JacketSE: An Offshore Wind

Turbine Jacket Sizing Tool

Theory Manual and Sample Usage
with Preliminary Validation
Rick Damiani
National Renewable Energy Laboratory

NREL is a national laboratory of the U.S. Department of Energy

Office of Energy Efficiency & Renewable Energy
Operated by the Alliance for Sustainable Energy, LLC
This report is available at no cost from the National Renewable Energy
Laboratory (NREL) at www.nrel.gov/publications.

Technical Report
NREL/TP-5000-65417
February 2016

Contract No. DE-AC36-08GO28308

 JacketSE: An Offshore Wind
Turbine Jacket Sizing Tool
Theory Manual and Sample Usage
with Preliminary Validation
Rick Damiani
National Renewable Energy Laboratory
Prepared under Task No. WE11.5084

NREL is a national laboratory of the U.S. Department of Energy

Office of Energy Efficiency & Renewable Energy
Operated by the Alliance for Sustainable Energy, LLC
This report is available at no cost from the National Renewable Energy
Laboratory (NREL) at www.nrel.gov/publications.

National Renewable Energy Laboratory Technical Report

15013 Denver West Parkway NREL/TP-5000-65417
Golden, CO 80401 February 2016
303-275-3000 • www.nrel.gov
Contract No. DE-AC36-08GO28308
 NOTICE

This report was prepared as an account of work sponsored by an agency of the United States government.
Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty,
express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of
any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately
owned rights. Reference herein to any specific commercial product, process, or service by trade name,
trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation,
or favoring by the United States government or any agency thereof. The views and opinions of authors
expressed herein do not necessarily state or reflect those of the United States government or any agency thereof.

This report is available at no cost from the National Renewable Energy

Laboratory (NREL) at www.nrel.gov/publications.

Available electronically at SciTech Connect http:/www.osti.gov/scitech

Available for a processing fee to U.S. Department of Energy

and its contractors, in paper, from:

U.S. Department of Energy

Office of Scientific and Technical Information
P.O. Box 62
Oak Ridge, TN 37831-0062
OSTI http://www.osti.gov
Phone: 865.576.8401
Fax: 865.576.5728
Email: [email protected]

Available for sale to the public, in paper, from:

U.S. Department of Commerce

National Technical Information Service
5301 Shawnee Road
Alexandria, VA 22312
NTIS http://www.ntis.gov
Phone: 800.553.6847 or 703.605.6000
Fax: 703.605.6900
Email: [email protected]

Cover Photos by Dennis Schroeder: (left to right) NREL 26173, NREL 18302, NREL 19758, NREL 29642, NREL 19795.

NREL prints on paper that contains recycled content.

Executive Summary

This manual summarizes the theory and preliminary verifications of the JacketSE module, which is an offshore
jacket sizing tool that is part of the Wind-Plant Integrated System Design & Engineering Model toolbox. JacketSE
is based on a finite-element formulation and on user-prescribed inputs and design standards’ criteria (constraints).
The physics are highly simplified, with a primary focus on satisfying ultimate limit states and modal performance
requirements. Preliminary validation work included comparing industry data and verification against ANSYS, a
commercial finite-element analysis package. The results are encouraging, and future improvements to the code are
recommended in this manual.

iv
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Acknowledgments

This work was supported by the U.S. Department of Energy under Contract No. DE-AC36-08GO28308 with the
National Renewable Energy Laboratory. Funding for the work was provided by the DOE Office of Energy Efficiency
and Renewable Energy, Wind and Water Power Technologies Office.

Table of Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Lattice Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.1 Member Definition and Tube Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Soil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Piles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3.1 Axial Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3.2 Pile Head Stiffness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Legs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.5 X-Braces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.6 Mud-Brace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.7 Top Brace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.8 transition piece (TP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.9 Tower . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.10 Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.10.1 rotor nacelle assembly (RNA) Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.10.2 Aerodynamic Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.10.3 Hydrodynamic Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.10.4 Loads Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.11 FEA Solver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.12 Utilization Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.12.1 Safety Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.12.2 Tower Utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.12.3 Jacket Utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3 Preliminary Verification and Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

5 Conclusions and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

List of Figures

Figure 1. JacketSE’s main geometry groups definitions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Figure 2. Leg-pile connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

v
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Figure 3. Defintions of key variables in the geometry used by JacketSE . . . . . . . . . . . . . . . . . . . . 14
Figure 4. Schematics of the intersection between two x-braces . . . . . . . . . . . . . . . . . . . . . . . . 17
Figure 5. Diagram showing the structural simplification adopted by JacketSE to represent the TP with a
frame of beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Figure 6. The main design variables and parameters for the tower model adopted by JacketSE . . . . . . . . 22
Figure 7. Main reference system used in JacketSE and principal sources of loading and their general areas
of application. Original illustration by Joshua Bauer, NREL (modified here) . . . . . . . . . . . . . 23
Figure 8. Typical unit vectors associated with x-joint and k-joint used in JacketSE to determine IP and OP
bending moment components with respect to the joint plane . . . . . . . . . . . . . . . . . . . . . . . . . 39
Figure 9. The element (F F 2 ), intermediary (F F 1 ), and global (F F 0 ) coordinate system used in Jack-
etSE. The generic element is shown as an AB cylindrical rod; the other symbols are explained in the text . 40
Figure 10. Results from the ANSYS analysis of a jacket-tower-pile configuration for a 10 MW offshore
wind turbine (OWT); (a) shows the first eigenmode; (b) shows the von-Mises stress distribution in the
jacket members. From Damiani and Song (2013) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Figure 11. Jacket steel mass trend for 6 MW turbine configurations. The study 81 data points are denoted
by filled circles, with colors indicating RNA mass as denoted by the legend in (b). The surface (bilinear
interpolation of the data points) is color-coded by jacket mass tonnage, and the legend is given in (a). In
all the plots, the z-axis shows jacket mass made non-dimensional with its average across all the 6 MW
cases. Other symbols indicate: existing installations of 6 MW offshore turbines (triangles); predictions
from the Crown Estate study BVG Associates 2012 (diamonds); and predictions from GL Garrad Hassan
2012 (squares). From Damiani et al. (2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Figure 12. Diagram of the jacket-tower configuration as calculated by the optimizer for this case study . . . . 44
Figure 13. Calculated tower utilizations for the mass-optimized jacket-tower configuration used in this case
study. VonMises Util1 and Util2 refer to utilizations with respect to yield strength, GL Util1 and Util2
refer to global (Eulerian) buckling criteria, and EUsh Util1 and Util2 refer to the shell buckling criteria.
GL and EUsh Util1’s refer to the first design load case (DLC), while GL and EUsh Util2’s refer to the
second . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

List of Tables

Table 1. Examples of JacketSE’s Geometry Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Table 2. Tube Class Variables and Parameters Used in the Definition of the Members in JacketSE . . . . . . 6
Table 3. Typical Stratigraphy Arrangement used by JacketSE . . . . . . . . . . . . . . . . . . . . . . . . . 7
Table 4. Variables and Parameters Used in the Definition of the Pile Members in JacketSE . . . . . . . . . 8
Table 5. Design Parameters for Cohesionless Soil (API 2014) . . . . . . . . . . . . . . . . . . . . . . . . . 9
Table 6. Coefficient of Subgrade Reaction, ks , for Cohesionless Soils as a Function of Friction Angle (from
API 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Table 7. Variables and Parameters Used in the Definition of the Leg Members in JacketSE . . . . . . . . . 14
Table 8. Variables and Parameters Used in the Definition of the X-brace Members in JacketSE . . . . . . . 16
Table 9. Variables and Parameters Used in the Definition of the Mud-brace Members in JacketSE . . . . . . 18

vi
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Table 10. Variables and Parameters Used in the Definition of the Top-brace Members in JacketSE . . . . . . 18
Table 11. Variables and Parameters Used in the Definition of the TP Members in JacketSE . . . . . . . . . . 20
Table 12. Variables and Parameters Used in the Definition of the Tower in JacketSE . . . . . . . . . . . . . . 21
Table 13. Additional Parameters Used in the Definition of the Loads in JacketSE . . . . . . . . . . . . . . . 25
Table 14. Input Parameters for FRAME3DD Handled by JacketSE’s Inputs—For More Details See Gavin
(2010) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Table 15. Examples of ultimate limit state (ULS), serviceability limit state (SLS), and fatigue limit state
(FLS) Load partial safety factors (PSFs)—From IEC 2005 . . . . . . . . . . . . . . . . . . . . . . . . . 30
Table 16. Examples of Minimum γn As a Function of Component Class—From IEC (2005) . . . . . . . . . 31
Table 17. Examples of Minimum γm As a Function of Failure Mode—From IEC (2005) . . . . . . . . . . . 31
Table 18. Values for λ̂0 ,η, and β p As a Function of Stress Type—From European Committee for Standardi-
sation (1993) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
d jb
Table 19. Values for c1 ,c2 , and c3 As a Function of Loading Conditions and Joint Type and β j = d jc —From
API (2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
d jb
Table 20. Values for Qub As a Function of Brace Loading and Joint Type and β j = d jc —From API (2014) . . 38
Table 21. Turbine and Substructure Parameters for the Case Study . . . . . . . . . . . . . . . . . . . . . . . 45
Table 22. Environmental Parameters Used for the Case Study . . . . . . . . . . . . . . . . . . . . . . . . . 45
Table 23. Key Constraints Used in This Study for the Sizing Tool . . . . . . . . . . . . . . . . . . . . . . . 46
Table 24. Values of the Design Variables for the Calculated Minimum Mass Configuration . . . . . . . . . . 47

List of Acronyms and Symbols

Acronyms

t Wall thickness

AHSE Aero-hydro-servo-elastic

BOS Balance-of-system

CAE Computer-aided engineering

CapEx Capital expenditure

DLC Design load case

FAST v8 National Renewable Energy Laboratory (NREL)’s aero-elastic tool

FEA Finite-element analysis
FLS Fatigue limit state

LCOE Levelized cost of energy

LRFD Load Resistance Factor Design

MSL Mean sea level

vii
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
NOAA National Oceanic and Atmospheric Administration
NREL National Renewable Energy Laboratory

O&G Oil and gas

O&M Operation and maintenance
OD Outer diameter
OEM Original equipment manufacturer
OR Outer radius
OWT Offshore wind turbine

PSF Partial safety factor

RNA Rotor nacelle assembly

SACS A commercial package by Bentley for the analysis of offshore fixed-bottom structures
SLS Serviceability limit state
SSt Support structure

TP Transition piece

ULS Ultimate limit state

WISDEM Wind-Plant Integrated System Design & Engineering Model

Symbols

2D Two-dimensional
AB Distance between the first two leg joints at the bottom bay
A1 First point of segment A1 A2
A2 Second point of segment A1 A2
Ap Surface area of the pile tip
As Side surface area of the pile
Abrc Brace cross-sectional area
A jnt Factor in the calculation of Q f c per API 2014
Aleg Leg cross-sectional area
Amid Area inscribed by the midthickness line
A Member cross-sectional area
B1 First point of segment B1 B2
B2 Second point of segment B1 B2
CM Center of mass
CPf g Flag indicating whether (True) or not (False) the legs are considered clamped at the seabed
Cτ Factor in the expression of τzθ ,Rcr
Cθ Factor in the expression of σθ ,Rcr
Cc Reduction factor in the calculations of the axial stress allowables for members under compression per
API 2014
Cm Reduction factor in the calculations of the utilization for members under compression and bending per
API 2014
Cz Factor in the expression of σz,Rcr
Cθ ,s Factor in the expression of σθ ,Rcr
C f xe Reduction factor in the calculation of the elastic buckling stress per API 2014, which is normally equal to
0.3

viii
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Cl2g Transformation matrix from local to global coordinate system
DT Rb Tower-base Diameter to thickness ratio (DT R)
DT Rt Tower-top DT R
DT R Diameter to thickness ratio
D1 Bottom Outer diameter (OD) of a tapered shell member
D2 Top OD of a tapered shell member
Db Tower-base OD
Dp Pile outer diameter
Dt Tower-top OD
DT Pbrc TP cross-brace OD
Dbrc1 Dbrc for brace 1 at the k-joint
Dbrc2 Dbrc for brace 2 at the k-joint
Dbrc Brace outer diameter
Dgir TP girder OD
Dleg Leg outer diameter
Dlt Leg top OD
Dmbrc Mud-brace outer diameter
Dsh Shell OD
Dstmp TP stump OD
Dstrt TP strut OD
Dtbrc Top-brace outer diameter
Dxbrc X-brace outer diameter
D Generic member OD
Ep Pile Young’s modulus
Es Soil modulus or modulus of subgrade reaction (N m−2 )
E Young’s modulus
Fa Allowable axial compressive stress
Fb Allowable bending stress
Fd Generic, design (factored) load within the Load Resistance Factor Design (LRFD) approach
Fe t Euler stress divided by a safety factor per AISC 1989
Fk Generic, characteristic load within the LRFD approach
Fp,x Reaction force along x at pile head
Fp,y Reaction force along y at pile head
Fp,z Reaction force along z at pile head
FxRNA Force from the RNA along the x-axis
FyRNA Force from the RNA along the y-axis
FzRNA Aerodynamic force from the RNA along the z axis
Gf Gust factor
Gp Pile material shear modulus
Gs Initial soil shear modulus at depth of interest
Gs,c Soil shear modulus at torsional critical depth
G Shear modulus
IP In-plane
Ixx Mass second moment of inertia about the local x axis
Ixy Mass cross-moment of inertia about the x–y axes
Ixz Mass cross-moment of inertia about the x–z axes
Iyy Mass second moment of inertia about the local y axis
Iyz Mass cross-moment of inertia about the y–z axes
Izz Mass second moment of inertia about the local z axis
Jxx,p Pile cross-sectional, second area moment of inertia
Jxx Member cross-sectional, second area-moment of inertia

ix
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
KLRbrc Brace KLR
KLR Buckling parameter kbuck ltube /rgyr
Kg Stiffness matrix referred to global coordinate system
Kl Stiffness matrix referred to local coordinate system
Kp Modulus ratio used in Pender’s method
KAPI Coefficient of lateral earth pressure
La Pile active length
Lc Pile critical embedment length for torsion response
Lp Pile embedment length
L Generic member length
Ma Allowable capacity for brace bending load
Md Design (factored) bending moment load at the station of interest
Mj Brace bending load at the joint
Mp Bending moment load resistance at the tower station of interest
Mx Component of the bending moment along the x-axis at the station of interest
My Component of the bending moment load along the y-axis at the station of interest
Mz Torsion moment load along the z-axis at the station of interest
Mc,IP Chord IP bending load at the joint
M jnt,x Local x component of Mjnt
M jnt,y Local y component of Mjnt
M jnt,z Local z component of Mjnt
Mmod Method to be used for dyanmic eigenvalue: 1=Subspace-Jacobi iteration; 2= Stodola (matrix iteration)
method Gavin 2010
M p,x Reaction moment along x at pile head
M p,y Reaction moment along y at pile head
M p,z Reaction moment along z at pile head
M pc Plastic moment capacity of the chord at the joint
MxRNA RNA aerodynamic moment along the x-axis
MyRNA RNA aerodynamic moment along the y-axis
MzRNA RNA aerodynamic moment along the z-axis
Nd Design (factored) normal load at the tower station of interest
Ne Elastic buckling resistance
Np Normal (axial) load resistance at the tower station of interest
Nq Dimensionless bearing capacity factor
OP Out-of-plane
P−Δ P-Δ effect
PPf g Flag indicating whether (True) or not (False) the piles are considered ’plugged’
Pa Allowable capacity for brace axial load
Pc Chord axial load at the joint
Pj Brace axial load at the joint
Pyc Yield capacity of the chord at the joint
Q−z Nonlinear spring treatment of soil-pile end-bearing, axial stiffness
Qd Ultimate bearing capacity of pile
Qf Skin friction resistance
Qg Gap factor at a joint per API 2014
Qp End bearing resistance
Qbeta Geometric factor at a joint per API 2014
Qfc Chord load factor per API 2014
Qub Strength factor per API 2014
Q Meridional compression fabrication quality parameter European Committee for Standardisation 1993
R( fd ) Probability distribution of the generic, design (factored) material resistance within the LRFD approach

x
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Rdx Method for matrix condensation: 0= none; 1= static; 2=Guyan; 3=dynamic Gavin 2010
S(Fd ) Probability distribution of the generic, design (factored) load within the Load Resistance Factor Design
approach
T Plvl Level of automatic build for the TP
Td Design (factored) shear load at the station of interest
Tw Wave spectral period
Tx Component of the shear load along the x-axis at the station of interest
Ty Component of the shear load along the y-axis at the station of interest
Tmr Relative stiffness factor from Matlock and Reese 1960
V Pf g Flag indicating whether the piles are vertical (True) or battered (False)
Wp Cross-sectional bending modulus
[Cφ ] Matrix describing the transformation of coordinate systems from F F 1 to F F 2 via a rotation about the
local x
[Cψ,θ ] Matrix describing the transformation of coordinate systems from F F 0 to F F 1 via a rotation θel about
the global y and a rotation ψel about global z
[Cel ] Matrix identifying the local element coordinate system, i.e., the transformation from F F 0 to F F 2 ,
with local x along the member axis and local y, z along the cross-section principal axes of inertia
[Cjnt ] Direction cosine matrix identifying the joint plane, with x, y in the joint plane and z normal to it
Ĝ p Equivalent shear modulus used for piles
î Unit vector along the x-axis
jˆ Unit vector along the y-axis
CMoff Distance vector from the tower-top flange to the CM of the RNA
Cel,x First row of [Cel ]
Cel,y Second row of [Cel ]
Cel,z Third row of [Cel ]
Cjnt,x First row of [Cjnt ]
Cjnt,y Second row of [Cjnt ]
Cjnt,z Third row of [Cjnt ]
Dstem Array of TP central shell ODs (one per shell member)
Faero Force vector originating at the rotor
MIP IP component of the bending moment vector for the generic member at the joint of interest
MOP OP component of the bending moment vector for the generic member at the joint of interest
Maero Moment vector originating at the rotor
Mjnt Bending moment vector for the generic member at the joint of interest
TPmas TP lumped mass, including mass, mass tensor (Ixx –Izz ) and CM offset from the center of the deck
Uw Vectorial sum of the wave and current velocity
Uhub Wind velocity at hub height
U Wind velocity vector
δthoff Distance vector from the tower-top flange to the hub center
îbrc Unit vector identifying the x-brace longitudinal axis
îch Unit vector identifying the chord longitudinal axis
fa Force per unit length caused by wind aerodynamic drag
fw Force per unit length as a result of wave and current kinematics
hbys Array containing the bay heights
hstem Array of TP central shell lengths (one per shell member)
tstem TP central shell Wall thicknesses (t’s) (one per shell member)
uc Current velocity
uw Wave velocity
FF0 Global coordinate system
FF1 Auxiliary coordinate system
FF2 Local coordinate system

xi
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 L Ratio of the pile length to its diameter
R Ratio of pile Young’s modulus to soil modulus at pile tiplength to diameter
νs Soil Poisson’s ratio
A1 A2 Vector connecting A1 to A2
B1 B2 Vector connecting B1 to B2
ABc Vector resulting from the cross product of A1 A2 and B1 B2
A0 Set of coordinates for intersection between A1 A2 and B1 B2
A1 Set of coordinates for first point of segment A1 A2
gg Inertial frame acceleration for FRAME3DD
b Two-dimensional batter, i.e., the vertical-to-horizontal ratio of the jacket-leg slope on a 2D projection
c1 Factor in the calculation of Q f c per API 2014
c2 Factor in the calculation of Q f c per API 2014
c3 Factor in the calculation of Q f c per API 2014
cd Drag coefficient (wind or water)
cm Added mass coefficient
cu Undrained shear strength
cd,a j Air drag coefficient jacket
cd,at Air drag coefficient tower
cd,w j Water drag coefficient jacket
dw Water depth
d jb Brace OD in a joint verification (also txbrc )
d jc Chord OD at a joint (also Dleg )
dckw Deck-side length
f0 First natural frequency in Hz
fd Generic, design (factored) resistance within the LRFD approach
fk Generic, characteristic material resistance within the LRFD approach
fy Characteristic yield strength
fθ θ Flexibility matrix coefficient at pile head, representing the rotation about y associated with a unit moment
along y
fθ x Flexibility matrix coefficient at pile head, representing the displacement about y associated with a unit
moment along y
fmax Upper bound for th eshaft friction value for cohesionless soils
fstem Ratio between the TP central-shell wall thickness and tb
fxθ Flexibility matrix coefficient at pile head, representing the lateral displacement along x associated with a
unit moment along y
fxc Inelastic local buckling stress
fxe Elastic local buckling stress
fxx Flexibility matrix coefficient at pile head, representing the lateral displacement along x associated with a
unit force along x
fy2 Allowable used in local buckling axial stress determination per API 2014
fyb Brace allowable stress used in joint verification
fyc Chord allowable stress used in joint verification
f Skin friction
gp Gap between two braces at a k-joint per API 2014
gm f g Geometric stiffness effect flag for FRAME3DD
g Gravity acceleration
hc Height of a tapered shell member
hw Wave height, (peak-to-peak distance)
h2 f Fraction of tower length at constant cross section
hb,1 Height of the bottom bay
hb,i Height of the i-th bay, counting from the bottom bay up

xii
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
h jckt Height of the jacket available to the bays
hlb Distance from the leg-toe to the first joint with an x-brace
hstmp TP stump length
htwrb Tower buckling effective length, shortest distance between flanges
htwr Tower length
hydc Hydrostatic constant
i Generic index
kτ Exponent factor for the shear stress ratio, in the local buckling utilization calculation
kθ Exponent factor for the hoop stress ratio, in the local buckling utilization calculation
ki Interaction (axial-hoop stresses) factor in the local buckling utilization calculation
ks Coefficient of subgrade reaction
kw Dynamic pressure factor to calculate hoop stresses function of cylinder dimensions and external pressure
buckling factor per European Committee for Standardisation (1993)
kz Exponent factor for the axial stress ratio, in the local buckling utilization calculation
k θx y Stiffness matrix coefficient, representing the moment about x associated with a unit displacement along y
k θx θx Stiffness matrix coefficient, representing the moment about x associated with a unit rotation about x
k θy x Stiffness matrix coefficient, representing the moment about y associated with a unit displacement along x
k θy θy Stiffness matrix coefficient, representing the moment about y associated with a unit rotation about y
k θz θz Stiffness matrix coefficient, representing the torsional moment about z associated with a unit rotation
about z
kbuck Buckling parameter or effective length factor
kxθy Stiffness matrix coefficient, representing the lateral force along x associated with a unit rotation about y
kxx Stiffness matrix coefficient, representing the lateral force along x associated with a unit displacement
along x
kyθx Stiffness matrix coefficient, representing the lateral force along y associated with a unit rotation about x
kyy Stiffness matrix coefficient, representing the lateral force along y associated with a unit displacement
along y
kzz Stiffness matrix coefficient, representing the axial force along z associated with a unit displacement along
z
lm Method for mass modeling: 0= consistent mass matrix method; 1=lumped mass matrix method Gavin
2010
lb,1 Length of the bottom x-brace
ltube Tube object unsupported length
mRNA RNA mass
mat Object class defining material properties
nA Auxiliary quantity used in the calculation of the intersection between A1 A2 and B1 B2 segments
nbays Number of bays
ndiv,T Pbrc Number of elements in each of the TP cross-brace members
ndiv,gir Number of elements in each of the TP girder members
ndiv,leg Number of elements in the leg member
ndiv,mud Number of elements in the mud-brace member
ndiv,pile Number of elements in the pile member
ndiv,stmp Number of elements in each of the TP stumps
ndiv,strt Number of elements in each of the TP struts
ndiv,tbrc Number of elements in the top-brace member
ndiv,twr Number of elements in each of the two tower segments
ndiv,xbrc Number of elements in the x-brace member
ndiv Number of elements in the member under consideration
nlegs Number of legs in the substructure (equal to the number of piles)
nmod Number of eigenmodes to be calculated by FRAME3DD
nstem Number of TP shell members

xiii
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
p−y Nonlinear spring treatment of soil-pile lateral stiffness
pto Overburden pressure at the depth of interest
qa Soil allowable bearing-capacity
qp Unit end bearing capacity
qmax Maximum wind dynamic pressure
q p,max Upper bound for the unit end bearing capacity
req Equivalent untapered Outer radius (OR) of a tapered shell member
rgyr Cross-section radius of gyration
rig f lg Flag selecting how to treat the rigid connection at tower-top with the RNA
rtr Ratio of req to teq
sk Buckling length factor, set equal to 2 for the tower
sh f g Shear deformation effect flag for FRAME3DD
t −z Nonlinear spring treatment of soil-pile axial stiffness
t1 Bottom t of a tapered shell member
t2 Top t of a tapered shell member
tb Tower-base t
tp Pile wall thickness
tt Time variable
tT Pbrc TP cross-brace t
teq Equivalent untapered t of a tapered shell member
tgir TP girder t
t jb Brace t in a joint verification (also txbrc )
t jc Chord thickness in a joint verification (also tleg )
tleg Leg wall thickness
tmbrc Mud-brace wall thickness
tsh Shell t
tstmp TP stump t
tstrt TP strut t
ttbrc Top-brace wall thickness
txbrc X-brace wall thickness
tube Tube object class
tube Number of elements per member in each leg
u p,x Pile head displacement along x
u p,y Pile head displacement along y
u p,z Pile head displacement along z
wb Width of the jacket base (length of one side) at the seabed
wb,1 Horizontal distance between the leg-to-brace joints at bay 1 (bottom bay), i.e., bay-1 bottom width
wb,2 Second bay width
wb,i Width of i-th bay (bays counted from bottom up, 1..nbays
wdr Allowance for weldments as a function of member diameter
xleg Generic leg joint or node coordinate along x
x Global (or local) x-axis
yleg Generic leg joint or node coordinate along y
y Global (or local) y-axis
zd Depth below the seabed
zw Distance from the sea surface, positive upwards
zRNA Z coordinate of the RNA CM
zcmo f f Distance from the tower-top flange to the RNA CM along z
zdbot Deck underside elevation Mean sea level (MSL)
zhub Hub height above MSL
zlb Elevation of the leg-toe above the seabed

xiv
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
zleg Generic leg joint or node coordinate along z
ztb Elevation MSL of tower-base
ztho f f Distance from the tower-top flange to the hub center along z
z Global (or local) z-axis
z Altitude above MSL

ANSYS Ansys commercial Finite-element analysis (FEA) package

FRAME3DD Open-source FEA package, Gavin 2010

JacketSE Offshore jacket sizing tool, part of Wind-Plant Integrated System Design & Engineering Model (WIS­
DEM)

k-joint Joint at the intersection between leg and x-braces

mud-brace Mud-brace

pyFrame3DD Python wrapper for FRAME3DD

Quattropod® Patented jacket configuration by OWEC Tower AS

TowerSE Tower and monopile sizing tool, part of WISDEM

x-brace X-brace
x-joint Joint at the intersection betwee x-braces

Greek Symbols

Δn Factor accounting for member slenderness in the global buckling utilization calculation
Δz Step size for internal force calculations along the member axis for FRAME3DD
Δwk Characteristic imperfection amplitude European Committee for Standardisation 1993
Δzmx Maximum FEA element length for the tower elements
Φ Factor used in the flexural buckling reduction factor calculation
αb Imperfection factor used in the buckling calculation, set equal to 0.21 in JacketSE
αs Factor used in cohesive soils to calculate shaft friction
αbat,2D Two-dimensional batter angle
αbat,3D Three-dimensional batter angle
αbl1 Angle between brace 1 and leg at the k-joint
αbl2 Angle between brace 2 and leg at the k-joint
αimp Elastic imperfection reduction factor from European Committee for Standardisation 1993
α Wind power law exponent
λ̄ Reduced slenderness (see Germanischer Lloyd 2005)
βc Cone angle for the typical tapered shell element
βj Ratio of brace diameter to chord diameter
βm Bending moment coefficient in the global buckling utilization calculation
βp Plastic range factor in the shell buckling verification
β2D Brace-to-leg angle as measured on a vertical projection
β3D Actual brace-to-leg angle
χθ Buckling reduction factor for hoop strength in the shell buckling verification
χz Buckling reduction factor for axial strength in the shell buckling verification

xv
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
χ Generic buckling reduction factor in the shell buckling verification
δf Dynamic amplification factor
δs Soil-to-steel friction angle
δel,x Component along global x of δel
δel,y Component along global y of δel
δel,z Component along global z of δel
δsh f Frequency shift factor for rigid-body modes Gavin 2010
η Interaction exponent in the shell buckling verification
γb Safety factor used in buckling verification, usually set equal to 1.1 (Germanischer Lloyd 2005)
γc Ratio of d jc to twice the t jc at the joint
γf Generic load PSF
γm Material PSF
γn Consequence of failure PSF
γs Soil unit weight
γfa Aerodynamic load PSF
γfg Gravitational load PSF
γfw Hydrodynamic load PSF
γin Angle described by the horizontal projection of one side of the jacket base and a line connecting the
center of the polygon at the base and one end of that side
γ j1 Brace load PSF in a joint verification, which is normally set at 1.6
γ j2 Chord PSF in a joint verification, which is normally set equal to 1.2
λ̂0 Squash limit for reduced slenderness
λ̂ p Plastic limit for the reduced slenderness
λ̂ Reduced slenderness
κw Wave number
κ Reduction factor in the global buckling utilization calculation
δel Distance vector between two adjacent nodes of a FEA element
ν Poisson’s ratio
ωb Dimensionless length parameter for shell buckling calculations
ωw Wave frequency
φs Soil friction angle
φel Eulerian rotational angle about the local x
φrot Rotation angle about the local x-axis
ψs Factor used in cohensionless soils for the determination of the pile shaft friction
ψel Rotational angle about the global z (per FRAME3DD’s convention)
ψrot Rotation angle about the local z axis
ψ Rotor yaw angle about global z
ρa Air density
ρw Sea water density
ρ Material density
σa Normal stress caused by axial force
σb Normal stress caused by total bending moment
σθ ,Ed Hoop design (factored) stress
σθ ,Rcr Critical buckling hoop stress
σθ ,Rd Hoop buckling strength
σb,x Normal stress caused by bending about local x
σb,y Normal stress caused by bending about local y
σvm von-Mises stress
σz,Ed Axial (meridional) design (factored) stress
σz,Rcr Critical buckling axial stress
σz,Rd Axial (meridional) buckling strength

xvi
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
τzθ ,Ed Shear design (factored) stress
τzθ ,Rcr Critical buckling shear stress
τzθ ,Rd Shear buckling strength
θj Smaller angle described by the brace’s and chord’s axes at the joint
θel Rotational angle about the global y (per FRAME3DD’s convention)
θ p,x Pile head rotation along x
θ p,y Pile head rotation along y
θ p,z Pile head rotation along z
ζ Dummy coordinate along the z-axis

xvii
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
1 Introduction

The European Wind Energy Association (EWEA 2015) reports that in Europe, by the end of 2014, 78.8% of the
installed substructures were monopiles, with lattice structures such as jackets accounting for 4.7%, and the remainder
were gravity foundations, tripods, and tripiles. Although monopiles are still considered as preferred substructures
because of their ease of fabrication and installation, it is unclear whether this trend will apply in the United States,
where challenging bathymetry, soil conditions, and sea states may make monopiles economically less attractive. The
first U.S. offshore wind installation (Deepwater Wind offshore of Rhode Island) is making use of jackets.
Several studies (e.g., Musial, Butterfield, and Ram 2006; De Vries et al. 2011), have shown that monopiles, likely the
most readily available solution for shallow waters, are progressively unfeasible as projects are sited in deeper water
and use larger turbine sizes (6 MW and above). Monopiles require large structural mass to guarantee system modal
performance, and can become expensive and difficult to manufacture and install. In contrast, a lattice substructure
can deliver needed structural stiffness by efficiently increasing its footprint.
Typical design practice for an offshore wind turbine (OWT) assumes a fixed turbine (and, in most cases, turbine-
tower) configuration and requires multiple iterations to arrive at a final layout for the support structure (i.e., the
substructure, foundation, and tower). The substructure’s (and foundation’s) design is generally carried out by a dif­
ferent engineering entity, requiring, at each iteration, an exchange of loads and stiffness data with the turbine original
equipment manufacturers (OEMs). The geometry of the substructure is determined by satisfying the serviceability
limit state (SLS), fatigue limit state (FLS), and ultimate limit state (ULS) under combined turbine loads and hydro­
dynamic loads. A change in the substructure design directly impacts the system dynamics and structural integrity,
thus it must be followed by a reverification of the turbine loads’ envelope. As a result of this sequential approach
to the support structure design, aspects of the fully-coupled dynamics may be missed along with the risk of achiev­
ing suboptimal solutions. Because of the lack of fully coupled analyses, important trade-offs in the design of the
subsystems are not fully considered, and the resulting system cost can be higher than that of the optimal solution.
Suboptimal designs of the support structure are particularly detrimental because they directly influence capital ex­
penditure (CapEx), balance-of-system (BOS) costs, and the operation and maintenance (O&M) costs. According
to Mone et al. 2015, the substructure and foundation are responsible for 14% of the total offshore wind plant lev­
elized cost of energy (LCOE), and the largest uncertainty in LCOE is attributed to its sensitivity to CapEx, which is
dominated by the support structure.
Capturing these cost relationships with respect to main environmental design drivers (Damiani et al. 2016) is a
formidable challenge that the wind industry faces in the quest for lower LCOE and improved reliability and perfor­
mance. An integrated design of the turbine and support structure may have the potential to significantly lower the
overall system cost. Adding the control system in the loop would further compound the opportunities for loads, ma­
terial mass, and cost reduction. To enable this system-level optimization, physics-based models of all major system
components are required to explicitly capture the trade-offs between their designs.
The National Renewable Energy Laboratory (NREL) developed the wind energy systems engineering toolbox WIS­
DEM (NREL 2015) to address some of the above issues. WISDEM integrates a variety of models for the entire
wind energy system, including turbine and plant equipment, O&M, and cost modeling (Dykes et al. 2011). Although
sophisticated load simulations conducted through aero-hydro-servo-elastic tools can account for all ULS and FLS
design load cases (DLCs), and for an accurate representation of all physical couplings between component dynam­
ics, these simulations are computationally expensive and time intensive. Simplified tools can guide the preliminary
design of components and of the overall system towards a configuration that minimizes the LCOE through multidis­
ciplinary optimization.
Within WISDEM, NREL developed physics-based models for the tower and monopile substructure (TowerSE). They
are relatively simple and mostly based on modal analysis and buckling verification of the main segments of a steel
tubular tower; however, these models do not directly port over to the analysis of a lattice structure. JacketSE was
developed to allow for the analysis of OWTs using jacket-based support structures.

1
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
JacketSE is based on an open-source finite-element analysis (FEA) package (FRAME3DD) that can handle Timo­
shenko beam elements arranged in a beam-frame configuration. A set of structural code checks based on API (2014)
is used to verify members and joints of the substructure, whereas the tower portion of the support structure uses
the same set as in TowerSE. Simplified hydrodynamics also is in common with the TowerSE module. Examples of
results and preliminary verification of the software can be found in Damiani and Song (2013) and Damiani et al.
(2016), but more validation remains necessary.
The most common jacket configuration is the Quattropod®1 , or four-legged lattice, with x-bracing between the legs
forming multiple bays (3–5 for water depths of 20-50 m), with a transition piece (TP) starting at deck-height and
terminating at the tower flange, and with piles secured via special plate-sleeves at the bottom of the legs. As a result,
this is the reference configuration modeled by JacketSE (depicted in Figure 1).
The constraints used on the jacket design follow industry experience but are still sufficiently relaxed to allow for
tighter optimizations. To arrive at more realistic estimates and detailed designs, standard ODs and DT Rs for the
lattice members and piles should be employed. More recently proposed solutions call for three-legged multimem­
ber substructures, as they save mass and construction labor costs. Yet, these configurations require other expedients
to generate the necessary stiffness, for example, by raking the legs as in the inward battered guided structure by
Keystone Engineering Inc. (BVG Associates 2012). These more complex layouts are outside the scope of the work
presented here and will be addressed in future software versions. Nevertheless, a basic three-legged, prepiled, bat­
tered jacket can still be modeled by JacketSE.
JacketSE can be used for either stand-alone support-structure analysis and design or as part of a larger wind turbine
or wind plant study. In stand-alone mode, JacketSE aids the designer in the search for an optimal preliminary con­
figuration of the substructure and tower, and for given environmental loading conditions, turbine dynamic loading,
modal performance targets, and standards design criteria. The optimization criteria (e.g., minimum subcomponent
mass or overall total structural mass) are customizable depending on the user’s needs. JacketSE also allows for
parametric investigations and sensitivity analyses of both external factors and geometric variables that may drive
the characteristics of the structure, thereby illustrating their impact on the mass, stiffness, blade/support clearance,
strength, reliability, and expected costs.
When used for optimization, the tool can size ODs and t’s for piles, legs, and braces; other design variables that
may be optimized are batter angle, pile embedment, tower base and top diameters, wall thickness schedule, and
tapering height. The design parameters (fixed inputs to the tool) include: water depth, deck and hub height, design
wind speed, design wave height and period, and soil characteristics (stratigraphy of undrained shear strength, friction
angles, and specific weight). Loads from the rotor nacelle assembly (RNA) are input to the model either from other
WISDEM modules or directly from the user. The user must also provide acceptable ranges for the design variables,
such as, maximum tower OD; minimum and maximum DT Rs for the various members; and maximum allowed
footprint at the seabed. Additional design criteria and constraints can be employed by the user if desirable.
As part of a system study within WISDEM, JacketSE allows for the full gamut of component investigations to arrive
at optimum LCOE wind turbine and/or power plant layout. For example, JacketSE can produce a design that meets
blade-tower/substructure clearance criteria while also meeting mass or cost targets.
The model has undergone preliminary verification (Damiani and Song 2013; Damiani et al. 2016), but an extensive
campaign against other codes and industry data has yet to be performed. A future version of the model will include
refined fatigue treatment, hydrodynamics loading, and automatic selection of standard dimensions for the various
subcomponents.
This document discusses the model details in Section 2, with an overview of the verification and validation efforts in
Section 3. A simple case study showing the key capabilities of the software is presented in Section 4. A summary of
the development thus far and recommended future research activities are provided in Section 5.

1 Patented jacket configuration by OWEC Tower AS.

2
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Figure 1. Diagram showing the main geometric and structural components Jack­
etSE refers to. Original illustration by Joshua Bauer, NREL (modified here)

3
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
2 Lattice Model Description

JacketSE is based on a modular code framework and primarily consists of the following submodules: geometry-
definition; load calculation; soil-pile-interaction model; FEA model; structural code check; and optimization. A
number of simplifications have been incorporated to allow for rapid analyses of multiple configurations on a per­
sonal computer. As such, complex hydrodynamics and associated variables (e.g., tidal range, marine growth, and
member-to-member hydrodynamic interaction) are ignored, and fatigue assessments are not carried out. Although
these aspects can very well drive the design of certain subcomponents and of the overall structure (Cordle, McCann,
and de Vries 2011; Zwick et al. 2014; Molde, Zwick, and Muskulus 2014), it is believed that the main structural and
mass characteristics should still be captured by the simplified model for the sake of preliminary design assessments
and trade-off studies, and with a level of accuracy limited to those goals. Further details on the code can be found
at https://github.com/WISDEM/JacketSE.git and Damiani and Song (2013). Additional conservatism can be pro­
vided by the choice of drag (cd ) and added mass (cm ) coefficients, the choice of a worst-case loading scenario, and
additional safety factors. For example: for the substructure, the cd and cm values could be doubled with respect to
those recommended by API (2014); the tower drag cd could be set equal to 2 to account for TP drag; and the wave
loads calculated on the main legs could be multiplied by a factor of four to account for hydrodynamics effects on
secondary members of the substructure otherwise not considered. Preliminary comparisons of loads to the peak loads
from dynamic simulations performed with SACS (a commercial package by Bentley for the analysis of offshore
fixed-bottom structures) and FAST v8 (NREL’s aero-elastic tool) of similar substructure configurations led to the
choice of those coefficients. Future studies will employ a refined FLS treatment.
The coupled geometry modules are implemented as components within OpenMDAO1 and they include (see also
Figure 1):
• Piles
• Legs
• Mud-braces
• X-braces
• Top-braces
• TP
• Tower.
A series of inputs are needed to define the entire geometry. Some of those inputs are parameters (i.e., they won’t
change throughout an optimization process), whereas others can be defined as design variables to be optimized. For
example, the height of the deck above mean sea level (MSL), water depth, wave height, and nominal gust speed
are fixed parameters; batter, deck width, tower-waist height, and leg OD are key geometric variables. In Table 1,
examples of the key geometric inputs are given, and more are provided in the subsequent Sections.

2.1 Member Definition and Tube Class

Most members within the substructure and tower can be represented as tubular beam elements. As such, a tube
object class is defined. The tube object is fully defined by (see also Table 2): ODs, t’s, kbuck ’s (buckling parameter
or effective length factor), ltube ’s (tube object unsupported length), and material properties. The latter are defined
via a dedicated object class (mat), which contains E (Young’s modulus), G (shear modulus), ρ (material density),
fy (characteristic yield strength), and calculates the associated ν (Poisson’s ratio). Related to this class are cross-
sectional properties such as surface area, shear area, area moments of inertia, bending moduli, cross-section radius of
1 National Aeronautics and Space Administration software for multidisciplinary analysis and optimization

4
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 1. Examples of JacketSE’s Geometry Inputs

Default
Input(a) Type Description Units
Value
dw parameter water depth – m
zdbot parameter height of TP deck 16 m
zhub parameter hub height - m
number of legs in the substruc­
nlegs parameter 4 –
ture (equal to the number of piles)
nbays parameter number of bays 5 –
flag indicating whether (True) or not (False)
CPf g parameter False –
the legs are considered clamped at the seabed
flag indicating whether the piles are
V Pf g parameter True –
vertical (True) or battered (False)
flag indicating whether (True) or not
PPf g parameter False –
(False) the piles are considered ’plugged’
T Plvl parameter level of automatic build for the TP 5 –
wdr parameter allowance for weldments as a function of member diameter 0.5 m
TP lumped mass, including mass, mass tensor kg,
TPmas parameter –
(Ixx –Izz ) and CM offset from the center of the deck kg m2
mRNA and kg,
parameter mRNA and mass tensor –
Ixx –Izz kg m2
distance vector from the tower-
CMoff parameter – m
top flange to the CM of the RNA
ψ parameter rotor yaw angle about global z 45 deg
two-dimensional batter, i.e., the vertical-to-horizontal
b variable 7 –
ratio of the jacket-leg slope on a 2D projection
dckw variable deck-side length 12 m
Dp variable pile outer diameter 1.5 m
tp variable pile wall thickness 0.035 m
Lp variable pile embedment length 40 m
Dleg variable leg outer diameter 1.5 m
tleg variable leg wall thickness 0.0254 m
Dmbrc variable mud-brace outer diameter 1 m
tmbrc variable mud-brace wall thickness 0.0254 m
Dxbrc variable x-brace outer diameter 0.8 m
txbrc variable x-brace wall thickness 0.0254 m
Dgir variable TP girder OD 1 m
tgir variable TP girder t 0.0254 m
Db variable tower-base OD 5 m
Dt variable tower-top OD 3 m
a Symbols used in this manual might differ from those used in the actual code, which are typeset from alphanumeric,
standard-set characters, but they can be easily referred to the variable names in the current version of JacketSE. See
https://github.com/WISDEM/JacketSE.git.

5
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 2. Tube Class Variables and Parameters Used in the Definition of the Members in JacketSE

Input(a) Type Description Units

OD variable outer diameter m
t variable wall thickness m
ltube parameter tube object unsupported length m
number of elements in the
ndiv parameter –
member under consideration
kbuck parameter buckling parameter or effective length factor –
ρ parameter material density kg m−3
E parameter Young’s modulus N m−2
G parameter shear modulus N m−2
ν parameter Poisson’s ratio m
fy parameter characteristic yield strength N m−2
a Symbols used in this manual might differ from those used in the actual code, which are typeset
from alphanumeric, standard-set characters, but they can be easily referred to the variable names
in the current version of JacketSE. See https://github.com/WISDEM/JacketSE.git.

gyration (rgyr ), and slenderness ratio (KLR) that are used by the FEA solver. The slenderness ratio (used in buckling
verifications) is defined as:
kbuck ltube
KLR = (2.1)
rgyr
A member is defined as a joint-to-joint structural entity (e.g., the member between two adjacent k-joint along the
leg). Each member can be given different material properties (e.g., to use different steels for the legs and x-braces in
the first jacket bay).
Multiple elements along each member may be defined to reduce discontinuities in the FEA mesh element sizes going
from one member to another. For the tower component, multiple elements are used to approximate the taper in OD
and t. In future versions of the code, this approach will help account for tapered members in the substructure as well.
The FEA solver can directly return the internal loads at various stations along a beam element, but this capability is
not currently exploited. For this reason, adopting a number of elements greater than one may be used to evaluate the
stress level in the substructure member with the current version of JacketSE.

2.2 Soil
The soil is described as either cohesive (clay) or cohesionless (sandy), and by a simple stratigraphy table (see Ta­
ble 3), which includes soil level depth (zd ), unit weight (γs ), undrained shear strength (cu ), friction angles (φs ), and an
average angle value representing steel-to-soil friction (δs ). Soil characteristics affect the axial pile capacity and the
stiffness of the soil-pile system described below.
In JacketSE, many assumptions are made to simplify the physics and behavior of the soil and foundation. In the
future, these models will be upgraded with higher fidelity ones. One of the main assumptions concerns the variation
of the Es —soil modulus or modulus of subgrade reaction (N m−2 )—with depth below the seabed. An Es ’s linear
trend, which is largely adopted in JacketSE, is mostly representative of consolidated clays, whereas cohesionless
soils tend to have a parabolic trend. Corrections are employed when sandy soils are considered, as in the Pender’s
method to calculate the pile-head stiffness (described in Section 2.3.2), and in the calculation of the coefficient
of subgrade reaction in Matlock and Reese’s method (see Section 2.3.2). It is further assumed, following Murthy

6
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 3. Typical Stratigraphy Arrangement used by JacketSE

Depth γs cu φs δs
m N m−3 N m−2 deg deg
-3 10,000 60,000 36
-5 10,000 60,000 33
-7 10,000 60,000 26
25
-15 10,000 60,000 37
-30 10,000 60,000 35
-50 10,000 60,000 37.5

(2002), that the Es can be described in terms of ks (coefficient of subgrade reaction):

Es = ks zd (2.2)

A default partial safety factor (PSF) equal to two is set for calculations involving soil properties, but, given the
uncertainty in the geotechnical data and modeling assumptions, larger values are encouraged.

2.3 Piles
The piles can be either vertical or driven through the legs, and therefore slanted (battered), as is normal practice
in the oil and gas (O&G) industry. In the case of a more common offshore wind jacket, the piles are either driven
through a template on the ground prior to lowering and grouting the lattice (prepiled jacket version), or driven
through and grouted to pile sleeves that are built in at the foot of each leg (postpiled version). JacketSE assumes
that the members’ input accounts for any eventual concentric, grouted configuration. Thus, if the piles are driven
through the legs, the input must provide equivalent material properties and wall thickness for the legs’ members to
best model the stiffness of the nonhomogeneous pile-grout-leg cross section. In the case of vertical piles, the physi­
cal connection (see Figure 2) is replaced by an idealized moment connection node between the pile and the leg, but
user’s input should at least account for the extra mass associated with the grouted connection. Similarly, in the case
of a connection via the sleeve plate, JacketSE does not automatically include these subcomponents, and the user
should make provisions for extra steel at the base of the jacket. Future releases will improve these aspects of the
substructure and foundation modeling.
The piles are fully identified by b (the two-dimensional batter, i.e., the vertical-to-horizontal ratio of the jacket-
leg slope on a 2D projection), D p (the pile outer diameter), t p (the pile wall thickness), L p (the pile embedment
length), and PPf g , which is a boolean flag that indicates whether or not they are considered ‘plugged’ (see also
Tables 1, 2, and 4). To determine the embedded length of the pile, JacketSE performs an axial load capacity check,
which normally drives the design of jacket piles. Piles should also be verified for lateral capacity, and in the future
this verification will be included. The piles properties are also used, together with the soil properties, to estimate
equivalent spring constants at the leg bottom.
In general, the stability of the piles at the seabed should also be verified, in which head displacement and rotation
would be checked against allowable values from the standards. This is not done in this version of the code, and will
be included in a future version along with lateral stability and capacity checks.

2.3.1 Axial Capacity

The normal force exchanged at the head of the pile must be reacted by friction along the outer surface of the pile
(also called shaft friction) and by the contact force at the pile tip. In the case of unplugged piles, friction developed

7
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 4. Variables and Parameters Used in the Definition of the Pile Members in JacketSE

Default
Input(a) Type Description Units
Value
Dp variable pile outer diameter 1.5 m
tp variable pile wall thickness 0.035 m
flag indicating whether (True) or not (False)
CPf g parameter False –
the legs are considered clamped at the seabed
flag indicating whether the piles are
V Pf g parameter True –
vertical (True) or battered (False)
flag indicating whether (True) or not
PPf g parameter False –
(False) the piles are considered ’plugged’
zlb parameter elevation of the leg-toe above the seabed 0 m
ndiv,pile parameter number of elements in the pile member 0 –
kbuck parameter buckling parameter or effective length factor 1 –
ρ parameter material density 7805 kg m−3
E parameter Young’s modulus 2.1e11 N m−2
G parameter shear modulus 7.895e10 N m−2
ν parameter Poisson’s ratio 0.3 –
fy parameter characteristic yield strength 345 N m−2
a Symbols used in this manual might differ from those used in the actual code, which are typeset from
alphanumeric, standard-set characters, but they can be easily referred to the variable names in the current
version of JacketSE. See https://github.com/WISDEM/JacketSE.git.

8
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 5. Design Parameters for Cohesionless Soil (API 2014)

Soil Density δs fmax Nq q p,max

deg kPa MPa
Very loose-medium 15 47.8 8 1.9
Loose-dense 20 67 12 2.9
Medium-dense 25 81.3 20 4.8
Dense-very dense 30 95.7 40 9.6
Dense-very dense 35 114.8 50 12

along the inner surface may also be included in the axial capacity of the pile. The ultimate bearing capacity, Qd , is
calculated following API (2014) as:
Qd = Q f + Q p = f As + q p A p (2.3)
where Q f is the skin friction resistance, Q p is the end bearing resistance, f is the skin friction, As is the side surface
area of the pile, q p is the unit end bearing capacity, and A p is the surface area of the pile tip.
Depending on whether the soil is considered cohesive or cohesionless, two different methods are employed to calcu­
late the shaft friction.
For cohesive soils, the procedure makes use of the ψs and αs factors defined as:

αs = min (1, 0.5ψs−0.5 ) if ψs ≤ 1

with ψs = cu /pot (2.4)
αs = min (1, 0.5ψs−0.25 ) if ψs > 1

where pto is the overburden pressure at the depth of interest.

For cohesionless soils, f is calculated as:

f = KAPI pto tan δs (2.5)

where KAPI is the coefficient of lateral earth pressure, which can be taken as 1.0 for plugged and 0.8 for unplugged
piles.
The end bearing capacity is also calculated differently for the two soil types. For clay soils, the unit end bearing is
given by:

q p = 9cu (2.6)

For sandy soils, q p is given by Eq. (2.7) and in any case limited by q p,max given in Table 5:

q p = pto Nq (2.7)

where Nq is a dimensionless bearing capacity factor and recommended values are given in Table 5. The total end
bearing capacity is calculated based on the pile wall annulus, or the gross cross-sectional area if the pile is consid­
ered plugged.

2.3.2 Pile Head Stiffness

To account for the soil-pile interaction in a linearized fashion, two models are available in JacketSE: 1) Matlock
and Reese (1960) and 2) Pender (1993). Through these methods, an equivalent stiffness matrix (or the inverse of a

9
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 6. Coefficient of Subgrade Reaction, ks , for Cohesion-
less Soils as a Function of Friction Angle (from API 2014)

φs (deg) 28 29 30 33 36 38 40 42.5 45
ks (below water table, MN m−3 ) 1.36 3.39 9.33 16.54 25.45 33.08 42.41 49.2 60.23
ks (above water table, MN m−3 ) 0.1 3.39 12.72 25.02 43.26 57.68 75.92 88.22 102.64

flexibility matrix) is devised to be applied at the leg foot:

⎡ ⎤ ⎡ ⎤⎡ ⎤
Fp,x kxx 0 0 0 kxθy 0 u p,x
⎢ Fp,y ⎥ ⎢ 0 kyy 0 kyθx 0 0 ⎥ ⎢u p,y ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ Fp,z ⎥ ⎢ 0 0 kzz 0 0 0 ⎥ ⎢ u p,z ⎥
⎢M ⎥ = ⎢ 0 (2.8)
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ p,x ⎥ ⎢ k θx y 0 kθx θx 0 0 ⎥ ⎢θ ⎥
⎥ ⎢ p,x ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎣M p,y ⎦ ⎣kθy x 0 0 0 kθy θy 0 ⎦ ⎣θ p,y ⎦
M p,z 0 0 0 0 0 kθz θz θ p,z
where Fp,x -Fp,z are the pile head forces, M p,y -M p,z are the pile head moments, u p,x -u p,z are the pile head displace­
ments, θ p,x -θ p,z are the pile head rotations, kxx (=kyy for symmetry)-kzz are the forces at the pile head when unit dis­
placements are imposed at the pile head, kxθy (=kyθx for symmetry) are the forces when unit rotations are imposed at
the pile head, kθx y (=kθy x for symmetry) are the moments when unit displacements are imposed at the pile head, and
kθx θx (=kθy θy for symmetry)-kθz θz are the moments when unit rotations are imposed at the pile head. For reciprocity,
kθx y =kyθx , thus only five terms (kxx ,kzz ,kxθy ,kθy θy ,kθz θz ) are unique in the stiffness matrix of Eq. (2.8).
The leg-to-pile connection is assumed to be of the grouted type, as shown in Figure 2, and not capable of securing
head fixity. This assumption renders the connection and the soil-pile stiffness conservatively less rigid. The two
semiempirical models are useful for design purposes, but a refined analysis is needed in detailed design, for which
p − y (nonlinear spring treatment of soil-pile lateral stiffness), t − z (nonlinear spring treatment of soil-pile axial
stiffness), and Q − z (nonlinear spring treatment of soil-pile end-bearing, axial stiffness) curves should be considered
(API 2014).
Based on a nondimensional analysis, Matlock and Reese (1960) identify a relative stiffness factor Tmr , as shown in
Eq. (2.9), which in turn is used to obtain the flexibility matrix coefficients, as shown in Eq. (2.10):
E p Jxx,p 1/5
Tmr = ks (2.9)
with Es = ks zd
3
fxx = 2.43 E pTJmrxx,p
2
fxθ = fθ x = 1.62 E pTJmrxx,p (2.10)
fθ θ = 1.75 E pTJmrxx,p
where E p is the pile Young’s modulus, Jxx,p is the pile cross-sectional, second area moment of inertia, and zd is the
depth below the seabed. Note that Eq. (2.9) assumes a linear dependence of Es with zd .
For sands, ks may be determined by following the trend recommended by API (2014) as a function of relative density
and friction angle, as shown in Table 6.
For cohesive soils, Bowles (1997) proposes values that are interpolated as a function of soil allowable bearing-
capacity (qa ): ⎧
12000
⎨ 12000 + 100 kPa (qa − 100 kPa) if qa ≤ 200 kPa
⎪
24000
ks = 24000 + 600 (qa − 200 kPa) if 200 kPa < qa ≤ 800 kPa (2.11)
⎪
48000 if qa > 800 kPa
⎩

10
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 (a)

(b)

Figure 2. Diagrams of the grouted connections at the leg footing shown to­
gether with the definitions of zlb and hlb , see text for more details: (a) shows
the slanted (battered) pile configuration; (b) is the vertical pile configuration.

11
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
where units for ks are kN m−3 .
In Pender’s method, a modulus ratio, K p , and an active length, La , of the pile are first defined as:
Ep Ep
Kp = = (2.12)
Es (D p ) ks D p
La = 1.3D p K p0.222 (2.13)
If the pile can be considered ‘long’ (i.e., flexible), then the flexibility coefficients become:

if L ≥ La :
⎧p
K p−0.29
⎧
f = 2.14
⎪
⎪ ⎪
xx
⎪ ⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ Es (D p )D p
K p−0.53
⎪
⎪ ⎪
⎨
⎪
fxθ = fθ x = 3.43 for cohesionless soils (quadratic variation of Es with zd )
⎪
⎪
Es (D p )D2p
⎪
⎪
⎪
⎪ ⎪
⎪
K p−0.77
⎪
⎪ ⎪
⎪
⎪ ⎪
⎩ fθ θ = 12.16
⎪
⎪ ⎪
⎪
Es (D p )D3p
⎪
⎪
⎪
⎨ (2.14)
or
K p−0.333
⎪
⎪ ⎧
⎪
f = 3.2
⎪ ⎪
xx
⎪ ⎪
⎪
Es (D p )D p
⎪
⎪ ⎪
⎪
⎪ ⎪
⎪
K p−0.556
⎪
⎪ ⎪
⎨
⎪
fxθ = fθ x = 5 for cohesive soils (linear variation of Es with zd )
⎪
⎪
Es (D p )D2p
⎪
⎪
⎪
⎪ ⎪
⎪
K p−0.778
⎪
⎪ ⎪
⎪
⎪ ⎪
⎩ fθ θ = 13.6
⎪
⎪ ⎪
⎪
Es (D p )D3p
⎩

If the pile can be considered ‘short’ (i.e., rigid), then:

⎧ if L p ≤ 0.07D p K p :
L −0.333
fxx = 0.7
⎪
⎪
⎪
⎪
⎪
⎪ Es (D p )D p
L −0.88
⎪
⎨ (2.15)
fxθ = fθ x = 0.4
⎪
⎪ Es (D p )D2p
L −1.67
⎪
⎪
⎪
⎩ fθ θ = 0.6 E (D )D3
⎪
⎪
s p p

where L is the ratio of the pile length to its diameter.

For intermediate piles—i.e., those with lengths between the bounds shown in Eqs. (2.14) and (2.15)—, the flexibility
coefficients are approximated by applying a 1.25 factor to those of Eq. (2.14).
Note that both of these treatments assume a linear dependence of Es with zd .
To get the terms kxx , kxθy , and kθy θy of the pile head stiffness matrix to apply at the seafloor, the flexibility matrix is
inverted to give:
kxx kxθy 1 fθ θ − fxθ
= (2.16)
k θy x k θy θy fxx fθ θ − fx2θ − fθ x fxx
An additional, axial spring constant is computed following Pender (1993) for the linear variation of Es with depth,
and given by:
−R L
p R
kzz = 1.8ks L p D0.55 (2.17)
Ep
R= (2.18)
ks L p

12
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Finally, the torsional stiffness is approximated following Randolph (1981) as:
√ 0.5
π 2Gs,c D3p Ĝ p
kθz θz c (2.19)
16 Gs,c

where Lc is the pile critical embedment length for torsion response, Gs,c is the soil shear modulus at torsional critical
depth, and Ĝ p is the equivalent shear modulus used for piles. These quantities are calculated as follows:
32G p Jxx,p
Ĝ p = (2.20)
πD4p
Gs,c = Gs (Lc ) (2.21)
1/3
Dp 2(1 + νs )Ĝ p
Lc c (2.22)
16 ks D p

where νs is the soil Poisson’s ratio.

If the piles are battered, the legs should be defined accounting for the presence of the pile, i.e., they should satisfy
inequality (2.23) (Chakrabarti 2005):
Dleg − 2tleg ≥ D p + 0.09 m (2.23)
where Dleg and tleg denote the jacket leg’s OD and t, respectively. Furthermore, to account for the batter, the stiffness
matrix is rotated by ψrot about the global z, and by φrot about the resulting new x axis, to arrive at:
T
Kg = Cl2g Kl Cl2g (2.24)

where Cl2g is given by:

⎡ ⎤
cos γin cos αbat,3D − sin γin − sin αbat,3D cos γin
Cl2g = ⎣ sin γin cos αbat,3D cos γin − sin αbat,3D sin γin ⎦ (2.25)
⎢ ⎥

sin αbat,3D 0 cos αbat,3D

where γin is the angle described by the horizontal projection of one side of the jacket base and a line connecting
the center of the polygon at the base and one end of that side (= π/2 for four-legged jackets), and αbat,3D is the
three-dimensional batter angle (see Figure 3):
√
2
αbat,3D = arctan (2.26)
b

2.4 Legs
Each leg is made up of nbays +2 members (nbays is the number of bays), which can be individually defined in terms
of OD, t, kbuck (default=1), and material properties (see also Table 7). Normally, one set of dimensions is used for
multiple bays to reduce fabrication complexity, but the code allows for tapered legs.
As mentioned earlier, the presence of internal piles, in the case of battered ones, requires the user to provide ade­
quate equivalent material properties and wall thickness for the leg members. For instance, a simple approach calls
for an equivalent member that has the same mass, axial, and bending stiffness as the original concentric member
arrangement.
The geometry of the legs is completely tied to the overall jacket layout. With a few given values for the overall
geometric variables (e.g., deck-side length, height of the jacket available to the bays, and two-dimensional batter),

13
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Figure 3. Defintions of key variables in the geometry used by JacketSE

Table 7. Variables and Parameters Used in the Definition of the Leg Members in JacketSE

Default
Input(a) Type Description Units
Value
Dleg variable leg outer diameter 1.5 m
tleg variable leg wall thickness 0.0254 m
zlb parameter elevation of the leg-toe above the seabed 0 m
distance from the leg-toe to
hlb parameter 1.5 · Dleg m
the first joint with an x-brace
ndiv,leg parameter number of elements in the leg member 3 –
kbuck parameter buckling parameter or effective length factor 1 –
ρ parameter material density 7805 kg m−3
E parameter Young’s modulus 2.1e11 N m−2
G parameter shear modulus 7.895e10 N m−2
ν parameter Poisson’s ratio 0.3 –
fy parameter characteristic yield strength 345 N m−2
a Symbols used in this manual might differ from those used in the actual code, which are typeset from alphanu­
meric, standard-set characters, but they can be easily referred to the variable names in the current version of
JacketSE. See https://github.com/WISDEM/JacketSE.git.

14
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
the joints and nodes of the legs can be identified via trigonometric functions—see also Eq. (2.32). The joints are the
intersections of leg members with either braces or other members. The nodes are defined as end nodes of the FEA
elements.
First, the bay heights must be calculated. The width of the jacket base (length of one side) at the seabed and the
horizontal distance between the leg-to-brace joints at bay 1 (bottom bay), i.e., bay-1 bottom width, are given by (see
also Figure 3):
wb = dckw − 2Dlt /2(1 + wdr) + 2(dw + zdbot )/b (2.27)
wb,1 = wb − 2 tan αbat,2D (zlb + hlb ) (2.28)
(2.29)
where dckw is the deck-side length, dw is the water depth, Dlt is the leg top OD, wdr is the allowance for weldments
as a function of member diameter, zdbot is the deck underside elevation MSL, αbat,2D is the two-dimensional batter
angle—that corresponds to b—, zlb is the elevation of the leg-toe above the seabed, and hlb is the distance from the
leg-toe to the first joint with an x-brace. The widths of each further bay can be recursively calculated as in Eq. (2.30):
tan (π/2 − β2D − αbat,2D )
wb,i = wb,i−1 1 − 2 tan (αbat,2D ) (2.30)
1 + tan (αbat,2D ) tan (π/2 − β2D − αbat,2D )
It can also be proved that the height of each bay (hb,i ) can be written as:
tan (π/2 − β2D − αbat,2D )
hb,i = wb,i (2.31)
1 + tan (αbat,2D ) tan (π/2 − β2D − αbat,2D )

where β2D is the brace-to-leg angle as measured on a vertical projection, which is an unknown, but is considered
fixed throughout the bays for ease of manufacturability. By making use of the above definitions and simultaneously
solving the system of equations in Eq. (2.32), the β2D angles can be calculated:
⎧
⎨ h jckt = zdbot + dw − zlb
⎪
hbys = h jckt − hlb (2.32)
⎪ nbays
hbys = h
⎩
∑1 b,i

In Eq. (2.32), h jckt is the height of the jacket available to the bays, and hbys is the array containing the bay heights.
The actual brace-to-leg angle β3D can be calculated from the extended Pythagorean theorem:
⎧
⎪
⎪
⎪ w2b,2 − AB2 + lb2,1 − 2 ∗ AB ∗ lb,1 cos (β3D ) = 0 with
h
⎪
AB = cos (αb,1 )
⎪
⎪
⎪
bat,3D
⎪
⎪
⎨ c
2 2 (2.33)
⎪ lb,1 = h2b,1 + wb,1 − hb,1 tan (αbat,2D ) + hb,1 ∗ tan (αbat,2D ) ∗ tan (γin )
⎪
⎪
wb,2 = wb,1 − 2hb,1 tan (αbat,2D )
⎪
⎪
⎪
⎪
tan (π/2−β2D −αbat,2D )
⎪
⎪
⎩ h =w
b,1 b,1 1+tan (αbat,2D)∗tan (π/2−β −α
2D )
bat,2D

Note that Eq. (2.33) is numerically solved for β3D , and AB is the distance between the first two leg joints at the
bottom bay, wb,2 is the second bay width, lb,1 is the length of the bottom x-brace, and hb,1 is the height of the bottom
bay.
Once hbys has been determined, the joints’ coordinates and the internal nodes for the legs and the braces can be
calculated. For the first leg, the joint coordinates are given by:
⎧ z −z
⎪
⎪ xleg i = −wb /2 + leg b lb
⎪
⎨ yleg i = xleg i tan γin
(2.34)
⎪
⎪
⎪ zleg i = zlb + ∑i1 hb,i
⎩
with i = 1..nbays

15
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 8. Variables and Parameters Used in the Definition of the X-brace Members in JacketSE

Default
Input(a) Type Description Units
Value
Dxbrc variable x-brace outer diameter 0.8 m
txbrc variable x-brace wall thickness 0.0254 m
ndiv,xbrc parameter number of elements in the x-brace member 1 –
kbuck parameter buckling parameter or effective length factor 0.8 –
ρ parameter material density 7805 kg m−3
E parameter Young’s modulus 2.1e11 N m−2
G parameter shear modulus 7.895e10 N m−2
ν parameter Poisson’s ratio 0.3 –
fy parameter characteristic yield strength 345 N m−2
a Symbols used in this manual might differ from those used in the actual code, which are typeset from
alphanumeric, standard-set characters, but they can be easily referred to the variable names in the current
version of JacketSE. See https://github.com/WISDEM/JacketSE.git.

The other legs’ coordinates are calculated starting from the above ones and rotating them about global z by the γin
angle and repeating the operation for nlegs − 1 times, where nlegs is the number of legs in the substructure (equal to
the number of piles). Internal nodes are calculated by subdividing the obtained joints’ coordinates proportionally to
the user-supplied number of elements (ndiv,leg ). The material properties are also replicated for all the elements, so
that tube objects can be produced for each of them.

2.5 X-Braces
With the leg joints defined, the x-braces are determined starting from the joints of two adjacent legs, and are assigned
their respective inputs, namely ODs, t’s, kbuck (default=0.8), and material properties (see also Table 8).
X-joints are defined at the three-dimensional intersections between the brace pairs, and calculated as in Eq. (2.35)
(see also Figure 4): ⎧
⎪ ABc = A1 A2 × B1 B2
⎪
⎪
⎨ nA
A0 = A 1 + A1 A2 (2.35)
⎪
⎪
⎪ AB c
⎩ with n = B B × B A · AB
A 1 2 1 1 c

where A1 and A2 are the two joints at the ends of one of the two braces in the pair, B1 and B2 are the analog joints for
the other brace, ABc is the vector resulting from the cross product of A1 A2 and B1 B2 , nA is an auxiliary quantity, and
A0 is the set of new coordinates for the x-joint.
Just like for the legs, if the input requests multiple elements (ndiv,xbrc > 1), then new nodes are created by proportion­
ally subdividing the coordinates of the joints in the member defined between the brace-leg joint and the x-joint.
The braces must satisfy a few constraints that are derived from engineering experience (Chakrabarti 1987) and

16
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Figure 4. Schematics of the intersection between two x-braces

shown in Eq. (2.36): ⎧

⎪
⎪ Abrc ≥ 0.1Aleg
⎪
⎨ Dxbrc ≥ 0.3Dleg
⎪
⎪
⎪
Dxbrc
txbrc ≥ 31 (2.36)
⎪ D 250.
txbrc ≤ hydc = (3.28dw )0.33
⎪
⎪ xbrc
⎪
⎪
⎪
KLRbrc ≤ KLRmax = 70 − 80
⎩

where Abrc the brace cross-sectional area, Aleg is the leg cross-sectional area, KLRbrc is the brace KLR, and hydc is
the hydrostatic constant. The first inequality of Eq. (2.36) ensures a rigid truss behavior of the bay and therefore an
adequate shear transfer from leg to leg; the second constraint improves the capacity of the joints; the third constraint
ensures the manufacturability of the brace and ensures positive buoyancy; the fourth inequality virtually removes
hydrostatic problems; and the last constraint in Eq. (2.36) improves the axial capacity of the brace and renders the
material utilization more efficient.

2.6 Mud-Brace
The mud-brace is a horizontal brace placed near the pile-to-leg joint. The mud-brace relieves the stress concentration
at the pile head, and further increases the torsional stiffness of the substructure. The mud-brace is fully defined by
its OD, t, kbuck (default=0.8), and material properties (see also Table 9). The position of the mud-brace changes
depending on the pile configuration (see also Figure 2). If the piles are of the prepiled, vertical type, then the mud-
brace joint with the leg is located at the assumed pile-to-leg joint. In the case of battered, in-leg piles, the mud-brace
is joined to the leg at the bottom x-brace joints.
Like other members, the mud-brace may be subdivided into multiple elements via the input ndiv,mud . The same
constraints as in Eq. (2.36) apply to the mud-brace.

2.7 Top Brace

A set of horizontal braces at the top of the main lattice structure can be enabled as an option. The top braces connect
the leg top joints together, and, similarly to the mud-braces, they relieve the stress concentration at the connection

17
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 9. Variables and Parameters Used in the Definition of the Mud-brace Members in JacketSE

Default
Input(a) Type Description Units
Value
Dmbrc variable mud-brace outer diameter 1 m
tmbrc variable mud-brace wall thickness 0.0254 m
number of elements in
ndiv,mud parameter 1 –
the mud-brace member
kbuck parameter buckling parameter or effective length factor 0.8 –
ρ parameter material density 7805 kg m−3
E parameter Young’s modulus 2.1e11 N m−2
G parameter shear modulus 7.895e10 N m−2
ν parameter Poisson’s ratio 0.3 –
fy parameter characteristic yield strength 345 N m−2
a Symbols used in this manual might differ from those used in the actual code, which are typeset from
alphanumeric, standard-set characters, but they can be easily referred to the variable names in the current
version of JacketSE. See https://github.com/WISDEM/JacketSE.git.

between the legs and TP stumps (defined in Section 2.8). The top bay’s x-brace’s OD and t are the default values
for the top brace’s OD and t, but the input can obviously be changed as needed. Completing the input set are kbuck
(default value =0.8) and material properties (see also Table 9). Note, the top-brace may be omitted, with its role
played by the TP girder.

Table 10. Variables and Parameters Used in the Definition of the Top-brace Members in JacketSE

Default
Input(a) Type Description Units
Value
Dtbrc variable top-brace outer diameter Dxbrc m
ttbrc variable top-brace wall thickness txbrc m
0 (not
ndiv,tbrc parameter number of elements in the top-brace member –
used)
kbuck parameter buckling parameter or effective length factor 0.8 –
ρ parameter material density 7805 kg m−3
E parameter Young’s modulus 2.1e11 N m−2
G parameter shear modulus 7.895e10 N m−2
ν parameter Poisson’s ratio 0.3 –
fy parameter characteristic yield strength 345 N m−2
a Symbols used in this manual might differ from those used in the actual code, which are typeset from
alphanumeric, standard-set characters, but they can be easily referred to the variable names in the current
version of JacketSE. See https://github.com/WISDEM/JacketSE.git.

18
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Figure 5. Diagram showing the structural simplification
adopted by JacketSE to represent the TP with a frame of beams

2.8 TP
The TP structure is modeled by considering a frame of beams in lieu of the actual structure, as shown in Figure 5. A
perimeter girder with two cross-braces make up the deck. A central beam (stem) represents the shell connecting to
the base of the tower. Four additional beams (struts) support the shell from the corners.
The various subcomponents, or tubular beams, are described by material properties and ODs, t’s, and kbuck ’s (default
value=0.8). The central stem needs also to be defined in terms of its length hstem , as its top node will connect to the
tower. The stem can also accommodate a concentrated mass at the connection with the cross-braces. Additionally,
vertical beams (stumps) can be used to represent spacers from the leg top-joints to the girders.
Key inputs for the TP are given in Table 11. Note that each subcomponent can be assigned different material proper­
ties, besides different values for the respective kbuck ’s.
JacketSE provides a simpler way to assign the geometry of the TP submembers. There are five levels of automatic
generation (T Plvl ) that can be selected:
1. Dstrt = Dstmp = Dleg , tstrt = tstmp = tleg
2. as 1. plus Dgir = DT Pbrc = Dxbrc , tgir = tT Pbrc = txbrc
3. Dgir = DT Pbrc = Dxbrc , tgir = tT Pbrc = txbrc only
4. Dgir = DT Pbrc , tgir = tT Pbrc only
5. as 1. + as 4.
For any of the above options, the elements of the arrays Dstem and tstem are set to Db and fstemtb , respectively, with
fstem given as input (=1.5 by default), and tb being the tower-base t.

2.9 Tower
The tower is assumed to be made up of two main sections (see also Figure 6): a bottom segment with a constant
cross section, and a top segment with both OD and t tapered. The main input variables are given in Table 12.

19
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 11. Variables and Parameters Used in the Definition of the TP Members in JacketSE

Default
Input(a) Type Description Units
Value
Dstmp variable TP stump OD Dleg m
tstmp variable TP stump t tleg m
hstmp variable TP stump length 0 m
Dgir variable TP girder OD Dxbrc m
tgir variable TP girder t txbrc m
DT Pbrc variable TP cross-brace OD Dxbrc m
tT Pbrc variable TP cross-brace t txbrc m
Dstrt variable TP strut OD Dleg m
tstrt variable TP strut t tleg m
Dstem variable array of TP central shell ODs (one per shell member) Db m
tstem variable TP central shell t’s (one per shell member) fstemtb m
hstem variable array of TP central shell lengths (one per shell member) tleg m
T Plvl parameter level of automatic build for the TP 5 –
TP lumped mass, including mass, mass tensor kg,
TPmas parameter -
(Ixx –Izz ) and CM offset from the center of the deck kg m2
nstem parameter number of TP shell members 3 –
fstem parameter ratio between the TP central-shell wall thickness and tb 1.5 –
0 (not
ndiv,stmp parameter number of elements in each of the TP stumps –
used)
ndiv,strt parameter number of elements in each of the TP struts 1 –
ndiv,gir parameter number of elements in each of the TP girder members 1 –
ndiv,T Pbrc parameter number of elements in each of the TP cross-brace members 1 –
kbuck parameter buckling parameter or effective length factor 0.8 –
ρ (b) parameter material density 7805 kg m−3
E (b) parameter Young’s modulus 2.1e11 N m−2
G(b) parameter shear modulus 7.895e10 N m−2
ν (b) parameter Poisson’s ratio 0.3 –
fy (b) parameter characteristic yield strength 345 N m−2
a Symbols used in this manual might differ from those used in the actual code, which are typeset from alphanumeric,
standard-set characters, but they can be easily referred to the variable names in the current version of JacketSE. See
https://github.com/WISDEM/JacketSE.git.
b Material properties are assigned for each subcomponent (i.e., stump, strut, brace, girder, and stem).

Additionally, the input should contain the effective length factor for buckling verification kbuck (default value=1),
the unsupported length htwrb (largest distance between consecutive flanges, e.g., 30 m), the ztho f f (distance from the
tower-top flange to the hub center along z) and the material properties of the shell. The material density should be
augmented for secondary steel and flanges, which are not directly accounted for, unless the discrete span input is
selected. If the input Δzmx (maximum FEA element length for the tower elements) is supplied, then ndiv,twr may be
overridden to ensure that every tower element length is below Δzmx . A minimum of two elements per segment is
allowed.

20
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 12. Variables and Parameters Used in the Definition of the Tower in JacketSE

Default
Input(a) Type Description Units
Value
Db variable tower-base OD 5 m
DT Rb variable tower-base DT R 120 –
Dt variable tower-top OD 3 m
DT Rt variable tower-top DT R 120 –
h2 f variable fraction of tower length at constant cross section 0.25 –
htwr parameter tower length – m
zhub parameter hub height above MSL – m
ztb parameter elevation MSL of tower-base – m
ndiv,twr parameter number of elements in each of the two tower segments [4, 10] –
distance vector from the tower-
CMoff parameter – m
top flange to the CM of the RNA
δthoff parameter distance vector from the tower-top flange to the hub center – m
ψ parameter rotor yaw angle about global z 45 °
tower buckling effective length,
htwrb parameter 30 m
shortest distance between flanges
flag selecting how to treat the rigid con­
rig f lg parameter False –
nection at tower-top with the RNA
Δzmx parameter maximum FEA element length for the tower elements – m
kbuck parameter buckling parameter or effective length factor 1 –
ρ (b) parameter material density 7805 kg m−3
E (b) parameter Young’s modulus 2.1e11 N m−2
G(b) parameter shear modulus 7.895e10 N m−2
ν (b) parameter Poisson’s ratio 0.3 –
fy (b) parameter characteristic yield strength 345 N m−2
a Symbols used in this manual might differ from those used in the actual code, which are typeset from alphanumeric,
standard-set characters, but they can be easily referred to the variable names in the current version of JacketSE. See
https://github.com/WISDEM/JacketSE.git.
b Material properties can be either assigned for the entire tower or specifically for each member along the length of the

tower.

If both htwr and zhub are provided, JacketSE will verify that the two values are compatible given ztb and ztho f f , oth­
erwise an error message is issued and the program quits. If ztho f f is not given or left at 0 m, it will be initialized at
zcmo f f (distance from the tower-top flange to the RNA CM along z). Note that ztb must coincide with the top node of
the TP.
The tower is assumed rigidly connected to the RNA. This is represented either by a very stiff, massless FEA element
(this is a legacy of older code and that will be discontinued in future versions), or it can be handled mathematically
(preferred). In the latter case, the RNA inertia tensor elements (mass and six second moments referred to the CM
(i.e., Ixx ,Iyy ,Izz ,Ixy ,Iyz , and Ixz ) and the CMoff (distance vector from the tower-top flange to the CM of the RNA) are
directly input into the mass matrix of the FEA solver. The input switch rig f lg is a boolean input that instructs the
code on how to treat the RNA connection. The presence of a ψ �= 0 is also accounted for, by rotating the RNA inertia
tensor accordingly.

21
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Figure 6. The main design variables and parameters for the tower model adopted by JacketSE

In place of Db , DT Rb , Dt , DT Rt , htwr , and h2 f , the tower may be defined by a set of discrete stations along the
span, which are identified by triplets containing z coordinate values, ODs, and t’s at the corresponding locations.
The coordinates input in this fashion will be directly used as FEA nodes, and no further mesh refinement is needed.
Concentrated, lumped masses can also be input in this case.
The resulting FEA elements are constant cross-section beam elements, thus, using an adequate number of elements
ndiv,twr (or a proper span discretization) is recommended to accurately capture the taper of the tower. Tapered ele­
ments will be made available in a future release of the code.
The ψ is used to properly rotate the RNA inertia tensor and pass that information to the FEA solver. Additionally,
δthoff (distance vector from the tower-top flange to the hub center) is also rotated by ψ so that RNA loads are placed
correctly.

2.10 Loads
The loads that are considered by JacketSE can be categorized into permanent actions (dead loads) and live loads.
Dead loads are the gravitational loads associated with the self-weight of the structure and secondary steel, such as
tower internals, deck appurtenances (such as transformers and cranes), anodes, platforms, boat landings, and so on.
Because of the vibrational and deflection characteristics of the OWT, even these so-called dead loads have very
important dynamic effects—for example, the P − Δ effect associated with the displacement of the RNA CM (center
of mass). Because JacketSE does not resolve these secondary and nonlinear effects, it is recommended that PSF and
load estimation be selected conservatively. Furthermore, secondary steel can only be accounted for by concentrated
masses at tower nodes and at the center of the TP deck. Applying a fictitious, increased steel density for the jacket
and the tower may help simulate the effect of hardware, cathodic protection, and coatings.
The live loads considered by JacketSE include: aerodynamic loads from the RNA, i.e., forces Faero and moments

22
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Figure 7. Main reference system used in JacketSE and principal sources of loading and their
general areas of application. Original illustration by Joshua Bauer, NREL (modified here)

Maero originating at the rotor (augmented for dynamic amplification and vibrational effects); drag loads from direct
action of the wind on the support structure; and hydrodynamic loads.
At this point in the development, other important sources of loads—e.g., installation loads, accidental loads, vortex-
induced vibrations, ice, and seismic loadsDamiani (2016)—are ignored, and a number of simplifications in the
calculation of live loads are employed, as discussed below.
In Figure 7, the approximate location of the loads’ application points from the RNA and hydrodynamics is shown
together with the main coordinate system adopted by JacketSE.
The primary simplification in JacketSE is the treatment of all loads as pseudo-static. This approximation reduces
computational time and resources, whereas an accurate determination of dynamic load components demands sophis­
ticated computer-aided engineering (CAE) tools and coupled numerical simulations. Thus, users must exercise care
in selecting loads and PSFs to compensate for the lack of a fully dynamic treatment. Furthermore, FLS verification
is not directly performed, which is a capability that will be implemented in future versions of JacketSE. In general,
fatigue loading tends to dominate the design of the flanges and welds for the tower, and the joints of the jacket,
whereas the main shells are driven by modal and buckling-strength requirements.
The primary loading source for the tower comes from the aerodynamic loads induced by the rotor. The substruc­
ture must resist the combination of RNA loads and hydrodynamics loads, with the latter becoming more and more
important as water depth and wave heights increase.

23
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Tower shear and bending moments from the direct wind action on the tower, as well as wave and current loads on
the jacket legs, are approximated following basic fluid-dynamics principles. Through consultations of the standards,
wind shear values (IEC 2009) and δ f s—or gust factors from DIN (2005), European Committee for Standardisation
(2010), and SEI (2005, 2010)—may be obtained to calculate and integrate the drag force along the tower span.
The default set-up for JacketSE allows for two main DLCs, usually taken as:
1. An operational DLC, similar to DLC 1.6 from IEC (2009)
2. A parked DLC, similar to DLC 6.1 from IEC (2009).
For these two cases, loads and environmental conditions can be input independently. Additional load cases can be
coded into the program relatively easily.
Some of the additional parameters needed to define the loads in JacketSE are given in Table 13, with more details
presented in the following Sections. More details about the load calculations are given in the Sections 2.10.1–
2.10.4.

2.10.1 RNA Loads

From a quasi-steady-state point of view, the RNA loads reduce to three forces and three moments along the main
coordinate axes—see Damiani (2016) for more details on load sources. These loads are generated by the rotor
aerodynamics in a turbulent and sheared inflow under tower interference effects and structural imbalance. In addition
to the normal operating loads, transients such as shutdowns, and blade/rotor/yaw faults can give rise to important
ULS loads for the support structure and should not be underestimated. Other important load contributions derive
from the gyroscopic effects when the turbine yaws with the rotor spinning.
The thrust is the biggest force responsible for the bending moment distribution along the tower and loads on the sub­
structure. For a downwind turbine, the additional effect of the gravitational load caused by a downwind offset of the
RNA CM from the tower centerline may significantly increase the tower utilization. In an upwind turbine, the RNA
mass contribution to the bending moment is minimal and any upwind offset of the RNA CM can be conservatively
ignored, as the P − Δ effect would tend to reduce that contribution anyway.
From a FLS standpoint, the aerodynamic loads associated with the RNA tend to dominate the design, with the
exception of cases in deep water sites where hydrodynamics excitation and low damping situations can also be
important.
Various methods exist to estimate RNA loads, and they are not discussed in this manual—for more details, see
Damiani (2016). JacketSE assumes that the input RNA loads account for dynamic amplification, such as those that
may be derived from aero-elastic analyses.

2.10.2 Aerodynamic Loading

Direct wind loading on the tower and jacket is caused by aerodynamic drag on the structural elements. The wind
profile can be taken as in Eq. (2.37), where α (wind power law exponent) is given by the standards—e.g., IEC
(2009)—and as input to JacketSE.
z α
|U(z)| = |Uhub | (2.37)
zhub
where U(z) is the wind velocity vector as a function of z, the altitude above MSL, and Uhub is the wind velocity at
hub height (i.e., zhub ).
For a typical tubular member of the tower or jacket, Eq. (2.38) can be used to calculate the pseudo-static loads on the
tower due to wind action on its cylindrical or conical segments.

fa = 0.5ρa πDsh cd G f U|U| (2.38)

24
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 13. Additional Parameters Used in the Definition of the Loads in JacketSE

Default
Input(a) Type Description Units
Value
Faero variable force vector originating at the rotor – N
Maero variable moment vector originating at the rotor – Nm
distance vector from the tower-
CMoff parameter – m
top flange to the CM of the RNA
δthoff parameter distance vector from the tower-top flange to the hub center – m
ψ parameter rotor yaw angle about global z 45 °
Uhub parameter wind velocity at hub height – m s−1
α parameter wind power law exponent 0.2 –
maximum (e.g., 50 yr) wave height(s)
hw parameter – m
under operation and parked conditions
maximum (e.g., 50 yr) wave nominal pe­
Tw parameter – s
riod under operation and parked conditions
uc parameter current velocity – m s−1
ρa parameter air density 1.225 kg m−3
ρw parameter sea water density 1025 kg m−3
Gf parameter gust factor – –
cd,at parameter air drag coefficient tower – –
cd,a j parameter air drag coefficient jacket – –
cm parameter added mass coefficient – –
cd,w j parameter water drag coefficient jacket – –
γf parameter generic load PSF 1.35 –
γfg parameter gravitational load PSF 1.1 –
γfa parameter aerodynamic load PSF 1.35 –
γm parameter material PSF 1. –
γn parameter consequence of failure PSF 1. –
a Symbols used in this manual might differ from those used in the actual code, which are typeset from alphanumeric,
standard-set characters, but they can be easily referred to the variable names in the current version of JacketSE. See
https://github.com/WISDEM/JacketSE.git.

where fa is the force per unit length caused by wind aerodynamic drag, ρa is the air density, Dsh is the OD of the
tower or jacket leg member, cd is the air drag coefficient (e.g., c 0.6 − 0.7 for tower segments), and G f is the gust
factor; all are potentially functions of the height above MSL.
JacketSE integrates Eq. (2.38) along the span of the tower, adds the thrust and moments from the RNA, and cal­
culates the total shears and bending moments along the tower segments. Normally, the thrust of the rotor under
operational DLCs is by far the dominant contributor to the bending stresses in the tower shell, and for FLS verifica­
tion the direct wind drag is less important; but parked cases under ULS conditions may give rise to important drag
loads that cannot be underestimated especially for high hub heights and smaller rotors.
The gust factor in Eq. (2.38) accounts for the effect on wind actions from the nonsimultaneous occurrence of peak
wind pressures on the structure surface together with the effect of the vibrations of the structure caused by turbu­
lence. Various methods to calculate G f are offered by the standards—e.g., DIN (2005), European Committee for

25
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Standardisation (2010), and SEI (2005, 2010)—, and more recent treatments with focus on wind turbine towers can
be found in Burton et al. (2005) and Murtagh, Basu, and Broderick (2007). Per SEI (2005), structures are defined as
dynamically sensitive (or flexible) if their first natural frequency is f0 < 1 Hz—more detailed criteria can be found in
European Committee for Standardisation (2010) Part 1-4. When this criterion is met, G f can be significantly larger
than one.
JacketSE requires the cd input value to account for G f . Furthermore, two values can be input: one for the jacket
(cd,a j ) and one for the tower (cd,at ). The aerodynamic drag is, in fact, computed also on the jacket leg members
above MSL following Eq. (2.38), where Dsh is replaced by Dleg . Because of the leg three-dimensional orientation,
the component of U normal to the leg is considered in Eq. (2.38) and the resulting force is further decomposed into
a horizontal and a vertical force. Because no aerodynamic force is computed for the cross members and the TP, the
cd,at and cd,a j should be augmented to account for the load contributions from these members.

2.10.3 Hydrodynamic Loading

The effects of wave and current kinematics on the loading of a jacket can be calculated via the Morison equation for
slender members:
fw = 0.25ρw πD2leg cm U
˙ w + 0.5ρw πDleg cd,w j Uw |Uw | (2.39)
where fw is the force per unit length as a result of wave and current kinematics, ρw is the sea water density, Uw =uw +uc
is the vectorial sum of the wave and current velocity (vector sum), and cd,w j and cm are the water drag coefficient
jacket and the added mass coefficient, respectively.
The coefficients cd,w j and cm should account for the presence of marine growth, and depend on Reynolds and
Keulegan-Carpenter numbers, thus on the specific DLC under consideration—see also ISO (2014).
JacketSE assumes a uniform (unsheared) current velocity aligned with the wave direction. The wave particle-
kinematic velocity is calculated via Airy theory—see Eq. (2.40), ISO (2014). This theory does not account for
any stretching, therefore the loads calculated through Eqs. (2.39)-(2.40) may be underestimated.
cosh κw (dw + zw )
uw = hw ωw
sinh κw dw (2.40)
sinh κw (dw + zw )
u̇w = hw ωw2 cos (κw x − ωwtt )
sinh κw dw

In Equation (2.40), hw is the wave height, (peak-to-peak distance), zw is the distance from the sea surface, positive
upwards, κw is the wave number, ωw is the wave frequency, x is the wave propagation direction, and tt is the time
variable. Two further simplifications are adopted in JacketSE. First, the hydrodynamic loads are solely calculated
on the main leg members, ignoring the direct contribution to the wave and current forces from the cross members.
Second, the maximum wave speed and acceleration are conservatively considered as acting simultaneously, thus
ignoring any phase effect and essentially dropping the time dependency, while also not including any structural
motion.
To account for additional loads on the cross-braces and the absence of wave stretching, applying larger-than-normal
hydrodynamic coefficients is recommended. The conservatism associated with the static treatment of the wave
kinematics helps further mitigate the consequences of a cursory treatment of the hydrodynamics.
In the literature, corrections to the Morison equation have been proposed for diffraction effects—e.g., the MacCamy-
Fuchs correction, Chakrabarti (1987)—, and for wave nonlinearities. In addition, the probability of breaking wave
occurrence (normally when hw /dw > 0.78) and associated slamming loads (DNV 2013) on the support structures
(SSts) should be verified, because these loads are not directly included in JacketSE.
In real structures, other concentrated loads may derive from the presence of boat landings and J-tubes, which may
attract more wave loads not accounted by JacketSE. These loads are normally less impactful, yet the welded connec­
tions to the principal steel may be a source of crack nucleation and corrosion.

26
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Because all of these aspects could be important, we again recommend using conservative estimates for cd,w j and cm
and PSFs. Future releases will refine the hydrodynamics loading model to overcome the limitations of this simplified
treatment.

2.10.4 Loads Integration

Analytically, the various contributions to the tower internal loads could be accounted for as in Eqs. (2.41)-(2.44).
The normal (axial, along the z axis) force can be written as:
L
Nd (z) = γ f FzRNA − γ f g mRNA g − γ f g ρAgdz (2.41)
z

where Nd is the design (factored) normal load at the tower station of interest, FzRNA is the aerodynamic force from
the RNA along the z axis, γ f is the generic load PSF, mRNA is the RNA mass, g is the gravity acceleration, γ f g
is the gravitational load PSF, ρ is the material density, L is the generic member length, A is the member cross-
sectional area, and z is the z coordinate along the tower span. The last two of the three terms on the right-hand side
of Eq. (2.41) are automatically calculated by the FEA solver (see Section 2.11), whereas the first term is part of the
input to JacketSE.
The shear components along x and y (reference system is that of Figure 7) can be calculated as:
L
Tx (z) = γ f FxRNA + γ f a ˆ
fa · idζ (2.42a)
z
L
Ty (z) = γ f FyRNA + γ f a ˆ
fa · jdζ (2.42b)
z

where FxRNA is the force from the RNA along the x-axis, FyRNA is the force from the RNA along the y-axis, fa is given
in Eq. (2.38), γ f a is the aerodynamic load PSF, î is the unit vector along the x-axis, jˆ is the unit vector along the y-
axis, and ζ is the dummy coordinate along the z-axis. Whereas FxRNA and FyRNA are inputs to JacketSE, the integrals
in Eq. (2.42) are replaced by a two-step approach. First, a discrete integration of the distributed wind pressure over
the tower elements is performed and equivalent shear loads are applied at the element nodes. Second, during uti­
lization calculation, these forces are numerically integrated via simple quadrature to obtain the actual internal shear
loads.
The bending moment components can be written as follows:
L
Mx (z) = γ f MxRNA − FyRNA (zRNA − z) − γ f a fa · jˆζ + γ f g ρAg(y(ζ ) − y(z)) dζ (2.43a)
z
L
My (z) = γ f MyRNA + FxRNA (zRNA − z) + γ f a fa · îζ + γ f g ρAg(x(ζ ) − x(z)) dζ (2.43b)
z

where Mx is the component of the bending moment along the x-axis at the station of interest, My is the component
of the bending moment load along the y-axis at the station of interest, MxRNA is the RNA aerodynamic moment along
the x-axis, MyRNA is the RNA aerodynamic moment along the y-axis, ρAg is the distributed mass in kg m−1 , and
(x(ζ ) − x(z)) and (y(ζ ) − y(z)) account for the deflection of the tower axis along x and y from the undeflected po­
sition. JacketSE, however, does not account for second order P − Δ effects, so the second terms within the integrals
are ignored. The remainder terms within the integrals are again replaced by a simple numerical quadrature of the dis­
tributed pressure over the span of the tower, and bending moment values are applied at the element nodes. Potentially
different PSFs for the various loading components can be employed following standards such as Germanischer Lloyd
(2005) and input into JacketSE. Finally, the torque about the z-axis is given by:
Mz (z) = γ f MzRNA − FxRNA (y(zRNA ) − y(z)) + FyRNA (x(zRNA ) − x(z)) (2.44)
where Mz is the torsion moment load along the z-axis at the station of interest. Note no further contribution to torque
along the tower axis is considered.

27
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
The shear and moment components can then be combined to arrive at characteristic design values:

Td (z) = Tx (z)2 + Ty (z)2 (2.45a)

Md (z) = Mx (z)2 + My (z)2 (2.45b)

Note that for the tower, the internal loads are calculated externally to the FEA solver (pyFrame3DD) while sep­
arating contributions from gravity loads and aerodynamics, but these loads could just as well be calculated by
pyFrame3DD.
For the jacket leg (and any exposed portion of the piling), wind and hydrodynamic loads are computed numerically
by integrating Eq. (2.39) below the water surface and Eq. (2.38) above the water line, and then by applying equiva­
lent concentrated forces at the leg nodes, as per Eq. (2.46):
L
Tx (zi ) = γ f w ˆ
fw · idζ (2.46a)
zi
L
Ty (zi ) = γ f a ˆ
fa · jdζ (2.46b)
zi

where γ f w is the hydrodynamic load PSF.

The internal loads are calculated by pyFrame3DD and used for utilization verification.
Rotation matrices are employed to account for batter in the legs and rotor yaw alignment. The DLCs discussed in
Section 2.10 consider wind and wave along the main diagonal of a four-legged jacket. In that case, the shear from air
and water drag on upstream and downstream legs generate both shear and tensile (compression) forces that are given
by the decomposition of wave forces assumed normal to the legs’ axes. For the sake of simplicity, the air drag is also
considered normal to the leg axis, which is more conservative, as most of the load goes into bending. For the other
two legs, the load is considered horizontal (i.e., normal to the vertical plane that contains them).
The internal loads are then used to derive stresses and calculate the utilization in both tower and jacket members and
joints as discussed below.
Although the above treatment is sufficient for conceptual and preliminary assessments, only rigorous loads and
FEA analyses can fully support a more detailed design. In particular, fully coupled loads analyses via aero-hydro­
servo-elastic (AHSE) simulations are the only way to capture important interactions and vibrational dynamics and to
perform FLS and ULS verifications for the certification of OWTs’ support structures.
The FLS verification will be implemented in future releases of the software.

2.11 FEA Solver

The FEA solver at the core of JacketSE is based on a modified version of FRAME3DD. FRAME3DD is an open-
source software for static and dynamic structural analysis of two- and three-dimensional beam frames and trusses
with elastic and geometric stiffness. It computes the static deflections, reactions, internal element forces, natural
frequencies, mode shapes, and modal participation factors of two- and three-dimensional elastic structures us­
ing direct stiffness and mass assembly. The theory is based on a Timoshenko beam element—see Timoshenko
(1970) and Panzer et al. (2009)—and the source code is in C. A python module called pyframe3DD (see https:
//github.com/WISDEM/pyFrame3DD) was developed to run FRAME3DD without the need for input and output
files. pyFrame3DD runs seamlessly with the JacketSE, and optional inputs for FRAME3DD are handled via Jack­
etSE’s inputs (see Table 14).
Furthermore, pyFrame3DD accepts arbitrary stiffness values at constraint nodes (rather than only rigid or free). This
allows for the soil-pile interaction stiffness to be incorporated into the JacketSE model, as discussed in Section 2.3.2.

28
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Table 14. Input Parameters for FRAME3DD Handled by JacketSE’s Inputs—For More Details See Gavin (2010)

Parameter Type Description Default Value

sh f g Boolean switch 0/1 = do not/do 1
include shear defor-
mation effects
Δz Floating point (m) Step size along -1 (ignore) (a)
member axis for
internal load calcula-
tions
gm f g Boolean switch 0/1 = do not/do 0
include geometric
stiffness effects
gg Three-element floating Inertial frame accel- [0, 0, −9.81]
point array eration
nmod Integer Number of eigenval- 6
ues/eigenmodes to
calculate
Mmod Integer switch Dynamic eigen- 1
value method: 1 =
subspace-Jacobi
iteration; 2 = Stodola
(matrix iteration)
lm Integer switch Mass modeling 0
method: 0 = consis-
tent mass matrix; 1 =
lumped mass matrix
δsh f Floating point Frequency shift 0
factor for rigid-body
modes in case of
partially restrained
structure
Rdx Integer switch Matrix condensation 2
method: 0 = none; 1
= static; 2 = Guyan;
3 = dynamic
a Internal forces are calculated at element nodes, i.e., more than one element (ndiv > 1) is
recommended with the current version of JacketSE.

29
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 15. Examples of ULS, SLS, and FLS Load PSFs—From IEC 2005

Unfavorable Loads Favorable Loads

Limit State
Normala Abnormal Transport and Installation All DLCs
ULS/SLS 1.35 1.1 1.5 0.9
FLS 1.0 1.0 1.0 1.0
a Normal or abnormal attributes for the various DLCs are given by the standards—e.g., IEC
(2009).

pyFrame3DD also allows for lumped masses to be distributed at model nodes (for instance, to simulate TP or RNA
inertia) affecting both modal and static load response.
Whereas FRAME3DD is capable of directly using linear and trapezoidal external force distributions along the el­
ements, and of calculating element internal forces, the current versions of JacketSE solely relies on nodal forces.
The load module condenses the distributed pressure loads from wind and waves at element nodes, and the verifica­
tions are performed at the node level. As a result, it is important to have enough resolution to resolve the pressure
loads. Future versions will take advantage of element internal load calculations and relax the need for higher element
resolution.

2.12 Utilization Calculation

2.12.1 Safety Factors
The generic ULS limit state verification may be expressed as follows (IEC 2005):

γn S(Fd ) ≤ R( fd ) (2.47a)
or explicitly (2.47b)
fk
γ f Fk ≤ (2.47c)
γn γm

where S(Fd ) is the probability distribution of the generic, design (factored) load within the Load Resistance Factor
Design approach, R( fd ) is the analogous function for the material-factored resistance, Fd (Fk ) is the factored (unfac­
tored) characteristic load, fd ( fk ) is the material-factored (unfactored) resistance, γn is the consequence of failure PSF
(or ‘importance’ factor based on the redundancy and fail-safe characteristics of the various components), γ f is the
generic load PSF, and γm is the material PSF.
The intent of JacketSE is not to deliver a final design, but to inspect the impact of various design choices under
some key DLCs from different perspectives, including LCOE aspects, as mentioned in Section 1. More accurate
AHSE loads analyses and subcomponent FEAs are needed to arrive at a detailed design. Design standards provide
guidance for determining characteristic loads through AHSE loads analyses (Germanischer Lloyd 2012; IEC 2005).
Furthermore, the standards provide recommended values for γ f and γn (see also Tables 15–16) (IEC 2009, 2005;
Germanischer Lloyd 2012; DNV 2013; ISO 2007; API 2014). These PSFs represent the uncertainty in the load
stochastic distribution and in the load assessment, and can be input to JacketSE. Material PSFs can be either taken
from specific, recognized design codes—e.g., ANSI (2010), European Committee for Standardisation (2005), ABS
(2014), and DNV (2010)—, or alternatively their minimum values may be taken as in Table 17 from IEC (2005).
Note that, per AWEA (2012), the OWT and its components should satisfy the ‘L2’ exposure category (API 2014),
which also requires that a 500 yr robustness check be carried out with unity PSFs. This check could be accomplished
by setting appropriate unit PSFs and environmental conditions as inputs into JacketSE; however, given the levels of
approximation in JacketSE, higher than normal PSFs are recommended.

30
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 16. Examples of Minimum γn As a Function of Component Class—From IEC (2005)

Component γn
Comment
Class ULS FLS SLS
‘Fail-safe’ structural components whose failures
1 0.9 1.0 1.0 do not result in the failure of a major part of
a wind turbine (e.g., replaceable bearings)
‘Nonfail-safe’ structural components
2 1.0 1.15 1.0 whose failures may lead to the fail-
ure of a major part of a wind turbine
‘Nonfail-safe’ mechanical components that
link actuators and brakes to main structural
3 1.3 1.3 1.3
components for the purpose of implementing
nonredundant turbine protection functions

Table 17. Examples of Minimum γm As a Function of Failure Mode—From IEC (2005)

γm
Failure Mode
ULS FLS SLS
Yielding of duc- 1.1 (welded 1.0 (if elastic
1.1
tile materials and structural properties proven
Global buckling steel) to 1.7 by full-scale
1.2
of curved shells (composites) testing); 1.1
Rupture from ex- (otherwise)
ceeding tensile or 1.3
compression strength

31
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
JacketSE recasts Eq. (2.47c) in utilization equations for the tower and the substructure. Although the utilization
checks should be performed for FLS, SLS, and ULS, the current version of JacketSE only performs ULS and SLS
verifications.
ULS structural checks ensure that the material utilization is below one, and that deflections are below limits specified
by the user. The verification is conducted on steel members under tension and compression-bending and following
the main standards of reference.

2.12.2 Tower Utilization

The normal stresses (σz,Ed along the meridional and σθ ,Ed along the circumferential directions), the shear stress
(τzθ ,Ed ), and the von-Mises equivalent stress (σvm ) for the generic cross section of the tubular tower can conserva­
tively be written as:
Nd Md Dsh
σz,Ed = + (2.48a)
A 2Jxx
Dsh − tsh
σθ ,Ed = γ f (kw − 1)qmax (2.48b)
2tsh
Mz
τzθ ,Ed = 2Td /A + (2.48c)
2Amid tsh
σvm = σz2,Ed + σθ2,Ed − σz,Ed σθ ,Ed + 3τz2θ ,Ed (2.48d)

where the argument z was dropped without losing generality, kw is the dynamic pressure factor to calculate hoop
stresses function of cylinder dimensions and external pressure buckling factor per European Committee for Standard­
isation (1993), tsh is the shell t, Amid is the area inscribed by the midthickness line, and all the other symbols were
introduced earlier except for qmax , which is the maximum wind dynamic pressure expressed as in Eq. (2.49):

qmax = 0.5 ∗ ρa |U|2 (2.49)

The tower can be considered verified if, for all of the segments, it can be proved that the stresses are kept below the
allowable yield strength, and stability is guaranteed at a global and local level. These constraints can be expressed by
the following equations (Germanischer Lloyd 2005; European Committee for Standardisation 1993):
fy
σvm ≤ (2.50)
γ m γn

Nd βm Md
+ + Δn ≤ 1 (2.51)
κN p Mp

σz,Ed ≤ σz,Rd (2.52a)

σθ ,Ed ≤ σθ ,Rd (2.52b)
τzθ ,Ed ≤ τzθ ,Rd (2.52c)
kz kθ kτ
σz,Ed σθ ,Ed σz,Ed σθ ,Ed τzθ ,Ed
+ − ki + ≤1 (2.52d)
σz,Rd σθ ,Rd σz,Rd σθ ,Rd τzθ ,Rd

Eq. (2.50) states the constraint on the yield resistance of the material ( fy ).
Eq. (2.51) is the global (Eulerian) buckling constraint, where Nd and Md are the design axial and bending moment
loads, as calculated by the load module (see Section 2.10.4), N p and M p are the characteristic, buckling-critical, re­
sistance values, κ and βm are reduction factors for the flexural buckling and bending moment coefficient, respectively

32
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
(Germanischer Lloyd 2005), and Δn is a function of the slenderness (λ̄ ) of the tower as in Eq. (2.53):

Δn = 0.25κλ̄ 2
N p γm (2.53)
λ¯ = Ne

κ is given by:

√1 ≤1 for λ̄ > 0.2

κ= Φ+ Φ2 +λ¯ 2
1 for λ̄ ≤ 0.2 (2.54)
with:
Φ = 0.5 1 + αb λ¯ − 0.2 + λ̄ 2
αb is the imperfection factor used in the buckling calculation, set equal to 0.21 in JacketSE. The elastic buckling
resistance Ne is taken as:
π 2 EJxx
Ne = (2.55)
1.1sk
where sk is the buckling length factor, set equal to 2 for the tower in JacketSE, E is the Young’s modulus, Jxx is the
member cross-sectional, second area-moment of inertia. The buckling resistance values are taken as in Eq. (2.56):
fy
Np = A (2.56a)
γb
fy
M p = Wp (2.56b)
γb

where Wp is the cross-sectional bending modulus, and γb is the safety factor used in buckling verification, usually set
equal to 1.1 (Germanischer Lloyd 2005).
Eq. (2.52) is the local (shell) buckling constraint; σz,Ed , σθ ,Ed , and τzθ ,Ed are the design values of the axial, hoop,
and shear stress, respectively, σz,Rd , σθ ,Rd , and τzθ ,Rd are the corresponding characteristic buckling strengths, kz ,
kθ , kτ , and ki are constants given by the standards (European Committee for Standardisation 1993). The buckling
strengths are given by:
χz fy
σz,Rd =
γm
χθ fy
σθ ,Rd = (2.57)
γm
χθ fy
τzθ ,Rd = √
3γm
where the generic buckling reduction factor in the shell buckling verification (χ) is given by:
⎧
⎪
⎪ 1 for λ̂ ≤ λ̂0
⎪ η
ˆ
⎨
χ= 1 − β p λ̂ −λ0 for λ̂0 < λ̂ ≤ λ̂ p
λ̂ p −λ̂0
⎪
⎪ (2.58)
⎩ αimp for λˆ p ≤ λ̂
⎪
λ̂ 2
with λˆ p =
αimp
1−β p

and the characteristic values for λ̂0 , β p , and η are shown in Table 18. In the table, the expressions of λ̂ are given as

33
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
functions of σz,Rcr , σθ ,Rcr , and τzθ ,Rcr , i.e., the critical buckling stresses shown in Eq. (2.59):

Cz
σz,Rcr = 0.605E (2.59a)
rtr
Cθ ,s
⎧
ωb
⎪
⎪
⎪ 0.92E ωb rtr for Cθ < 20
0.92E ωCθrtr ωb
⎨
σθ ,Rcr = for 20 ≤ Cθ ≤ 1.63rtr (2.59b)
b
4
Cθ
⎪
E ωb
0.275 + 2.03 ωb rtr for > 1.63rtr
⎪
⎪
rtr2 Cθ
⎩

1
τzθ ,Rcr = 0.75ECτ √ (2.59c)
ωb rtr

where ωb is the dimensionless length parameter for shell buckling calculations, rtr is the ratio of req to teq , and the
factors Cz , Cθ , Cθ ,s , and Cτ are shown in Eq. (2.60):

2L √
ωb = rtr (2.60a)
D
⎧
⎪ 1.36 − 1.83
⎪
2.07
ωb + ω 2 for ωb < 1.7
⎨ b
Cz = 1 for 1.7 ≤ ωb ≤ 0.5rtr (2.60b)
⎪
⎩ max 0.6, 1 + 0.2 1 − 2ωb
⎪
for ωb > 0.5rtr
6 rtr

Cθ = 1.5 assuming clamped-clamped boundary conditions (2.60c)

⎧
⎪
⎪ 1 + ω423 for ωb < 10
⎨ b
Cτ = 1 for 10 ≤ ωb ≤ 8.7rtr (2.60d)
⎪
⎩ 1 ωb
⎪
for ωb > 8.7rtr
3 rtr

The values for αimp to be used in Eq. (2.58) are also given in Table 18, as a function of meridional compression
fabrication quality parameter European Committee for Standardisation 1993 (Q) and the ratio of req to teq (rtr).
JacketSE assumes Q=25, i.e., class ‘B’ fabrication tolerance, and calculates rtr as:
req D2 + D1
rtr = =
teq cos βc 4 (t2 + t1 ) cos βc (2.61)
D2 − D1
with βc = arctan
2hc
where req and teq are the equivalent OR and t for a tapered member (a segment of the tower shell), taken as the
average OR and t, βc is the cone angle for the typical tapered shell element, and hc is the height of a tapered shell
member.
JacketSE returns the left-hand side of Eqs. (2.50), (2.51), and (2.52) as utilization values.
In addition to FLS and ULS checks, an important structural check for both blade and tower design is the verification
of potential blade strike on the tower, which is part of the SLS checks. Blade maximum deflection may occur under a
transient event as in the case of extreme gusts or fault situations. In order to verify that a safety margin remains in the
deflection of the blade before the tower strike, the maximum deflection across all ULS DLCs must be determined.
Standards such as IEC (2005) and Germanischer Lloyd (2012) offer guidance on the PSFs to employ in this calcula­
tion: IEC (2005) uses the same load PSF as for any other ULS DLC, whereas Germanischer Lloyd (2012) states that
the blade clearance shall not be less than 30% of the value under unloaded (at rest) conditions. JacketSE can return
the deflection at the top of the tower and that combined with the blade deflection can be used to check for this SLS
criteria.

34
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 18. Values for λ̂0 ,η, and β p As a Function of Stress
Type—From European Committee for Standardisation (1993)

Stress Type λ̂ λ̂0 βp η αimp

fy 0.62 (a)
Axial 0.2 0.6 1
σz,Rcr 1+1.91(Δwk /t)1.44
fy
Hoop σθ ,Rcr 0.4 0.6 1 0.65
c
f
Shear √ y 0.4 0.6 1
3τzθ ,Rcr

a Δwk is the characteristic imperfection

√ amplitude European Committee
for Standardisation 1993, Δwk = Q1 rtr.

2.12.3 Jacket Utilization

The substructure utilization is calculated at the joints and members per API (2014). For simplicity, all the joints
between two cross-braces are considered as x-joints, and all of the joints involving a brace and a leg are considered
as k-joints. Note that a rigorous definition of k-joint and x-joint in terms of geometry and loading conditions is given
in API (2014).
For members subject to combined axial compression and bending, the equations to satisfy are:

2 2
σa Cm σb,x + σb,y
+ ≤1 (2.62)
Fa 1 − σa F Fe t b

2 +σ2
σb,x
σa b,y
+ ≤1 (2.63)
0.6 fy Fb

where σa is the normal stress caused by axial force, Fa is the allowable axial compressive stress, fy is the charac­
teristic yield strength, σb,x is the normal stress caused by bending about local x, σb,y is the normal stress caused by
bending about local y, Fa is the allowable axial compressive stress, Fb is the allowable bending stress, Fe t is the Euler
stress divided by a safety factor per AISC 1989 and shown in Eq. (2.64), Cm is the reduction factor in the calculations
of the utilization for members under compression and bending per API 2014 and shown in Eq. (2.65).

12π 2 E
Fe t = (2.64)
23KLR
σa
Cm = 1 − 0.4 (2.65)
Fe t

35
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
The allowables are given by:
2
1 − KLR f
2C2 y
c
Fa = 3 for KLR < Cc (2.66a)
5/3 + 3KLR KLR
8Cc − 8C3 c

12π 2 E
Fa = for KLR ≥ Cc (2.66b)
23KLR2
0.5
2π 2 E
with Cc = (2.66c)
fy2
10,340 N m−2
⎧ D
⎪
⎪ 0.75 fy for t ≤ fy
10,340 N m−2 20,680 N m−2
⎨
fy D D
Fb = 0.84 − 1.74 Et fy for fy < t ≤ fy (2.66d)
20,680 N m−2
⎪
⎩ 0.72 − 0.58 fy D f D 300 N m−2
⎪
Et y for fy < t ≤ fy

where fy2 is the allowable used in local buckling axial stress determination per API 2014 given by:
fy2 = min ( fxc , fxe ) (2.67)
where the inelastic ( fxc ) and elastic ( fxe ) buckling stresses are given by:
D 0.25 D
fy 1.64 − 0.23 t for t > 60
fxc = (2.68)
D
fy for t ≤ 60
t
fxe = 2C f xe E (2.69)
D
where D is the generic member OD, C f xe is the reduction factor in the calculation of the elastic buckling stress per
API 2014, which is normally equal to 0.3.
For members under axial tension, the verification consists of the following inequality:
σa ≤ 0.6 fy (2.70)

The left-hand sides of Eqs. (2.62)-(2.63) and Eq. (2.70) are used by JacketSE to directly output utilization values at
all of the jacket members. The verification is conducted at all FEA nodes, and the stresses are calculated via basic
mechanics starting from the loads obtained from the FEA solver—see Eq. (2.71):
Nd
σa = (2.71)
A
Md D
σb = (2.72)
2Jxx
(2.73)
where σb is normal stress caused by total bending moment.
Note that the jacket legs are assumed to be flooded, thus hydrostatic effects are ignored for the leg members. The
mud- and cross-braces are sized through an engineering rule (see Sections 2.6 and 2.5) that grants sufficient strength
over hydrostatic effects. In a future release of JacketSE, stresses due to hydrostatic pressure will be directly incorpo­
rated into the utilization calculations.
Joints are verified by considering loads from the intersecting members, denoted as the chord and brace. For a k-
joint, the chord is the jacket leg; for an x-joint, the verification is repeated twice by having each of the two braces
alternately play the role of the chord. The joint utilization is calculated via Eq. (2.74):
M j 2 �� M j ��
� �
Pj
+ +� � ≤ 1 (2.74)
Pa Ma IP Ma OP

36
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 19. Values for c1 ,c2 , and c3 As a Function of Load-
d
ing Conditions and Joint Type and β j = d jbjc —From API (2014)

Joint type c1 c2 c3
k-joint under brace axial loading 0.2 0.2 0.3
x-joint under brace axial loading(a) β j ≤ 0.9 0.2 0 0.5
x-joint under brace axial loading(a) β j = 1 -0.2 0 0.2
a Linear interpolation between β j =0.9 and β j =1.

where Pj is the brace axial load at the joint, Pa is the allowable capacity for brace axial load, M j is the brace bending
load at the joint, Ma is the allowable capacity for brace bending load, IP stands for in-plane, OP stands for out-of­
plane.
The basic joint capacities are calculated as in Eq. (2.75):

fyct 2jc
Pa = Qub Q f c (2.75)
γ j1 sin θ j
fyct 2jc d jb
Ma = Qub Q f c (2.76)
γ j1 sin θ j
where Qub is the strength factor per API 2014, Q f is the chord load factor per API 2014, fyc is the chord allowable
stress used in joint verification, t jc is the chord thickness in a joint verification (also tleg ), θ j is the smaller angle de­
scribed by the brace’s and chord’s axes at the joint, γ j1 is the brace load PSF in a joint verification, which is normally
set at 1.6.
Q f c is calculated as in Eq. (2.77):
Pγ Mc,IP γ j2
Q f c = 1 + c1 Pc ycj2 − c2 M pc − c3 A2jnt
c
γ j2 Pc 2 γ j2 Mc,IP 2 (2.77)
with A jnt = Pyc + Mp

where c1 -c3 are given in Table 19, Pc is the chord axial load at the joint, Pyc is the yield capacity of the chord at the
joint, Mc,IP is the chord IP bending load at the joint, M pc is the plastic moment capacity of the chord at the joint, γ j2
is the chord PSF in a joint verification, which is normally set equal to 1.2. The average of the chord loads on either
side of the brace intersection should be used in Eq. (2.77). The chord axial load is positive in tension, and chord IP
bending is positive when it produces compression of the joint footprint.
The strength factor Qub is given by API (2014) as a function of brace loading type and joint type, as shown in Ta­
ble 20. Note that, in Table 20, a gap factor Qg is used, which is a function of the ratio between the adjacent brace
t
distance at the k-joint with the leg (g p ) over the chord OD (d jc ), and of the ratios of brace-to-chord t’s ( t jbjc ) and fy ’s
f
( fyb
yc
). In JacketSE the gap is calculated as:

d jc π π Dbrc2 Dbrc1
gp = tan − αbl1 + tan − αbl2 − − (2.78)
2 2 2 sin (αbl2 )/2 sin (αbl1 )/2
where αbl1 and αbl2 are the angles between the braces and the leg, Dbrc1 and Dbrc2 are the braces’s ODs at the k-joint
under examination.
As shown in Eqs. (2.74) and (2.77), it is necessary to calculate the IP and OP components of the bending moments
at the brace and chord. JacketSE performs a coordinate transformation from the FEA element local coordinate
system to the triad that identifies the plane of the joint.

37
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 d jb
Table 20. Values for Qub As a Function of Brace Loading and Joint Type and β j = d jc —From API (2014)

brace load
Joint type

axial tension axial compression IP bending OP bending

k-joint min 40β j1.2 Qg , (16 + 1.2γc ) β j1.2 Qg (a),(b)

(5 + 0.7γc ) β j1.2 2.5 + (4.5 + 0.2γc ) β j2.6

23β j for β j ≤ 0.9

x-joint 2.8 + (12 + 0.1γc ) β j Qbeta (c)

20.7 + β j − 0.9 (17γc − 220) for β j > 0.9

⎧
⎪
⎪
3 gp
⎪
max 1, 1 + 0.2 1 − 2.8g p /d jc ≥ 0.05
⎪
⎪
⎪
⎪ for d jc
⎪
⎪
⎪
⎪
⎪
⎨
aQ =
g t f √ gp
⎪
⎪ 0.13 + 0.65 t jbjc fyb
yc
γc for d jc ≤ −0.05
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩ linear interpolation otherwise
⎪
⎪

b d jc
γc = 2t jc
⎧
⎪
⎪
⎪
⎪ 0.3
⎪
⎪
⎨ β j (1−0.833β j )
for β j > 0.6
c Qbeta =
⎪
⎪
⎪
⎪
⎩ 1
⎪
⎪ for β j ≤ 0.6

38
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Figure 8. Typical unit vectors associated with x-joint and k-joint used in JacketSE to
determine IP and OP bending moment components with respect to the joint plane

The generic IP and OP bending moment vectors are calculated via simple geometry rules as in Eq. (2.79).

MOP = Mjnt · Cjnt,x Cjnt,x + Mjnt · Cjnt,y Cjnt,y (2.79)

⎡⎡ ⎤ ⎤
M jnt,x Cel,x
MIP = ⎣⎣M jnt,y Cel,y ⎦ · Cjnt,z ⎦ Cjnt,z (2.80)
M jnt,z Cel,z

In Eq. (2.79), Mjnt is the bending moment vector for the generic member at the joint of interest, whose components
are M jnt,x -M jnt,z in the local FEA element coordinate system, [Cel ] is the matrix identifying the local element coor­
dinate system, i.e., the transformation from F F 0 to F F 2 , with local x along the member axis and local y, z along
the cross-section principal axes of inertia, which identifies the local FEA element coordinate system with respect to
the global coordinate system, Cel,x -Cel,z denote the three-element rows of [Cel ], [Cjnt ] is the direction cosine matrix
identifying the joint plane, with x, y in the joint plane and z normal to it, with Cjnt,x -Cjnt,z denoting its three-element
rows (see Figure 8 and Eq. (2.81)):
⎤ ⎡ ⎤
îch
⎡
Cjnt,x
[Cjnt ] = ⎣Cjnt,y ⎦ = ⎣ îch × îbrc × îch ⎦ (2.81)
⎢ ⎥
Cjnt,z î × î
ch brc

where îch is the unit vector identifying the chord longitudinal axis, and îbrc is the unit vector identifying the x-brace
longitudinal axis. The unit vectors are easily obtained from the coordinates of two adjacent nodes in the member of
interest, which are also used to obtain δel , i.e., the distance vector between two adjacent nodes of a FEA element.
[Cel ] follows FRAME3DD’s convention and is given by:

⎡ ⎤
cos ψel cos θel cos θel sin ψel − sin θel
[Cel ] = [Cφ ][Cψ,θ ] = ⎣cos ψel sin φel sin θel − sin ψel cos φel cos φel cos ψel + sin φel sin ψel sin θel cos θel sin φel ⎦
cos ψel cos φel sin θel + sin ψel sin φel − sin φel cos ψel + cos φel sin ψel sin θel cos θel cos φel
(2.82)

39
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Figure 9. The element (F F 2 ), intermediary (F F 1 ), and global (F F 0 ) coordinate system used in Jack­
etSE. The generic element is shown as an AB cylindrical rod; the other symbols are explained in the text

with ⎡ ⎤ ⎡ ⎤
cos ψel cos θel cos θel sin ψel − sin θel 1 0 0
[Cψ,θ ] = ⎣ − sin ψel cos ψel 0 ⎦ [Cφ ] = ⎣0 cos φel sin φel ⎦ (2.83)
sin θel cos ψel sin θel sin ψel cos θel 0 − sin φel cos φel
where [Cψ,θ ] is the matrix describing the transformation of coordinate systems from F F 0 to F F 1 via a ro­
tation θel about the global y and a rotation ψel about global z, θel is the rotational angle about the global y (per
FRAME3DD’s convention) (see also Figure 9 for the definition of the F F 0 , F F 1 , and F F 2 coordinate sys­
tems), ψel is the rotational angle about the global z (per FRAME3DD’s convention), [Cφ ] is the matrix describing
the transformation of coordinate systems from F F 1 to F F 2 via a rotation about the local x, and φel is the eulerian
rotational angle about the local x, which is taken as 0 rad for tubular members. θel and ψel are calculated as in Eqs.
(2.84)-(2.85):

δel,y
ψel = arctan (2.84)
δel,x
δel,z
θel = arctan (2.85)
2 +δ2
δel,x el,y

where δel,x -δel,z are the components of the distance vector δel , see also Figure 9.
4
Finally, because API (2014) permits a 33% increase of the given allowables, JacketSE includes a 3 factor in all of the
allowables shown above for the jacket utilization.

40
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
3 Preliminary Verification and Validation

Damiani and Song (2013) show a comparison of the results in terms of modal and static loads analyses performed
by JacketSE and ANSYS for a few representative jacket-tower configurations. ANSYS models utilized second-order
beam and pipe elements but also shell elements for the TP. The results showed little dependence on the number of
elements used.
Two figures from Damiani and Song (2013) are reproduced in Figure 10. Figure 10a shows the first eigenmode
for a structure supporting a 10 MW OWT as calculated by ANSYS. The associated frequency (0.2132 Hz shown
in the graph) matched the JacketSE-calculated one of 0.21 Hz rather well. Figure 10b shows the von-Mises stress
distribution in the jacket. The maximum pile and leg axial forces were within 5% of those calculated by JacketSE’s
pyFrame3DD, thus lending credibility to the loads and utilization calculation. For more details on these and more
results and verification against ANSYS, see Damiani and Song (2013).
Damiani et al. (2016) examines the effect of varying turbine ratings, hub heights, water depths, waves, and tower-top
masses on the structural overall support mass for OWTs. The primary motivation of that research lay in determining
key relationships between environmental design drivers and costs associated with wind plant BOS, as well as with
the operation and maintenance.
The results obtained by running optimizations of jacket-tower configurations were compared to data available from
industry experience. Data from actual installations and other studies (GL Garrad Hassan 2012; BVG Associates
2012) that have predictions on steel quantities are plotted in the graphs of Figure 11. In that figure, the surface that
best fits the model-devised data was extrapolated towards lower hub heights to compare to the industry data. As
shown, the results of the model, extrapolated toward lower hub heights and water depths, agree relatively well with
the industry data for water depths between 20 m and 40 m, but underestimate and overestimate the mass for shallower
and deeper depths, respectively. This may be because of the coarse resolution in the data considered, which did not
include any points in the depth range between 45 m and 55 m and below 20 m. The gradient of the jacket mass with
respect to water depth above 40 m and below 10 m may therefore be artificially overestimated.

41
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 (a)

(b)

Figure 10. Results from the ANSYS analysis of a jacket-tower-pile configu­

ration for a 10 MW OWT; (a) shows the first eigenmode; (b) shows the von-
Mises stress distribution in the jacket members. From Damiani and Song (2013)

42
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 (a) 3D view (b) x-z projection of the same data

(c) y-z projection of the same data (d) Same data, but oriented to show the
distance of data points from best fit curve

Figure 11. Jacket steel mass trend for 6 MW turbine configurations. The study 81 data points are de­
noted by filled circles, with colors indicating RNA mass as denoted by the legend in (b). The sur­
face (bilinear interpolation of the data points) is color-coded by jacket mass tonnage, and the leg­
end is given in (a). In all the plots, the z-axis shows jacket mass made non-dimensional with its
average across all the 6 MW cases. Other symbols indicate: existing installations of 6 MW off­
shore turbines (triangles); predictions from the Crown Estate study BVG Associates 2012 (di­
amonds); and predictions from GL Garrad Hassan 2012 (squares). From Damiani et al. (2016)

43
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
4 Case Study

A simple case was set up to show the capabilities of JacketSE. A 6 MW turbine was imagined installed at a site
with a 40.8 m water depth. The main loads from the RNA and turbine parameters are given in Table 21, whereas
environmental parameters are shown in Table 22. The soil was assumed to be sand with stratigraphy as per Table 3.
Piles were considered unplugged. ASTM-992 steel was considered for all components (ASTM 2015), but its density
was taken as 8500 kg m−3 .
An optimization scheme that makes use of SNOPT—Sparse Nonlinear OPTimizer, a software package for solving
large-scale optimization problems (Gill, Murray, and Saunders 2005)—was implemented. The objective was set to
minimize the total structural mass. Bounds for the design variables are given in Table 23. Note that the tower top OD
was fixed at 3.51 m to match a given nacelle yaw bearing size. The jacket was assumed to be of the prepile type with
vertical piles, and a TP lumped mass of 75 t (metric tonnes) was also included. Five bays were selected based on past
experience with similar designs.
The initial-guess configuration did not meet either the pile capacity constraint or the eigenfrequency target, but
met utilization constraints. Results of the optimization are shown in Figures 12 and 13, and Table 24. Figure 12
shows the stick diagram of the resulting jacket-tower configuration; Figure 13 shows the outline of the tower and
the calculated utilization values along the span for the two DLCs. It can be seen from the graph in Figure 13, that
the local buckling criteria drove the tower design. In general, the operational conditions dominated the design when
compared to the parked DLC. Maximum utilization was seen just above the mid-height of the tower.
The obtained values for the design variables and the component mass schedule are given in Table 24, where they
can also be compared to the initial guess values. The batter did not change significantly, and, as a result, neither did
the footprint. The tower and TP mass decreased with respect to the starting configuration, whereas the piles’ and
the main lattice’s mass increased. This increment in mass was necessary to keep the first eigenfrequency within the
selected tolerance and to ensure the pile capacity constraint was satisfied, and only caused some 1% penalty on the
overall mass.

Figure 12. Diagram of the jacket-tower configuration as calculated by the optimizer for this case study

44
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 21. Turbine and Substructure Parameters for the Case Study

Parameter Unit Value Comments

Generator rating MW 6
zhub m MSL 97 From project data
zdbot m MSL 16 From project data
nbays - 5 Based on experience
TP lumped mass t (metric tonne) 75
RNA mass t 365
RNA [Ixx , Iyy , Izz ] kg m2 [7.92E7, 4.26E7, 4.04E7]
RNA [Ixy , Ixz , Iyz ] kg m2 [2.39E2, 1.78E6, 2]
RNA CMoff m [−4.1, 0, 3.04]
DLC 1.6 Faero N [1.94E6, −2.12E5, −1.46E5] Max thrust case
DLC 1.6 Maero Nm [5.35E6, 1.56E7, −1.8E6] "
DLC 6.1 Faero N [1.80E5, −4.21E4, 1.28E5] Max wind gust case
DLC 6.1 Maero Nm [−1.42E4, 1.11E7, −1.39E5] "
RNA δthoff m [−7.76, 0, 3.5]
Target eigenfrequency Hz 0.28

Table 22. Environmental Parameters Used for the Case Study

Parameter Unit Value Comments

Water depth m 40.8 Long Island (US)
50 yr wave height m 17.6 From NOAA buoy data
50 yr wave period s 12.5 From NOAA buoy data
50 yr gust speed m s−1 55 From actual project
Max-thrust wind speed m s−1 20 From actual project

45
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 23. Key Constraints Used in This Study for the Sizing Tool

Parameters Units Values

Pile OD min-max m 0.8-2.5
Pile t min-max m 0.0381-0.0762
Max embedment length m 60
Max footprint m 23
Leg OD min-max m 0.8-3
Leg t min-max m 0.0318-0.0635
X-brace OD min-max m 0.32-2.4
X-brace t min-max m 0.0191-0.0508
Mud-brace OD min-max m 0.32-2.4
Mud-brace t min-max m 0.0318-0.0508
Brace OD/Leg OD - 0.3
Brace DT R - ≥ 31 AND ≤ 50
Brace buckling-parameter KLRc - ≤ 70.
Girder OD min-max m 0.8-1.3
Girder t min-max m 0.0127-0.0635
Deck width/Tower-base OD min-max m 2-2.1
Tower-base OD min-max m 4-6
Tower-top OD min-max m 3.51 (fixed)
Tower DT R - 120-250
Tower waist/tower length max - 0.25
First eigenfrequency acceptable band above target % 5

Figure 13. Calculated tower utilizations for the mass-optimized jacket-tower configuration used in
this case study. VonMises Util1 and Util2 refer to utilizations with respect to yield strength, GL Util1
and Util2 refer to global (Eulerian) buckling criteria, and EUsh Util1 and Util2 refer to the shell buck-
ling criteria. GL and EUsh Util1’s refer to the first DLC, while GL and EUsh Util2’s refer to the second

46
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
 Table 24. Values of the Design Variables for the Calculated Minimum Mass Configuration

Parameters Units Initial Guess Optimized Values

Pile OD m 1.2 1.61
Pile t m 0.04 0.0381
Embedment length m 60 60
Piles’ mass t 297 385
Batter - 15 14.87
Footprint m 18.22 18
Leg OD m 0.9 0.98
Leg t m 0.04 0.0318
X-brace OD m 0.6 0.59
X-brace t m 0.02 0.0191
Mud-brace OD m 0.8 0.98
Mud-brace t m 0.04 0.0318
Lattice mass t 509 467
Deck width m 12.6 11.94
Girder OD m 0.9 1.08
Girder t m 0.04 0.03
Strut OD m 0.9 0.98
Strut t m 0.04 0.0318
Diagonal brace OD m 0.9 1.08
Diagonal brace t m 0.04 0.03
Shell OD m 6 5.97
Shell t m 0.04 0.048
TP length m 6 6
TP mass t 206 198
Tower-base OD m 6 5.97
Tower-base t m 0.03 0.0323
Tower-top OD m 3.51 3.51 (fixed)
Tower length m 71.5 71.5 (fixed)
Tower waist height m 17.5 8.6
Tower mass t 263 243
Total mass t 1276 1293
First and second eigenfrequencies Hz 0.3,0.31 0.29, 0.296

47
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
5 Conclusions and Future Work

The support structure plays an important role in defining the OWT system reliability and performance characteris­
tics, but it is also the primary driver in the BOS costs. Suboptimal designs of the support structure are particularly
detrimental because they directly influence CapEx and the O&M costs. Important trade-offs in the design of an OWT
subsystems could be easily missed by the industry practice, uncoupled design-process, in which each component
is designed independently. The system cost resulting from such an approach can be higher than that of the optimal
solution. In contrast, an integrated design of the turbine and support structure might significantly lower the overall
system cost. To enable this system-level optimization, we developed JacketSE, a CAE tool within NREL’s systems
engineering WISDEM toolbox, which connects physics-based models of all major system components, and allows
engineers and designers to explicitly capture the trade-offs between their designs. JacketSE leads to configurations of
towers, jackets, and piles for given environmental conditions and turbine parameters.
JacketSE is based on a simplified treatment of the physics, namely simplified hydrodynamics and material mechan­
ics. The main modules were described together with the math and standards equations used for the sizing, load
calculations, and member utilization calculations. Preliminary verification against the commercial package ANSYS
showed good accuracy in the modal and static loads analyses performed by the FEA solver in JacketSE.
A case study was run to show the capabilities when coupled with an optimization algorithm seeking minimum
overall mass. Only a couple of design driving load cases were considered for the optimization cycle, but results have
proven encouraging when compared to industry data in other studies (Damiani et al. 2016).
JacketSE can be used to evaluate LCOE trends as functions of main parameters and to assess CapEx and LCOE
sensitivity to: project characteristics (size and turbine spacing), technology (turbine size, substructure and foundation
type, electrical infrastructure), and geospatial properties (water depth and distance to shore).
Future research will enhance and validate the model capabilities, including revised FEA, hydrodynamics, fatigue,
and soil-pile interaction.

48
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
Bibliography

ABS. 2014. Guide for BUCKLING AND ULTIMATE STRENGTH ASSESSMENT FOR OFFSHORE STRUCTURES.
American Bureau of Shipping, ABS Plaza, 16855 Northchase Drive, Houston, TX 77060 USA.
AISC. 1989. AISC 335:1989 - Specification For Structural Steel Buildings - Allowable Stress Design, Plastic Design.
Superseded by AISC 360-10. AISC, American Institute of Steel Construction, One East Wacker Drive, Suite
700-Chicago, Illinois 60601-1802.
ANSI. 2010. Specification for Structural Steel Buildings. 3rd printing: 2013. ANSI/AISC, American Institute of Steel
Construction, One East Wacker Drive, Suite 700-Chicago, Illinois 60601-1802.
API. 2014. Planning, Designing and Constructing Fixed Offshore Platforms - Working Stress Design. API RECOM-
MENDED PRACTICE 2A-WSD. American Petroleum Institute, API Publishing Services, 1220 L Street, NW,
Washington, DC 20005.
ASTM. 2015. ASTM A992 / A992M - 11(2015) Standard Specification for Structural Steel Shapes. ASTM, West
Conshohocken, PA. www.astm.org.
AWEA. 2012. Offshore Compliance Recommended Practices Ű Recommended Practices for Design, Deployment
and Operation of Offshore Wind Turbines in the United States. AWEA.
Bowles, J. E. 1997. Foundation Analysis and Design. 5th. 1168 pp. The McGraw-Hill Companies, Inc.
Burton, T., et al. 2005. Wind Energy Handbook. 1st. 617. 111 River St., Hoboken, NJ 07030, USA: John Wiley /
Sons, Inc.
BVG Associates. 2012. Offshore wind cost reduction pathways Ű technology work stream. Technical Report. The
Crown Estate.
Chakrabarti, S. K. 1987. Hydrodynamics of Offhore Structures. 440 p. Southampton, UK: WIT Press.
– , ed. 2005. Handbook of Offshore Engineering. 2 volume set –1321 pp. The Boulevard Langford Lane, Kidlington,
Oxford OX5 lGB, UK: Elsevier.
Cordle, A., G. McCann, and W. de Vries. 2011. “Design drivers for offshore wind turbine jacket support structures”.
In ASME 2011 30th International Conference on Ocean, Offshore and Arctic Engineering- OMAE2011, 419–428.
Rotterdam, THe Netherlands: ASME.
Damiani, R. 2016. “Offshore WInd Turbine Design”. Chap. 8—Offshore Wind Turbine Tower Design, ed. by Else-
vier. In Press. Elsevier.
Damiani, R., and H. Song. 2013. “A Jacket Sizing Tool for Offshore Wind Turbines Within the Systems Engineering
Initiative”. In Offshore Technology Conference. Houston, Texas, USA: Offshore Technology Conference. doi:http:
//dx.doi.org/10.4043/24140-MS.
Damiani, R., et al. 2016. “Scenario Analysis For Techno-Economic Model Development of U.S. Offshore Wind
Support Structures”. In review, Wind Energy.
De Vries et al. 2011. Final Report WP4.2: Support Structure Concepts for Deep Water Sites. Technical Report Up-
Wind_WP4_D4.2.8_Final Report. Project Upwind Contract No. 019945 (SES6).
DIN. 2005. DIN 1055-4 (2005-03): Action on Structures - Part 4: Wind Loads. Deutsches Institut Fur Normung E.V.
DNV. 2010. Offshore Standard DNV-OS-C501 - Composite Components. Det Norske Veritas AS, Veritasveien 1 -
N-1322 HØVIK,Norway.
DNV. 2013. Design of Offshore Wind Turbine Structures. Offshore Standard. DNV, DET NORSKE VERITAS AS.
Dykes, K., et al. 2011. Applications of Systems Engineering to the Research, Design, and Development of Wind
Energy Systems. Technical Report TP-5000-52616. Contract No. DE-AC36-08GO28308. 1617 Cole Boulevard,
Golden, Colorado: NREL.
European Committee for Standardisation. 1993. Eurocode 3: Design of Steel Structures—Part 1-6: General rules—
Supplementary rules for the shell structures. European Committee for Standardisation.
– . 2005. Eurocode 3: Design of Steel Structures—Part 1-1: GenStructures and rules for buildings. European Com-
mittee for Standardisation.
– . 2010. Eurocode 1: Actions on Structures—Part 1-4: General actions—Wind actions. European Committee for
Standardisation.

49
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications
EWEA. 2015. The European offshore wind industry - key trends and statistics 2014. Tech. rep. European Wind
Energy Association.
Gavin, H. P. 2010. “User Manual and Reference for Frame3DD: A Structural Frame Analysis Program” [inlangC++].
Duke University. https://nees.org/resources/1650/download/frame3dd.pdf.
Germanischer Lloyd. 2005. Guideline for the Certification of Offshore Wind Turbines. Germanischer Lloyd, German-
ischer Lloyd.
– . 2012. Guideline for the Certification of Offshore Wind Turbines. Germanischer Lloyd, Germanischer Lloyd.
Gill, P. E., W. Murray, and M. A. Saunders. 2005. “SNOPT: An SQP Algorithm for Large-Scale Constrained Opti-
mization”. SIAM REVIEW 47 (1): 99–131. doi:10.1137/S0036144504446096.
GL Garrad Hassan. 2012. “Expected Offshore Wind Farm balance of Station Costs in the United States”. NREL’s
Subcontractor’s report.
IEC. 2005. 61400-1. Wind turbines - Part 1: Design requirements. International Electrotechnical Commission, IEC
TC-88.
– . 2009. 61400-3 Wind turbines - Part 3: Design requirements for offshore wind turbines. International Electrotech-
nical Commission, IEC TC-88.
ISO. 2007. 19902:2007 - Petroleum and natural gas industries – Fixed steel offshore structures. ISO, ISO, ISO
copyright office, Case postale 56, CH-1211 Geneva 20, Switzerland.
– . 2014. 19901-1:2005 (Modified) - Petroleum and natural gas industries – Specific requirements for offshore
structures – Part 1: Metocean Design and Operating Considerations. ANSI/API Recommended Practice 2MET.
ISO, ISO, ISO copyright office, Case postale 56, CH-1211 Geneva 20, Switzerland.
Matlock, H., and L. C. Reese. 1960. “Generalized solutions for laterally loaded piles.” J. Soil Mechanics & Founda-
tion Division 86 (5): 91–97.
Molde, H., D. Zwick, and M. Muskulus. 2014. “Simulation-based optimization of lattice support structures for
offshore wind energy converters with the simultaneous perturbation algorithm”. Journal of Physics: Conference
Series 555 (012075). doi:doi:10.1088/1742-6596/555/1/012110.
Mone, C., et al. 2015. Cost of Wind Energy Review. Technical Report TP-5000-63267. Golden, CO: NREL.
Murtagh, P. J., B. Basu, and B. M. Broderick. 2007. “Gust Response Factor Methodology for Wind Turbine Tower
Assemblies”. J. of Structural Engineering 133 (1): 139–144. doi:http : / / dx . doi . org / 10 . 1061 / (ASCE ) 0733 -
9445(2007)133:1(139).
Murthy, V.N.S. 2002. Geotechnical Engineering - Principles and Practices of Soil Mechanics and Foundation
Engineering. Civil and Environmental Engineering. 1056 pp. 270 Madison Ave., New York, NY 10016: Marcel
Dekker, Inc.
Musial, W. D., S. Butterfield, and B. Ram. 2006. “Energy from Offshore Wind”. In Offshore Technology Conference.
Houston, TX.
NREL. 2015. WISDEM - The Wind-Plant Integrated System Design and Engineering Model set. Accessed on
November 4, 2015; https://github.com/WISDEM.
Panzer, H., et al. 2009. Generating a Parametric Finite Element Model of a 3D Cantilever Timoshenko Beam Using
Matlab. Technical Reports on Automatic Control TRAC-4. Technische Universitat Munchen.
Pender, M. J. 1993. “Aseismic Pile Foundation Design Analysis”. Bullettin of the New Zealand National Society for
Earthquake Engineering 26 (1): 49–159.
Randolph, M. F. 1981. “Pile subject to torsion”. J. Geotech. Engrg. Div. 107:1095–1111.
SEI. 2005. Minimum Design Loads for buildings and other structures. 388 pp. ASCE Standard, ASCE/SEI 7-05.
American Society of Civil Engineers.
– . 2010. Minimum Design Loads for buildings and other structures (ASCE/SEI 7-10). 3rd printing. 388 pp. ASCE
Standard, ASCE/SEI 7-10. 1801 Alexander Bell Drive, Reston, Virginia 20191: American Society of Civil Engi-
neers.
Timoshenko, S. 1970. Theory of elasticity. 3rd. Engineering societies monographs. 567 pp. McGraw-Hill Kogakusha
Ltd.
Zwick, D., et al. 2014. “Comparison of different approaches to load calculation for the OWEC Quattropod jacket
support structure”. Journal of Physics: Conference Series 555 (012110). doi:doi:10.1088/1742- 6596/555/1/
012110.

50
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications