OTC-20674-PP

Qualification of a Semi-Submersible Floating Foundation for Multi-

Megawatt Wind Turbines
Christian Cermelli and Alexia Aubault, Marine Innovation & Technology, Dominique Roddier, Principle Power Inc,
and Timothy McCoy, DNV Global Energy Concepts Inc.

Copyright 2010, Offshore Technology Conference

This paper was prepared for presentation at the 2010 Offshore Technology Conference held in Houston, Texas, USA, 3–6 May 2010.

This paper was selected for presentation by an OTC program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material does not necessarily reflect any position of the Offshore Technology Conference, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Offshore Technology Conference is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of OTC copyright.

Abstract
Recent trends in the wind industry point to the use of increasingly larger and more powerful machines with rated power
ranging from 5 to 10MW exclusively designed for offshore use. Floating foundations offer greater flexibility in term of site
selection for wind farms, and if properly designed, may result in comparable availability with equivalent offshore turbines on
fixed foundations, while reducing the complexity and risks associated with offshore installation.

The WindFloat platform is a semi-submersible platform with three columns fitted with a large horizontal water-entrapment
plate at the base. The wind turbine and tower are fitted on one of the columns. The platform is designed to support
commercially available multi-megawatt wind turbines with no hardware modification to the turbine. The qualification
process followed for the development of a 150MW wind farm offshore Portugal is discussed.

Because of economic constraints, optimization of the platform is essential to achieve project financial targets. A rational and
comprehensive process was followed to optimize the system while maintaining the robustness required to survive in the
offshore environment. The design process is based on a combination of advanced numerical analysis and scale model
experimentation. Full-scale experimentation is ongoing.

Selected design codes and industry standards are applied. The return period of extreme events is adjusted based on
experience acquired by the wind industry.

Because of the considerable aerodynamic loads generated by the wind turbine and their effects on platform motion, the ability
to solve the combined aerodynamic and hydrodynamic problem is necessary. Additional factors, such as tower dynamics
and turbine controls must also be taken into account. Development of a coupled hydro-servo-aero-elastic model constitutes
a key element of the qualification process.

Rules for the design of Mobile Offshore Drilling Units are followed to verify platform stability in intact condition, and to
ensure that compartmentation is sufficient to withstand damaged conditions with any compartment flooded.

Time-domain coupled analysis of the platform and its mooring system are performed. The hydrodynamic response is also
validated with scale model tests.

A four-point mooring configuration is adopted. Reliability considerations are discussed to assess the level of redundancy
required in the mooring system, in particular with regards to the one-line damaged condition.

The platform structural design is conducted using applicable offshore codes from several classification societies, as well as
extensive finite-element analysis. Structural dynamics is critical to platform design due to the dynamic excitation of the
tower and the corresponding loading of the substructure primary members.
OTC 20674-PP 2

Introduction

The wind power industry has been growing rapidly in recent years due in large part to concerns related to the impact of fossil
fuels on climate change. While the net benefit of wind power generation on the environment is well-established, there are
several barriers encountered by many onshore wind projects. These include visual impact, noise, land availability, and
wildlife impacts. This has provided some of the incentives to develop wind farms offshore. The stronger and steadier wind
resource typically encountered offshore provides further incentive.

At the same time, major wind turbine manufacturers are spending considerable resources in increasing the rated power of
individual wind turbine. 5MW turbines have already been installed and turbines of up to 10MW rated power are expected to
be commercialized in a few years. These very large units with rotor diameters between 120 and 150m are almost exclusively
designed for offshore installation, as their transportation and installation inland is not practical. Offshore installation of large
wind turbine units requires complex logistics and large offshore vessels. Increasing power produced per device is a way of
improving the economics of offshore wind power.

While most offshore installations to date have been conducted in shallow water (20 m depth or less), site selection and
permitting of wind farms offshore is proving to be an issue since multiple parameters must be accounted for such as
proximity to grid connection or power demand, geotechnical conditions, visibility from shore, shipping lanes and other
marine activities like fishing and boating as well as ecological considerations. These parameters tend to drive the selection of
optimal wind farm sites further offshore and toward areas with greater water depth.

Efforts have therefore been expanded in providing foundations for multi-megawatt wind turbines in greater waterdepth, such
as the Beatrice wind farm demonstrator project in 45m of water in the North Sea, which consists of two jackets supporting
wind turbine towers. The cost of “fixed foundation” is expected to increase considerably with waterdepth, and economic
viability of such system in greater than 50m waterdepth will not likely be achieved. Beyond 50m depth, “floating
foundations” will likely result in lower cost. A variety of floating foundations concepts are currently being investigated [1],
[2]. The Hywind project is the first installation of a multi-megawatt turbine on a floater [3].

Floating foundation technology opens the door to very large coastal areas with attractive markets but few shallow water sites
suitable for offshore wind turbines. This encompasses the West coast of Europe from Portugal to Ireland, deep water regions
of the North Sea, the Northeastern and West coasts of the US, as well as coastal areas of Japan, Korea, etc.

The present paper discusses the process followed to qualify a semi-submersible platform for use as a floating foundation for
multi-megawatt wind turbines, also referred to as WindFloat in this paper. The WindFloat comprises three vertical
stabilizing columns with a large horizontal water-entrapment plate at their base. The wind turbine tower is supported on one
of the column. The columns are interconnected by tubular elements. The platform is anchored to the seabed with four to six
mooring legs.

The WindFloat platform has similarities with a large number of column-stabilized oil and gas platforms. It seems therefore
logical to make use of almost a half century of experience acquired by the oil and gas industry in the development of column-
stabilized platforms. This includes design guidelines issued by several classification societies. At the same time, there are
fundamental differences between the offshore wind power and oil and gas industries. The economics of oil and gas fields
allows for massive investments with good returns when the field conditions are favorable. The economics of offshore wind
power development remains marginal in present market conditions. Most offshore wind developments to date have been
subsidized. Risks in the offshore oil and gas business are much greater, including potential for significant loss of life, as
most exploration and production platforms are large manned structures, as well as major pollution. The risks linked to
offshore wind developments are considerably smaller since wind turbine structures are normally un-manned, except during
occasional maintenance, and there is no potential for significant pollution. Therefore indiscriminate application of offshore
oil & gas design codes will un-necessarily compromise the economics of offshore wind power projects.

Design guidelines for offshore wind turbine structures have been issued in recent years [4], [5], however these are essentially
applicable to fixed foundations. These have been created by merging design recommendations developed for onshore wind
turbines over a large number of years, with the experience of fixed oil and gas platforms. A similar approach is currently
adopted by several classification societies for the design of floating wind turbines, however, to date, no specific guidance has
been issued. Further complicating the issue for a structure such as the WindFloat is the complex interactions between the
turbine aero-servo -elastic responses, and the floating foundation hydrodynamic behavior. Large wind turbines are
complicated dynamic systems in which the tower structural dynamics is carefully tuned to avoid interactions with the
spinning turbine, and advanced control systems are implemented to maximize power output while preserving the integrity of
the equipment. When such a system is placed on a relatively small floating foundation, interactions between platform
motions and turbine performance are significant [6].
OTC 20674-PP 3

The lack of design standards for floating wind turbines and the severe economic constraints placed on such system require a
rigorous and scientific approach to the design to optimize the structural configuration while ensuring robustness in the
offshore environment. This paper discusses the main steps of the platform qualification process such as intact and damaged
stability analysis, hydrodynamics and model testing, mooring and cable analysis and structural analysis. These steps are
commonly used for qualification of oil and gas platforms; however key differences with such systems will be highlighted.
The paper places special emphasis on the development of a design software coupling hydrodynamic and aero-servo-elastic
responses, which is specifically made for floating wind turbine systems.

Many other important studies are also required to fully qualify the system. These include fabrication, installation and
transportation studies, geotechnical, and anchor sizing studies, operability studies and risk analysis, as well as environmental
impact assessment. These aspects have been studied during the qualification process; however they will not be presented
herein for lack of space.

System description

The WindFloat design described in this paper is based on currently available 5MW wind turbines with the aim of creating a
150MW offshore wind farm on the West coast of Portugal in 80m waterdepth; see artistic rendering in Figure 1.

Figure 1: Artistic rendering of a 150 MW floating wind farm

The main components of the WindFloat platform, shown in Figure 2, are briefly described: there are three vertical circular
columns interconnected by horizontal tubular members and bracing. The columns are stiffened-shell construction with ring
girders and vertical stiffeners. A water-entrapment plate is supported at the base of the columns and extends horizontally
outward. It is made of stiffened plates with radial stiffeners fixed to the column base and supported by vertical bracings
connecting the outer edge of the plate to the columns. The water-entrapment plate increases the platform added-mass and
damping. It is a key component of the platform hydrodynamic performance. The turbine tower is supported by one of the
columns. The turbine rotor and nacelle are fixed to the top of the tower. The hub height is 80m above mean water level.
There is a very minimal deck between the columns used to support some electrical equipment, which is more practical to fit
and maintain inside a deck house rather than inside the tower.

An active ballast water system is installed in the platform to transfer water between columns to compensate the change in
overturning moment due to change in mean wind speed or direction. The ballast system has no piping connection to the
outside to minimize risks associated with malfunction of ballasting equipment. The ballast system is only designed to
compensate mean overturning moment. It has the ability to transfer up to 200 tonnes of active ballast water in approximately
30 minutes using two independent flow paths with redundant pumping capability. One active ballast compartment is located
in the upper half of each column.
OTC 20674-PP 4

Figure 2: Water-entrapment plate (left), deck layout (center), and top of column (right)

Intact and damaged stability

The WindFloat intact and damaged stability was conducted in accordance with ABS MODU rules [7]. The RhinoMarine
software is used to model the hull and compute the righting moment curves and downflooding angles. Site specific
environmental criteria are considered for this analysis. These are based on hindcast wind and waves from the fine mesh
Global Re-analysis of Ocean Waves model for the North Atlantic basin (GROW-FAB) from Oceanweather Inc which
includes 50 years of continuous hindcast.

The main difference with drilling semi-submersible platform analysis is the definition of design conditions selected to
perform the stability analysis: the maximum wind speed conditions may not yield the maximum wind loads; indeed when
wind speed exceeds 25m/s, the turbine is normally shut-down. One may be tempted to conservatively assume that the
controller has likely failed and the turbine is still spinning in wind speed greater than 25m/s, however if this was the case, the
turbine would suffer catastrophic damage, possibly leading to the collapse of the tower. It is therefore critical that sufficient
redundancy be built-in the emergency systems of the turbine to prevent overspeeding of the rotor.

In addition, it is conservatively assumed that the ballast system has failed or that a sudden change in wind direction has
occurred, and as a result, the platform is improperly ballasted. This condition occurs simultaneously with the largest mean
overturning moment with the turbine shut-down. In this case, the downflooding angle is 20 degrees, which provides more
than 7 degrees of margin over the extreme heel angle.

Damaged conditions are also considered to verify that platform compartmentation is sufficient. In this case, it is assumed
that any single compartment (at or below the waterline) is flooded, due to collision, dropped object or other structural
problem.

In the case where an active ballast compartment is punctured, the ballast system may attempt to pump water to the other side
to reduce platform heel. This will result in flooding of the three active ballast compartments. This case is also considered to
assess platform survivability in damaged condition.

Compartmentation of the platform is designed to meet, in all damaged cases, criteria defined in [7] regarding downflooding
angle and righting moments.

In Northern Europe, extreme wind speeds during storms are such that they typically generate the highest wind loads due to
drag forces on the blades, nacelle and tower, even with the turbine shut-down; however this is not the case at the selected
Portugal location. Since the peak wind gusts offshore Portugal are relatively low compared with more Northern locations,
the mean overturning moment generated by the wind turbine is higher in operational conditions and reaches a maximum for
wind speed around 12 m/s. However, the platform control system is designed to shut-down the turbine if the platform heels
more than 10 deg, and therefore stability design cases are more onerous in extreme wind conditions.

Hydrodynamic analysis and model testing

The WindFloat hull design is based on the MiniFloat technology developed to provide a minimal platform for application to
deepwater marginal oil and gas fields. It has undergone comprehensive hydrodynamic qualification through numerical
analysis and model testing [8].
OTC 20674-PP 5

Numerical modeling of the WindFloat platform is conducted using two different pieces of software described in later
sections. A fully-coupled time-domain OrcaFlex model is generated whose primary purpose is to design the mooring system
and electrical cable. This is described in the “Mooring and cable analysis” section. Additionally, a time-domain algorithm
using the TimeFloat software was coupled to the FAST aerodynamic software to yield the fully coupled responses of the
floating wind turbine, including the blade aerodynamics, tower structural response, wind turbine controls as well as all
hydrodynamic quantities such as diffraction-radiation and viscous effects, and the mooring system. This work is described in
the section titled: “Coupling of hydrodynamic and aero-servo-elastic analysis”.

A model testing campaign was conducted at the University of California Berkeley Ship Model Testing Facility. The primary
objectives of the model tests were to:
- Verify numerical hydrodynamic model and platform motion response
- Determine clearance between extreme wave crest and platform deck
- Measure interactions between wave-induced dynamics and tower vibrations
- Measure hydrodynamic loading on the heave plate
- Measure mooring line tensions

A 1/67 scale geometrically similar model of the underwater hull section was fabricated in acrylic; see Figure 3. Stiffening
details towards the edge of the heave plates were scaled properly. The platform hull is modeled precisely to the top of
column, which is beyond the expected reach of the largest wave crests. The platform mass properties were scaled and verified
experimentally. Lead ballast was then added to the bottom of the columns and steel disks were placed on the heave plate to
fine tune the ballast. Steel disks were also added to the top of column to model the active ballast system.

Figure 3: Model in front of the wind machine (left), top of tower instrumentation (top right),
and strain gage setup on water-entrapment plate (bottom right)
Wind loads represent a very large component of the external loading onto the WindFloat platform. A correct representation
of wind loads is therefore important to the proper realization of the model test. These loads must also be Froude-scaled for
consistency with the rest of the model; which means, that the force should be proportional to the third power of the scale
factor. Because the Froude-scale airfoil would operate in a completely different Reynolds number range than the prototype,
the resulting aerodynamic forces would not match their target. In addition, the turbine control system, which includes a
constant adjustment of the blade pitch angle, would be exceedingly complicated to model at such small scale. A compromise
OTC 20674-PP 6

was therefore adopted in the aerodynamic representation of scaled turbine loads on the platform: the turbine was modeled
with a 27” diameter circular disk sized to provide the same thrust force as the prototype turbine (using Froude scaling). The
drag coefficient of a disk perpendicular to the incident flow is well-known and virtually Reynolds-number-independent over
a very large flow regime.

A wind machine composed of 6 high-velocity fans was built to generate sufficient wind speed to achieve the target wind
forces. 8”x8” vanes were built at the exhaust of the wind machine to reduce turbulence. The fan exhaust covered a 6 ft wide
by 4 ft tall area and was located approximately 3 ft in front of the drag disk. The circular drag disk was fixed at the top of the
tower with its center aligned with the rotor hub.

The following instrumentation was placed at the top of the tower:

- a load cell connects the drag disk to the top of tower assembly and measures the force perpendicular to the disk axis.
- a single axis accelerometer is placed at the hub height directly above the center of the tower. It measures accelerations in
line with the disk perpendicular axis. The accelerometer also measures gravity along its axis and can be used as an
inclinometer in static test, or can provide the mean heel angle of the platform.
- because the drag disk is fixed, the inertia of the spinning turbine blades is not modeled by the disk. Blade rotational inertia
on a floater may result in gyroscopic effects if the spinning turbine undergoes rotations along axes perpendicular to their axis
of rotation. For instance, yaw of the platform may result in a roll moment applied at the tower top when the turbine is
spinning. In order to model this effect, the rotational inertia of the blades is modeled using masses placed at the edges of a
spinning rod forced to rotate by an electrical motor.

The total weight of the instrumentation, drag disk, and rotor was designed to match the weight of the turbine and its
equipment at the top of tower. The tower top assembly is mounted on a connector that can be rotated around the tower top,
such that the disk remains always perpendicular to the dominant wind direction irrespective of the model/wind machine
orientation.

The tower dynamics due to its structural deformations was modeled by properly sizing a thin aluminum rod which was
guided at the top of column and anchored at the bottom of the column. In order to obtain proper stiffness, a short sleeve with
very tight fit, was slipped onto the lower section of the tower. Decay tests were conducted to match the required tower
bending frequencies.

Model motions were measured using an optical tracking system with two video cameras. The airgap was determined visually
using video camera. Black markers were placed every meter between the mean waterline and the top of column (Figure 4).

Figure 4: Airgap measurements using video recordings

Hydrodynamic load at the edge of the heave plate was measured using a strain gage set-up. Slits were cut in the heave plate,
so that loads on the outer section of the heave plate were transferred through to the remaining plate section through two small
tabs that were instrumented with a strain gage bending bridge. The load cells were calibrated by applying various lead
weights along the edge of the heave plate.

The mooring line configuration was kept as close as possible to the prototype mooring, however the mooring line azimuth
had to be modified because the UC Berkeley ship model testing facility is a towing tank and is not sufficiently large to
include the entire mooring lines. In addition the mooring lines were truncated at the middle of the lower rope section. The
stiffness of the mooring section was adjusted to yield an equivalent stiffness curve as the prototype mooring. The top section
of the mooring line including top chain and clump weight is the section that undergoes the most dynamics due to platform
motion; it was modeled exactly with the exception of the mooring line azimuth. Hydrodynamic simulations were conducted
OTC 20674-PP 7

to verify that the prototype and equivalent mooring systems responses were similar. Mooring tensions were monitored at the
anchors using load cells.

Calibration tests were conducted to adjust the wave spectra and wind forces to match their target. System identification tests
were carried out to validate the model test set-up and basic model physical and hydrodynamic properties; these were decay
tests in each degree of freedom, including tower vibration decay test, static offset tests, incline test and regular wave tests.

Tests were then run to determine the system responses in irregular wave conditions:
- Operational sea-state; i.e. 2.5m significant wave height (corresponding to the 90% non-exceedence at the site) and
12m/s wind, which leads to the largest wind overturning moment.
- Maximum operational sea-state: one year storm; i.e. 5.5m Hs with 12m/s wind – it is assumed that the turbine
remains operational in the one year storm at the site
- Extreme storm; i.e. 9.8m Hs with 25m/s wind (one hour mean at 10m elevation)
- Emergency stop of the turbine in 12m/s wind.

The tests were conducted with 0 and 90 deg wave directions with wind collinear with the waves. An additional series of tests
was performed with the wind direction 30 deg from the wave direction.

Mooring and cable analysis

The proposed mooring system consists of 4 mooring lines composed of chain with a clump weight at the top, rope in the
intermediate section and a section of chain at the bottom attached to a drag-embedment anchor. Two lines are connected to
the column supporting the wind turbine, as it is subject to higher loads than the other two columns.

A medium voltage dynamic electrical cable is daisy-chained from one floating turbine to another one. A lazy-wave
configuration is adopted to reduce strain on the electrical cable.

Fully-coupled time-domain simulations of the platform and tethers are conducted using the OrcaFlex software; Figure 5.
Wamit diffraction-radiation software was run as a preprocessor. Viscous effects on the columns, tubular members and water-
entrapment plates are modeled using specially defined elements with drag coefficients in different directions adjusted based
on model test results.

Figure 5: OrcaFlex model of the WindFloat mooring and electrical cable system

Safety factors were based on API-RP2SK [9]. A 50 year return period for the design storm was selected based on similar
return periods recommended by various offshore wind turbine codes although none of these specifically refer to floating wind
turbines and their mooring system.
OTC 20674-PP 8

Mooring systems of permanent oil and gas platforms are always required to withstand the design storm with any mooring line
broken, albeit with a reduced safety factor. With a four-point mooring system, this requirement is onerous since there is
inherently less redundancy with fewer lines, hence the remaining lines must be made larger.

In the case of a floating wind turbine where the risks to human life or the environment are much lower in case of loss of
position, the requirement to hold position in the design storm assuming one mooring line broken may be excessive.

Historical data related to mooring failures has shown the following trends [10]:
- Most failures have been associated with terminations, fairleads, connectors, pin retention details (i.e. "one off"
design details not standard chain/wire elements).
- Where there have been line failures there has typically been evidence of degradation in all similar components.
- The causes have included poor detailing, fatigue sensitivity, inadequate corrosion protection, inappropriate materials
and lack of adequate manufacturing process.
- Most of the failures first appeared within the first three years of operation

These trends point to the fact that many failures are the results of engineering/fabrication/installation shortcomings (either
due to lack of specification or non-respect of specifications) and are likely to result in early failure of the mooring. Although
these are more likely to occur in adverse weather conditions when the mooring tension is higher, they are associated with
lack of performance of the mooring component and not necessarily correlated with extreme loads (i.e. since these issues
occur due to “defects”, it is likely that a “minor storm” will result in failure). Additionally, assuming such defect is present,
and the platform is subjected to the design storm with a mooring line broken, it is likely that a second mooring line will break
when the load is transferred from the first damaged line. As a result, meeting the required safety factor with one mooring
line damaged during the design storm may not lead to significantly higher reliability level for the mooring system.

It is not suggested that the mooring system does not require to maintain position with a failed mooring line, but rather that the
damaged line case should be applied to a reduced design storm. Quantitative Risk Assessment (QRA) studies are required to
verify and quantify the above argument. The ability to detect damages or failures of mooring lines shortly after they happen,
using instrumentation or more frequent inspection, is a critical element to enhance the reliability of the mooring system.

With the present design, rather than adjusting the mooring to fulfill the damaged line condition, the safety factor achieved
with a broken line was computed during the design storm. The resulting factor was greater than 1.0 although it did not reach
the required 1.25 factor as per API RP2SK. Future work is required to define the reduced storm environment for the
damaged case and corresponding safety factor.

Structural analysis

The WindFloat structural design was performed using a combination of analytical tools and design codes for offshore
structures. Special attention was given to the design of the truss and tower which carry large aerodynamic loads induced by
the wind turbine. The main structural component characteristics and design codes are summarized in Table 1.

Table 1: Design Codes for Main Structural Components

Internal Type of Stiffening Dimension to
Description Type Design Code
stiffening Members determine
Tubular thickness API RP2A – WSD
Truss No
members diameter [11]
thickness
Cylindrical Ring stiffeners ABS Guide [12]
Columns Yes stiffener spacing
Shells Stringers Section 4
stiffener sizing
Water Girders thickness
ABS Guide [12]
Entrapment Stiffened Plates Yes Stringers stiffener spacing
Section 3
Plate Tubular members stiffener sizing
Unstiffened thickness DNV RP C202
Tower No
Cylindrical Shell diameter [13]

Extreme, survival and fatigue loading cases were defined based on combination of wind speed (and turbulence level), status
of the turbine (operating, parked), wave height and period, and ballast condition of the platform. These were primarily based
on the analysis of the 50 year hindcast database (GROW-FAB) and grouping of similar wind and wave regimes in a
simplified wind-wave scatter diagram.
OTC 20674-PP 9

Finite element analysis was performed for the selected design cases using FEA software SAP2000©. The forces and stresses
on the structure were determined for all loading cases using advanced numerical tools and the algorithm presented in
Figure 6.

External loads include the dynamic wind forces and moments generated by the wind turbine and the hydrodynamic,
hydrostatic and inertia loads from wave motions. Local pressure coefficients are obtained from hydrodynamic program
WAMIT©. WAMIT is also used as a pre-processor for a time-domain hydrodynamic solver which computes mooring loads,
wave and viscous loads and the 6 degree of freedom motions for a floater in any given sea-state. To combine the
hydrodynamic and the aerodynamic loads on the WindFloat, in-house program TimeFloat was combined with aerodynamic
program FAST; see next section. The resulting software, FastFloat, generates time-dependent global loads on the structure
for a given set of environmental conditions. An interface was also created to convert these loads into distributed loads on
individual components of the structural model. Although the input loads are identical, the interface is adapted to the finite
element model. For frame analysis, the pressure is integrated on parts of the columns to be applied locally on the frame
element, whereas for shell analysis, the pressure is matched between the mesh from WAMIT and the mesh from the FEA
model.

Figure 6: Flow chart of structural, aerodynamic and hydrodynamic load interface

Table 2: Summary of Finite Element Models

Extent of Type of Structural

Structural Part to Size Type of Analysis
Structural Element for FEM
Model
Truss Global Frame Elements Time-domain linear dynamic analysis
Can at connection on truss for Shell elements on can &
Global Static Analysis on 3-minute events
SCF Frame elements
- Modal analysis
Tower Global Frame Elements
- Time-domain linear dynamic analysis

Shell elements on Column

Column Global Static Analysis on extreme events
1 & Frame elements
Water Entrapment Plate Global Shell elements Static Analysis on extreme events

Slender components, such as the truss and the tower are designed using frame elements, which are based on beam theory in
SAP 2000. The stress on structural details such as the can at the connection and larger components were determined from
geometric models with shell elements when they fall outside the range of recognized semi-empirical models, such as the
Efthymiou formulaes [11]. A dynamic analysis is necessary to account for the effect of the structural modes on the behavior
of the truss and tower. But a static analysis is sufficient to capture the structural strength of the column and water entrapment
plates since they will not be excited by the high-frequency resonance of the tower. Table 2 summarizes the type of finite
element model used to design each component of the WindFloat.
OTC 20674-PP 10

Sample output obtained from the various FEA models are shown below. Figure 7 shows the structural deformation (amplified
50 times) when the design wind force is applied for two different wind directions. The contribution of the substructure
deformations is significant. Figure 8 left shows the Von Mises stress on the platform global FEA model with highly stressed
areas at the bottom of tower and in the upper half of the column supporting the tower. The fine mesh model results of a
tubular node are shown on Figure 8 right.

Figure 7: Deformed Shape of WindFloat (x50) with design wind load - Left: 0 deg heading; Right: 90 deg heading

Figure 8: Von Mises stress on shell (left) tubular joint fine mesh model (right)
OTC 20674-PP 11

Coupling of hydrodynamic and aero-servo-elastic analysis

The importance of considering interactions between turbine aerodynamic loads and platform hydrodynamic loads became
obvious in the early phases of the project. Significant efforts were then placed on developing a coupled analysis tool. To that
effect, the following existing numerical tools were combined:

Marine Innovation & Technology (MI&T) has developed the TimeFloat algorithm since 2003 to compute the motion of
floating vessels. The software has been used on a wide range of projects including all types of floaters and has been
validated against a number of model testing campaigns and other numerical tools. In particular, TimeFloat has been used
extensively to conduct the design of the MiniFloat platform, the precursor of the WindFloat structure. TimeFloat is a time-
domain numerical algorithm that solves the equation of motion of a floater subjected to wind, wave and current forces, and
includes the effects of various mechanical connections, such as mooring lines, fenders, tendons, etc. The program has been
written in FORTRAN.

FAST, which stands for Fatigue, Aerodynamics, Structures and Turbulence is a comprehensive aeroelastic simulator capable
of predicting both the extreme and fatigue loads of two- and three-bladed horizontal axis wind turbines (HAWT). FAST was
developed over many years by the National Renewable Energy Laboratory (NREL). It was originally created by combining
various software including AeroDyn, which provide subroutines to compute the aerodynamic behavior of HAWT. In 2005,
FAST with AeroDyn were evaluated by Germanischer Lloyd WindEnergie and found suitable for the calculation of onshore
wind turbine loads for design and certification. FAST is a FORTRAN open-source software available from NREL.

Access to the source code provides flexibility to test, modify and run numerical algorithms. It was therefore decided to
merge the two computer programs. The resulting package, named FastFloat, combines the full hydrodynamic capability of
TimeFloat with the full aerodynamic and structural capability of FAST.

A feature recently added to FAST which allows the computation of the turbine support platform motion via user-provided
mass, damping, stiffness and external forces on the platform, was key to the integration of the TimeFloat algorithm into
FAST. In essence, TimeFloat solves the same Newton’s equation of motion for the platform including the aerodynamic and
inertial forces due to the wind turbine provided at the base of the tower.

TimeFloat was re-written to fit the format of the FAST user-specified subroutines for platform motion parameters. Minimal
changes were made to FAST.

One of the complications is that the two pieces of software operate on different time-scales. Typical time-steps for FAST are
on the order of 100 time-steps per second; however TimeFloat is normally run at only a few time-steps per second.
TimeFloat must solve the mooring equations at each fractional time-step (in its Runge-Kutta algorithm), which includes
hundreds of finite-difference elements, used to discretize the mooring lines. In addition, the drift force formulation, and
wave-kinematics model for viscous forces is time-consuming, and therefore the program cannot be run efficiently with time-
steps on the order of 0.01 sec. Fortunately, floater responses do not generally require such small time-steps since the typical
frequencies of motions are not very high. The “full” Timefloat algorithm was therefore run only at selected time-steps of the
FAST algorithm and an interpolation scheme was set-up to prevent discontinuities in the external forces at the time-steps for
which a full TimeFloat solution is not obtained.

The turbine controller was modified with a filtering and estimation scheme aimed at suppressing platform pitch resonance
induced by negative damping of the motion due to the blade pitch controller algorithm. The blade pitch control is used to
regulate rotor speed and power in the presence of wind turbulence. This issue is well known and can be mitigated by
complex control algorithms [14]. The present approach is based on an algorithm that estimates the rotor speed variations that
can be attributed to the apparent wind generated by platform pitch motions. The estimator uses the platform pitch
measurement and a high and low-pass filtered (bandpass) speed signal in a feed-back loop. These estimated speed variations
are removed from the rotor speed measurement that is sent to the standard pitch control and instead are sent to an alternative
control that has reduced over all gain. This simpler scheme works well on the WindFloat because of the high damping level
of the platform in pitch generated by viscous effects on the water-entrapment plate. A flow chart showing the modified
control algorithm in relation with the original controller is provided in Figure 9. This result does come at the expense of
slightly higher rotor speed and power variations.

Time-series of the platform motion are shown in Figure 10 based on the original controller and the modified controller. With
the original control algorithm, a steady-state pitch oscillation of approximately 5 deg is obtained; the new algorithm
eliminates the issue completely, and the platform reaches a stable steady state.
OTC 20674-PP 12

Figure 9: modified turbine controller flowchart for suppression of platform pitch resonance

10
PtfmRDyi

-10

-20
0 100 200 300 400 500 600
10
PtfmRDyi

-10

-20
0 100 200 300 400 500 600
Figure 10: time-series of platform pitch (deg) in steady wind (12m/s) with original and modified controller

The FastFloat algorithm was run for the extreme and fatigue design cases. A baseline comparison with an identical turbine
on a fixed base providing the same structural bending modes was established to assess the viability of floater-based wind
turbines. Comparisons are given in Table 3 for selected sea-states based on 20-minute time-domain simulations.
Fatigue results are primarily driven by the root means square (RMS) of the dynamic component of the responses; 20-minute
simulations provide good accuracy of the RMS response.

Table 3 summarizes the wave and wind speed statistics in the mean wind direction for the 8, 12 and 16m/s wind cases. The
wind and wave direction is along the platform port/starboard symmetry. Spectral analysis has been carried out to determine
the RMS in the following frequency ranges:

1. High Frequency (HF) with periods smaller than 4 seconds – there is very limited wave energy for periods shorter
than 4 sec; this frequency range includes the tower and blade resonant modes.
2. Wave Frequency (WF) includes periods between 4 and 25 seconds, which is almost all the wave energy.
3. Low Frequency (LF) encompasses all periods larger than 25 seconds, which includes the pitch/roll and horizontal
periods (surge/sway/yaw).

Wind speed RMS is dominated by low-frequency variations. Platform motion statistics are presented for surge, heave and
pitch. Sway and yaw are very small in these runs since the wind and wave direction is along the port/starboard symmetry
axis. Small roll oscillations (less than 1 degree) are observed, due to rotor-induced moments.

Heave is the vertical motion at the base of the tower. It is driven mainly by low-frequency wind induced pitch. Pitch motion
is also dominated by low-frequencies, even in the large sea-states. This is due to changes in wind speed induced by wind
turbulence. The model was run with a high level of turbulence to be on the conservative side; atmospheric turbulence is a
critical driver of the platform response.

Mean power produced by the floating turbine is over 98% of that produced by the fixed turbine in 8m/s wind, over 99% in
12m/s wind and 100% in 16m/s wind. Power surge beyond the rated power are up to 3.3% higher in the floating case.
OTC 20674-PP 13

There are no noticeable differences in generator torque in the fixed and floating case.

The mean shaft speed is identical in runs which produce rated power. The maximum shaft speed is slightly larger in the
floating cases with spikes that are 3.3% larger than in the fixed case.

The Out-Of-Plane blade deflection reaches a maximum 4% higher in the floating case. The In-Plane deflection is the same
whether the turbine is fixed or floating. The clearance between the tower and the blades will therefore not require
modification of the turbine. Blade pitch dynamics is very similar in the fixed and floating cases.

Tower base loads are significantly higher in the floating case due to the effect of turbulence-induced pitch. The tower
diameter and wall thickness were increased to absorb the additional loading.

Loads on the shaft are also summarized in Table 3. The axial force (Fx) is slightly larger in the floating case, due to the
weight component of the hub and rotor applied because of the tower inclination and its inertia. Shaft forces in the transverse
direction are very small and the ratios are not meaningful. In the vertical direction, the load in the fixed and floating case is
almost identical.

Table 3: Results of coupled aero-hydro analysis for selected sea-states

Run Name ftg03 ftg08 ftg12 Generator ftg03 ftg08 ftg12 Tower Base Loads ftg03 ftg08 ftg12
Power Output Ratio Mx Ratio
Sea-State mean 0.984 0.991 1.000 mean 2.037 1.946 1.997
Hs (m) 1.5 2.5 4 max 0.920 1.027 1.033 max 1.172 1.424 1.302
Tp (sec) 7 9 11 rms 0.931 1.037 1.620 rms 1.478 0.851 0.859
gamma 2.4 2.4 2.4 rms_HF 0.970 1.219 1.096 rms_HF 0.748 0.530 0.523
rms_WF 0.965 1.089 1.342 rms_WF 0.545 0.583 0.620
Wind speed rms_LF 0.918 1.023 1.998 rms_LF 2.225 2.630 2.687
Vx (m/s) Torque Ratio My Ratio
mean 7.8 12.4 16.6 mean 0.992 0.997 1.000 mean 0.876 0.820 0.981
max 13.0 18.0 24.4 max 0.947 1.000 1.000 max 1.704 1.381 1.564
rms 1.79 1.97 2.58 rms 0.944 1.003 1.000 rms 2.219 2.173 1.921
rms_HF 0.38 0.55 0.81 rms_HF 0.973 1.250 1.000 rms_HF 1.195 0.967 0.903
rms_WF 0.60 0.83 1.26 rms_WF 0.972 0.996 1.000 rms_WF 3.710 2.182 1.949
rms_LF 1.42 1.57 1.95 rms_LF 0.933 0.989 1.000 rms_LF 2.306 2.341 2.396
Shaft Speed Ratio Mz Ratio
Platform motion ftg03 ftg08 ftg12 mean 0.996 0.994 1.000 mean 1.148 1.485 1.718
Surge (m) max 0.972 1.027 1.033 max 0.939 1.017 1.109
mean 3.0 5.5 4.2 rms 0.945 1.084 1.625 rms 0.952 0.976 0.983
max 6.5 8.3 8.0 rms_HF 0.988 1.161 1.095 rms_HF 0.949 0.991 0.990
rms 1.46 1.17 0.92 rms_WF 0.985 1.205 1.342 rms_WF 0.988 1.012 1.009
rms_HF 0.02 0.00 0.02 rms_LF 0.933 1.080 1.998 rms_LF 0.946 0.954 0.969
rms_WF 0.10 0.23 0.51
rms_LF 1.24 1.11 0.71 Blade#1 ftg03 ftg08 ftg12 Shaft Loads ftg03 ftg08 ftg12
Heave (m) OOP Deflection Ratio Fx Ratio
mean 0.0 0.5 0.0 mean 0.950 0.971 1.004 mean 0.959 0.952 0.999
max 2.3 2.1 1.6 max 0.928 0.991 1.040 max 1.089 1.056 1.253
rms 0.91 0.79 0.62 rms 0.955 0.990 1.030 rms 1.236 1.247 1.301
rms_HF 0.01 0.00 0.01 rms_HF 0.940 1.019 0.998 rms_HF 1.012 1.037 1.024
rms_WF 0.04 0.14 0.30 rms_WF 0.958 1.308 1.168 rms_WF 1.248 1.523 1.400
rms_LF 0.79 0.76 0.50 rms_LF 0.964 0.939 1.021 rms_LF 1.254 1.235 1.298
Pitch (deg) IP Deflection Ratio Fy Ratio
mean -0.1 -1.1 -0.1 mean 1.000 1.002 1.001 mean 3.015 4.062 2.260
max 5.1 2.8 5.0 max 1.028 1.073 1.016 max 0.638 0.766 0.599
rms 1.95 1.68 1.23 rms 0.997 1.007 1.004 rms 1.036 0.980 0.988
rms_HF 0.02 0.01 0.01 rms_HF 0.916 1.006 1.004 rms_HF 0.905 0.884 0.862
rms_WF 0.09 0.25 0.41 rms_WF 1.000 1.021 0.987 rms_WF 0.711 0.828 0.858
rms_LF 1.68 1.61 1.07 rms_LF 0.933 0.886 1.014 rms_LF 1.327 1.293 1.378
Blade Pitch Ratio Fz Ratio
mean 1.421 1.064 0.995 mean 0.999 1.002 1.000
max 1.350 0.996 1.010 max 0.989 1.007 0.996
rms 1.343 0.945 1.055 rms 1.038 0.991 1.026
rms_HF 1.735 1.173 1.073 rms_HF 0.988 0.952 1.001
rms_WF 1.414 1.196 1.092 rms_WF 1.641 1.679 1.688
rms_LF 1.418 0.951 1.049 rms_LF 1.214 1.240 0.941
OTC 20674-PP 14

Conclusion

Key ingredients of the WindFloat qualification process have been described in this paper, with specific elements related to its
function as a floating foundation for multi-megawatt wind turbines. A combination of design codes and standards used for
floating oil and gas platforms and fixed offshore wind turbines has been followed.

The economics of wind turbines offshore cannot afford un-necessary conservatism; therefore a comprehensive qualification
process has been adopted to optimize the system using advanced numerical tools and rational approaches, while maintaining
the system robustness required to survive the offshore environment. A design has been developed for a WindFloat platform
carrying a 5MW wind turbine for deployment in a 150MW wind farm offshore the West coast of Portugal.

Existing rules for the design of Mobile Offshore Drilling Units (MODU) have been followed to ensure intact and damaged
stability of the system.

The hydrodynamic response has been validated by using two different time-domain algorithms as well as scale model tests.

The design of the mooring and electrical cable has been based on results of fully-coupled analysis. A four-point mooring
configuration has been selected. Offshore mooring codes used in the oil and gas industry have been used as a basis for the
design; however issues associated with the redundancy of the system in case of a damaged mooring line have been
highlighted. Adjustment of the design criteria associated with the damaged mooring case is recommended based on overall
system reliability considerations.

Extensive finite-element analyses have been carried out for the structural design of the platform. An interface with advanced
aero-hydrodynamic tools has been developed to determine loading on the various parts of the structure.

Because of the large aerodynamic loads generated by the wind turbine, it became apparent that a coupled algorithm, able to
simultaneously solve the aero-servo-elastic problem of the wind turbine, tower and control system as well as the platform
hydrodynamic response was required. An existing computer program for design of wind turbines was then coupled with a
hydrodynamic software developed to predict the motion of floating structures. Coupled analyses have shown that the
installation of an existing wind turbine on such floater is viable provided the tower design is adjusted accordingly.

Acknowledgments

The authors wish to thank Principal Power Inc for permission to publish this work. The National Renewable Energy
Laboratory (NREL) contribution to this work by providing the FAST algorithm is gratefully acknowledged.

References

[1] Sclavounos PD and Wayman EN (MIT), Butterfield S, Jonkman J and Musial W (NREL) “Floating Wind Turbine Concepts”
European Wind Energy Association Conference (EWAC), Athens, Greece, 2006.
[2] Jensen JJ and Mansour AE "Extreme Motion Prediction for Deepwater TLP Floaters For Offshore Wind Turbines", Proc of the 4th Intl.
HydroElasticity Conf., 2006.
[3] Nielsen FG, Hanson TD, and Skaare B, “Integrated Dynamic Analysis of Floating Offshore Wind Turbines” Proceedings of
OMAE2006 25th International Conference on Offshore Mechanics and Arctic Engineering, Hamburg, Germany, 2006.
[4] Germanischer Lloyd, “Guideline for the Certification of Offshore Wind Turbines”, Rules and Guidelines, 2005.
[5] Det Norske Veritas (DNV), “Design of offshore wind turbine structures”, Offshore Standard, DNV-OS-J101, Oslo, Norway, 1st
Edition, 2004.
[6] Jonkman JM and Sclavounos PD, “Development of Fully Coupled Aeroelastic and Hydrodynamic Models for Offshore Wind Turbines,”
44th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 2006.
[7] American Bureau of Shipping, “Rules for Building and Classing Mobile Offshore Drilling Units”, Part 3, Hull Construction and
Equipment, 2001.
[8] Cermelli CA and Roddier DG, “Experimental and Numerical Investigation of the Stabilizing Effects of a Water-Entrapment Plate on a
Deepwater Minimal Floating Platform”, Proc. 24th International Conference on offshore Mechanics and Arctic Engineering, Halkidiki,
Greece, 2005.
[9] The American Petroleum Institute “Design and Analysis of Stationkeeping Systems for Floating Structures” – API RP 2SK Third
Edition – 2007.
OTC 20674-PP 15

[10] Morandini C and Legerstee F, “Consistent Integrity of Mooring System”, Proceedings of the Nineteenth (2009) International Offshore
and Polar Engineering Conference Osaka, Japan, 2009.
[11] The American Petroleum Institute, “Recommended Practice for Planning and Constructing Fixed Offshore Platforms – Working
Stress Design”, 2000.
[12] American Bureau of Shipping, “Guide for Buckling and Ultimate Strength Assessment for Offshore Structures”, 2008.
[13] Det Norske Veritas (DNV), “Buckling Strength of Shells” Recommended Practice DNV RP-C202, 2002.
[14] Larsen TJ and Hanson TD “A method to avoid negative damped low frequent tower vibrations for a floating, pitch controlled wind
turbine” - Journal of Physics: Conference Series 75, 2007.