13. Copyright ASCE. For personal use only; all rights reserved.

(no strength reduction) around the monopile. The model was verified by analyzing a
case study from the literature (Abdel-Rahman and Achmus, 2005).

Figure 2: Finite element mesh used in analyses, constructed using Plaxis 3D

Foundation.

The p value is determined by integrating, over a unit depth, the stresses on the leading
edge of the pile and subtracting the integral of the stresses along the trailing edge of
the pile as a function of pile deflection. Figure 3 shows a detailed plan view of the
nodes and stress points adjacent to the pile at 10 meters depth. Only the stresses
parallel to the applied force on the pile were included in this analysis, assuming that
transverse stresses cancel each other out. Equation 4 describes this integration:
d d
Soil Resistance = ∫ σ x dx − ∫ σ x dx Equation 4
0 0
LeadingEdge TrailingEdge

where d is the diameter of the pile and σx are the stresses on the pile, and the resulting
soil resistance value represents the stresses over a unit depth interval of soil. The
leading edge unit forces are larger than the trailing edge because these are the forces

GeoFlorida 2010
 GeoFlorida 2010: Advances in Analysis, Modeling & Design
(GSP 199) © 2010 ASCE 1861

resisting the lateral load while the trailing edge forces are relieved due to the pile
displacement. The integration is performed for a range of instantaneous horizontal
loads and displacements to back-calculate a p-y curve in this manner.
Downloaded from ascelibrary.org by TEMPLE UNIVERSITY on 04/30/13. Copyright ASCE. For personal use only; all rights reserved.

Figure 3: 3-D perspective and plan views of the nodes adjacent to the pile used to
integrate stresses, constructed using Plaxis 3D Foundation.

Finite Element Model Results

Figure 4 presents p-y curves back-calculated from the FEA analyses of the 5.1
m diameter monopile at depths of 1.5, 5, and 10 meters below the seafloor Also
shown are the p-y curves estimated the by the API (2000) method, Equation 3.These
p-y curves become steeper and higher with increasing depths, consistent with the p-y
curves derived from experiments by Reese et al (1974). However, the curves
estimated by the API (2000) method are much steeper and higher than the FEA
generated curves.

GeoFlorida 2010
 GeoFlorida 2010: Advances in Analysis, Modeling & Design
(GSP 199) © 2010 ASCE 1862

8000
Depth = 10, Plaxis
7000 Depth = 10, API
Depth = 5, Plaxis
6000 Depth = 5, API
Soil Resistance, p (kN/m)
Depth = 1.5, Plaxis
Depth = 1.5, API
5000

4000
Downloaded from ascelibrary.org by TEMPLE UNIVERSITY on 04/30/13. Copyright ASCE. For personal use only; all rights reserved.

3000

2000

1000

0
0.00 0.02 0.04 0.06 0.08 0.10 0.12
Deflection, y (m)

Figure 4: Comparison of API and FEA d p-y curves. k= 16,300kN/m3 , d- 5.1 m

Figure 5 compares the API (2000) and FEA generated p-y curves from an
analysis at three depths, 1.5, 5, and 10 m, for a pile diameter of 2 meters. The back-
calculated API (2000) p-y curves at the depths of 1.5 and 5 m indicate better
agreement with the FEA generated curves for this smaller diameter pile than with the
5.1 diameter pile of Figure 4. For both pile diameters, the p-y curves show better
agreement with decreasing depth.

8000
Depth = 10, Plaxis
7000 Depth = 10, API
Depth = 5, Plaxis
Soil Resistance, p (kN/m)

6000 Depth = 5,API

Depth = 1.5, Plaxis
5000 Depth = 1.5, API
4000

3000

2000

1000

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Deflection, y (m)

Figure 5: Comparison of API and FEA p-y curvess. k= 16,300kN/m3, d= 2 m

GeoFlorida 2010
 GeoFlorida 2010: Advances in Analysis, Modeling & Design
(GSP 199) © 2010 ASCE 1863

These observations are consistent with the previous studies by Abdel-Rahman

and Achmus(2005) who have shown that the API (2000) p-y curves agrees better with
the results of finite element analysis for smaller diameter piles than large monopiles.

Influence of Soil Stiffness on p-y Curves

A parametric study was performed to evaluate the effect of the soil stiffness
(k) on p-y curves. The k-values used for this analysis are 80,000 kN/m3,16,300
Downloaded from ascelibrary.org by TEMPLE UNIVERSITY on 04/30/13. Copyright ASCE. For personal use only; all rights reserved.

kN/m3, 3,600 kN/m3, and 2,000 kN/m3 to represent extremely dense, medium-dense,
loose, and very loose sand densities, respectively. All of the p-y curves in this section
were calculated at 1.5 meters below the seafloor for a 2 meter pile with a 6 MN
vertical and a 4 MN horizontal equivalent static load. Byrne and Houlsby(2003)
describe these as representative equivalent static loads for large offshore wind
turbines. The resulting p-y curves are presented in Figure 6.
The curves for the two highest stiffnesses, Figure 6(a) and 6(b), show a very
large difference in the ultimate soil resistances. The FEA curve, Figure 6(a),
indicates an ultimate soil resistance of approximately 325 kN/m while the API (2000)
curve suggests an ultimate soil resistance of nearly 500 kN/m. The FEA and API
(2000) curves for the lower stiffnesses, Figure 6(c) and 6(d), show much better
agreement.

600
600
Soil Resistance, p (kN/m)
Soil Resistance, p (kN/m)

500
400 400

300
200 200

100 API API

FEA FEA
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0 0.05 0.1
Deflection, y (m) Deflection, y (m)

(a) k =80,000 kN/m³ (b) k =16,300 kN/m³

600 600
Soil Resistance, p (kN/m)

Soil Resistance, p (kN/m)

500 500

400 400
300 300
200 200
100
API API
100
FEA FEA
0 0
0 0.1 0.2 0.3 0 0.2 0.4 0.6 0.8
Deflection, y (m) Deflection, y (m)

(c) k =3,600 kN/m³ (d) k =2,000 kN/m³

Figure 6: Effects of stiffness modulus(k) on p-y curves, depth = 1.5 m, diameter = 2 m

GeoFlorida 2010
 GeoFlorida 2010: Advances in Analysis, Modeling & Design
(GSP 199) © 2010 ASCE 1864

Discussion of p-y parametric studies

Figures 4, 5, and 6 compare the p-y curves back calculated by FEA, with the p-
y curves from the API (2000) method. These figures indicate the API method over
predicts soil resistance in most cases, especially at large depths and for the larger
diameter monopile. These differences may occur because the API (2000) p-y curves
Downloaded from ascelibrary.org by TEMPLE UNIVERSITY on 04/30/13. Copyright ASCE. For personal use only; all rights reserved.

are based on limited experimental data on a 0.61 meter pile. The API and FEA p-y
curves agree better for the smaller diameter monopile, as shown by Figure 5. There
may be some issues associated with the FEA, including the effects of numerical error
due to mesh discretization. Also Plaxis 3D computations include the effects of the
structural deformation of the pile. This is illustrated in Figure 7 which shows that the
deflections at the seafloor in the direction of loading are larger at the trailing side than
the leading side. This occurs because Plaxis 3D models the structural behavior of the
monopile and computes its in-plane deflections (ovaling) as well as the secondary
moments. This may be important for the structural design of the monopile.

Figure 7: Deflections in the z-direction. Direction of loading in + z direction

Summary and Conclusions

This paper reports on the development of a 3D finite element model to analyze

large diameter monopile foundations for the support of offshore wind turbines. It
presents a method for back-calculating p-y curves from 3D FEA for monopiles in
sand. It compares these results with p-y curves estimated by the API (2000) method
and also performs parametric studies on the effects on pile diameter and soil stiffness
on these p-y curves. These studies show that the API (2000) method may compute p-
y curves that are higher and steeper than the p-y curves back-calculated from FEA,

GeoFlorida 2010
 GeoFlorida 2010: Advances in Analysis, Modeling & Design
(GSP 199) © 2010 ASCE 1865

especially for large diameter piles and for dense soils. The differences occur because
the API (2000) recommendations for p-y curves in sand are based on a very limited
data base. FEA models the structural behavior of the monopile and computes its in-
plane deflections (ovaling) as well as the secondary moments. This may be important
for the structural design of the monopile.
Downloaded from ascelibrary.org by TEMPLE UNIVERSITY on 04/30/13. Copyright ASCE. For personal use only; all rights reserved.

References

Abdel-Rahman, K. and Achmus, M. (2005) “Finite element modeling of horizontally

loaded monopile foundations for offshore wind energy converters in
Germany” Frontiers in Offshore Geotechnics: ISFOG: 391-396.
API (2000) American Petroleum Institute. Recommended Practice for Planning,
Design and Constructing Fixed Offshore Platform, API Recommended
Practice 2A-WSD (RP2A-WSD) 21st edition, Dallas.
Byrne, B. and Houlsby, G.; (2003)” Foundations for Offshore Wind Turbines“.
Philosophical Transactions: Mathematical, Physical and Engineering
Sciences, Vol. 361, No. 1813 pp.02909-2930.
Ensoft Incoroporated (2006) LPile Plus for Networks, version 5.0.29
Lesny, K. and J. Wiemann. (2005) “Design aspects of monopiles in German
offshore wind farms.” Frontiers in Offshore Geotechnics; ISFOG: 383-389.
Plaxis 3D Foundation.(2008) User’s Manual. Version 2.
Reese, L.C., Cox, W.R., and Koop, F.D. (1974) “Analysis of laterally loaded piles in
sand,” 6th Annual Offshore Technology Conference, Vol. 2: 473-484.
Reese, L.C. and Welch, R.C. (1975) “Lateral Loading of Deep Foundations in Stiff
Clay”, Journal of the Geotechnical Engineering Division, American Society
of Civil Engineers, Vol. 101, No. GT7, February, pp. 633-649.
Reese, Lymon C. and Van Impe, William, F. (2001) Single Piles and Pile Groups
Under Lateral Loading. Published by CRC Press.

GeoFlorida 2010