Guidelines For Design of Wind Turbines Copenhagen PDF
Guidelines For Design of Wind Turbines Copenhagen PDF
Guidelines For Design of Wind Turbines Copenhagen PDF
for
Design
of
Wind Turbines
A publication from
DNV/Ris
Second Edition
Guidelines for Design of Wind Turbines
2nd Edition
Det Norske Veritas, Copenhagen ([email protected]) and Wind
Energy Department, Ris National Laboratory ([email protected]) 2002.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means, electronical, mechanical, photocopying,
recording and/or otherwise without the prior written permission of the publishers.
This book may not be lent, resold, hired out or otherwise disposed of by way of trade in any
form of binding or cover other than that in which it is published, without the prior consent of
the publishers.
ISBN 87-550-2870-5
Guidelines for Design of Wind Turbines DNV/Ris
Preface i
Guidelines for Design of Wind Turbines DNV/Ris
Contents
ii Contents
Guidelines for Design of Wind Turbines DNV/Ris
Contents iii
Guidelines for Design of Wind Turbines DNV/Ris
iv Contents
Guidelines for Design of Wind Turbines DNV/Ris
Contents v
Guidelines for Design of Wind Turbines DNV/Ris
vi Contents
Guidelines for Design of Wind Turbines DNV/Ris
much smaller solidity. To compensate for the one illustrated in Figure 1-2, are that the
the slenderness of the blades, modern generator and gearbox are placed on the
turbines operate at high tip speeds. ground and are thus easily accessible, and
that no yaw mechanism is needed. Among
1.2.1 Vertical axis turbines the disadvantages are an overall much lower
level of efficiency, the fact that the turbine
Vertical axis wind turbines (VAWTs), such
needs total dismantling just to replace the
as the one shown in Figure 1-2 with C-
main bearing, and that the rotor is placed
shaped blades, are among the types of
relatively close to the ground where there is
turbine that have seen the light of day in the
not much wind.
past century.
1.2.2 Horizontal axis turbines
Horizontal axis wind turbines (HAWTs),
such as the ones shown in Figure 1-3,
constitute the most common type of wind
turbine in use today. In fact all grid-
connected commercial wind turbines are
today designed with propeller-type rotors
mounted on a horizontal axis on top of a
vertical tower. In contrast to the mode of
operation of the vertical axis turbines, the
horizontal axis turbines need to be aligned
with the direction of the wind, thereby
allowing the wind to flow parallel to the axis
of rotation.
aligned with the direction of the wind. more fluctuating loads because of the
Downwind rotors are placed on the lee side variation of the inertia, depending on the
of the tower. A great disadvantage in this blades being in horizontal or vertical
design is the fluctuations in the wind power position and on the variation of wind speed
due to the rotor passing through the wind when the blade is pointing upward and
shade of the tower which gives rise to more downward. Therefore, the two and one-
fatigue loads. Theoretically, downwind bladed concepts usually have so-called
rotors can be built without a yaw teetering hubs, implying that they have the
mechanism, provided that the rotor and rotor hinged to the main shaft. This design
nacelle can be designed in such a way that allows the rotor to teeter in order to
the nacelle will follow the wind passively. eliminate some of the unbalanced loads.
This may, however, induce gyroscopic loads One-bladed wind turbines are less
and hamper the possibility of unwinding the widespread than two-bladed turbines. This is
cables when the rotor has been yawing due to the fact that they, in addition to a
passively in the same direction for a long higher rotational speed, more noise and
time, thereby causing the power cables to visual intrusion problems, need a counter-
twist. As regards large wind turbines, it is weight to balance the rotor blade.
rather difficult to use slip rings or
mechanical collectors to circumvent this
problem. Whereas, upwind rotors need to be
rather inflexible to keep the rotor blades
clear of the tower, downwind rotors can be
made more flexible. The latter implies
possible savings with respect to weight and
may contribute to reducing the loads on the
tower. The vast majority of wind turbines in
Figure 1-4. Three, two and one-bladed wind turbine
operation today have upwind rotors. concepts. From www.windpower.org (2000), Danish
Wind Turbine Manufacturers Association.
1.2.3 Number of rotor blades
The three-bladed concept is the most 1.2.4 Power control aspects
common concept for modern wind turbines. Wind turbines are designed to produce
A turbine with an upwind rotor, an electricity as cheap as possible. For this
asynchronous generator and an active yaw purpose, wind turbines are in general
system is usually referred to as the Danish designed to yield a maximum power output
concept. This is a concept, which tends to be at wind speeds around 15 m/s. It would not
a standard against which other concepts are pay to design turbines to maximise their
evaluated. power output at stronger winds, because
such strong winds are usually too rare.
Relative to the three-bladed concept, the two However, in case of stronger winds, it is
and one-bladed concepts have the advantage necessary to waste part of the excess energy
of representing a possible saving in relation to avoid damage on the wind turbine. Thus,
to the cost and weight of the rotor. However, the wind turbine needs some sort of power
their use of fewer rotor blades implies that a control. The power control is divided into
higher rotational speed or a larger chord is two regimes with different concepts:
needed to yield the same energy output as a power optimisation for low wind speeds
three-bladed turbine of a similar size. The power limitation for high wind speeds
use of one or two blades will also result in
These regimes are separated by the wind rated power of the generator at high wind
speed at which the maximum power output speeds. Disadvantages encompass extra
is achieved, typically about 15 m/sec. complexity due to the pitch mechanism and
high power fluctuations at high wind speeds.
Basically, there are three approaches to
power control: Active stall-controlled turbines resemble
stall control pitch-controlled turbines by having pitchable
pitch control blades. At low wind speeds, active stall
active stall control turbines will operate like pitch-controlled
turbines. At high wind speeds, they will
Stall-controlled wind turbines have their pitch the blades in the opposite direction of
rotor blades bolted to the hub at a fixed what a pitch-controlled turbine would do
angle. The stall phenomenon is used to limit and force the blades into stall. This enables a
the power output when the wind speed rather accurate control of the power output,
becomes too high. This is achieved by and makes it possible to run the turbine at
designing the geometry of the rotor blade in the rated power at all high wind speeds. This
such a way that flow separation is created on control type has the advantage of having the
the downwind side of the blade when the ability to compensate for the variations in
wind speed exceeds some chosen critical the air density.
value. Stall control of wind turbines requires
correct trimming of the rotor blades and Figure 1-5 shows iso-power curves for a
correct setting of the blade angle relative to wind turbine as a function of the blade angle
the rotor plane. Some drawbacks of this and the mean wind speed. The ranges for
method are: lower efficiency at low wind pitch control and active stall control are
speeds, no assisted start and variations in the separated at a blade angle of 0 with the
maximum steady state power due to rotor plane. At low wind speeds, the optimal
variation in the air density and grid operation of the wind turbine is achieved at
frequencies. a blade angle close to 0. At higher wind
speeds, the turbine will overproduce if the
Pitch-controlled wind turbines have blades blade angle is not adjusted accordingly.
that can be pitched out of the wind to an With pitch control, the blade is pitched
angle where the blade chord is parallel to the positively with its leading edge being turned
wind direction. The power output is towards the wind. With active stall control,
monitored and whenever it becomes too the blade is pitched negatively with its
high, the blades will be pitched slightly out trailing edge turned towards the wind. The
of the wind to reduce the produced power. power control and, in particular, the power
The blades will be pitched back again once limitation at higher wind speeds are
the wind speed drops. Pitch control of wind indicated in an idealised manner for both
turbines requires a design that ensures that control approaches by the dashed curves in
the blades are pitched at the exact angle Figure 1-5. The dashed curves illustrate how
required in order to optimise the power the transition between operation with 0
output at all wind speeds. Nowadays, pitch blade angle at low wind speeds and power-
control of wind turbines is only used in limiting operation along an iso-power curve
conjunction with variable rotor speed. An at high wind speeds can be achieved for a
advantage of this type of control is that it three-bladed rotor at a rated power of
has a good power control, i.e. that the mean 400 kW. In this example, the rated power is
value of the power output is kept close to the reached at a wind speed of about 12 m/sec.
1.3 Economical aspects The choice of rotor size and generator size
depends heavily on the distribution of the
The ideal wind turbine design is not dictated
wind speed and the wind energy potential at
by technology alone, but by a combination
a prospective location. A large rotor fitted
of technology and economy. Wind turbine
with a small generator will produce
manufacturers wish to optimise their
electricity during many hours of the year,
machines, so that they deliver electricity at
but it will only capture a small part of the
the lowest possible cost per unit of energy.
wind energy potential. A large generator
In this context, it is not necessarily optimal
will be very efficient at high wind speeds,
to maximise the annual energy production, if
but inefficient at low wind speeds.
that would require a very expensive wind
Sometimes it will be beneficial to fit a wind
turbine. Since the energy input (the wind) is
turbine with two generators with different
free, the optimal turbine design is one with
rated powers.
low production costs per produced kWh.
Figure 1-5. Iso-power curves for a wind turbine at 26.88 rpm vs. blade angle and mean wind speed
Efficiency [%]
Power [kW]
Thus the primary factor affecting the power
performance is the rotor size. Secondary
factors are the control principle such as stall-
or pitch control and single- dual- or variable
speed.
Wind speed
The power being produced by any type of The produced power varies with the wind
wind turbine can be expressed as speed as can be seen from the blue graph in
Figure 1-6. The form of the graph varies
P = V3ACP slightly from different concepts. Assuming
constant efficiency (e.g. constant tip speed
P output power ratio) the graph basically consists of a third
air density degree polynomial up to the rated wind
V free wind speed speed at which the nominal power is
A rotor area reached. At this point the power regulation
CP efficiency factor sets in, either by the blades stalling or by
pitching the blades to attain an
The power coefficient CP is a product of the approximately constant power. The power
mechanical efficiency m, the electrical curve and the power efficiency curve are
efficiency e, and of the aerodynamic often presented in the same graph, with the
efficiency. All three factors are dependent power and the efficiency scales on each side
on the wind speed and the produced power, of the graph as shown in Figure 1-6.
respectively. The mechanical efficiency m
is mainly determined by losses in the Figure 1-7 illustrates the controlled power
gearbox and is typically 0.95 to 0.97 at full curve of a wind turbine, in the case of 1)
load. The electrical efficiency covers losses stall controlled, fixed speed configuration,
in the generator and electrical circuits. At and 2) pitch controlled, variable speed
full load e = 0.97 - 0.98 is common for configuration.
configurations with an induction generator.
It can be shown that the maximum possible Power Rated power
[Kw]
value of the aerodynamic efficiency is
16/27 = 0.59, which is achieved when the
turbine reduces the wind speed to one-third Wind power resource
Stall regulation
regulated &&fixed
fixedspeed
speed
of the free wind speed (Betz' law). Pitch control & variable speed
Wind speed
5 10 15 20 25 [m/s]
The efficiency factor CP typically reaches a For a particular wind turbine, the power
maximum at a wind speed of 7-9 m/sec and, production is defined by the power curve as
normally, it does not exceed 50%. The described above. By combining the power
electric power typically reaches the rated curve with the wind distribution the actual
power of the turbine at a wind speed of 14- energy production is yielded, often
16 m/sec. expressed in terms of the annual energy
production Eyear.
In the Danish approval scheme as well as in
the IEC wind turbine classification system, Vstop
the power curve is required to be determined
from measurements.
E year = N 0 P(u) f (u) du
Vstart
A(1 + 1 / k )
Vave =
A / 2, if k = 2 1.5 Configurations and sizes
Wind turbines are erected as stand-alone
in which is the gamma function. turbines, in clusters of multiple turbines, or
on a larger scale in park configurations.
In the Danish standard it is common to Before the 1980s, wind energy
describe the wind climate in terms of the development focused on the individual wind
roughness class, and an explicit correlation turbine. By the late 1980s, this perspective
exists between the roughness class and the began to change as attention shifted to
average wind speed. collective generation of electric power from
an array of wind turbines located in the high population density and thus with
vicinity of each other and commonly difficulties in finding suitable sites on land.
referred to as wind parks or wind farms. In Construction costs are much higher at sea,
the early 1980s, the typical size of a wind but energy production is also much higher.
turbine was about 55 kW in terms of rated Currently, wind energy from turbines
power, whereas turbine sizes today have erected on fixed foundations in up to 15 m
exceeded 2 MW. Table 1-1 gives examples water depth is considered economically
of typical combinations of rotor diameter feasible. Figure 1-9 shows an example of an
and rated power for a number of different early offshore wind farm.
tower heights.
REFERENCES
they are onshore, and that more emphasis consider a 20-year design lifetime for wind
will eventually be placed on financial factors turbines. Characteristic capacities are
when it comes to safety issues and usually chosen as low quantiles in the
determination of acceptable safety levels. associated capacity distributions. The partial
Limit state design is used to achieve the safety factors that are applied in the design
prescribed level of safety. It is common to account for the possible more unfavourable
verify the safety of a wind turbine with realisation of the loads and capacities than
respect to the following limit states: those assumed by the choice of
ultimate limit state characteristic values. Note in this context
serviceability limit state that some of the partial safety factors, which
accidental limit state are specified in standards, are not safety
For this purpose, design loads are derived factors in the true sense, but rather reduction
from multiplying characteristic loads by one factors which account for degradation
or more partial safety factors, and design effects, scale effects, temperature effects,
capacities are derived from dividing etc. and which happen to appear in the
characteristic capacities by one or several design expressions in exactly the same
other partial safety factors. Partial safety manner as true partial safety factors.
factors are applied to loads and material
strengths to account for uncertainties in the With structural safety being a major goal of
characteristic values. Verification of the the design, it is important to make sure that
structural safety is achieved by ensuring that the characteristic values of load and material
the design load, or the combination of a set quantities, which have been assumed for the
of design loads, does not exceed the design design, are achieved in practice. Non-
capacity. In case a combination of loads is destructive testing of completed structural
used, it should be noted that it is common parts plays a role in this context, and control
always to combine one extreme load with of workmanship another. Material
one or several normal loads. Two or certificates also come in handy in this
several extreme loads are usually not context. In general, one may say that
combined, unless they have some inspection is an important part of the safety
correlation. philosophy. Not least, as it will allow for
verification of assumptions made during the
Characteristic loads and characteristic design and for taking remedial actions if
capacities constitute important parameters in averse conditions are detected during the
the design process. Characteristic loads for service life of the wind turbine.
assessment of the ultimate limit state are
usually determined as load values with a 50- For details about structural safety and limit
year recurrence period, and they are state design, reference is made to Section
therefore often interpreted as the 98% 2.3. For details about combinations of
quantile in the distribution of the annual design situations and external conditions, as
maximum load. This choice does not well as a definition of load cases, reference
necessarily imply that a design lifetime of is made to Chapter 4.
exactly 50 years is considered. It is more a
matter of tradition and convenience. Nor
should it be taken as a 50-year guarantee 2.2 System safety and operational
within which failures will not occur. For reliability
assessment of fatigue, a design lifetime is A wind turbine is to be equipped with
needed, and in this context it is common to control and protection systems which are
meant to govern the safe operation of the voltage and frequency at mains connec-
wind turbine and to protect the wind turbine tion
from ill conditions. Some components will connection of the electrical load
act in both the control and the protection power output
function, but distinction is made in that the cable twist
control system monitors and regulates the yaw error
essential operating parameters to keep the brake wear
turbine within defined operating range
whereas the protection system ensures that The control system is meant to keep the
the turbine is kept within the design limits. wind turbine within its normal operating
The protection system must takes range. As a minimum, the normal operating
precedence over the control system. range should be characterised by the
following properties and requirements:
2.2.1 Control system a maximum 10-minute mean wind
Controls are used for the following speed at hub height, Vmax, i.e. the stop
functions: wind speed below which the wind
to enable automatic operation turbine may be in operation
to keep the turbine in alignment with a maximum long-term mean nominal
the wind power Pnom, interpreted as the highest
to engage and disengage the generator power on the power curve of the wind
to govern the rotor speed turbine in the wind speed interval
to protect the turbine from overspeed or [Vmin;Vmax], where Vmin denotes the start
damage caused by very strong winds wind speed for the turbine
to sense malfunctions and warn a maximum nominal power Pmax, which
operators of the need for maintenance on average over 10 minutes may not be
or repair exceeded for a wind speed at hub height
of V10min,hub < Vmax
The control system is meant to control the a maximum operating frequency of
operation of the wind turbine by active or rotation nr,max for the wind turbine
passive means and to keep operating a maximum transient frequency of
parameters within their normal limits. rotation nmax for the wind turbine
Passive controls use their own sensing and a wind speed below which the wind
are exercised by use of natural forces, e.g. turbine may be stopped
centrifugal stalling or centrifugal feathering. The wind turbine is kept within its normal
Active controls use electrical, mechanical, operating range by means of the control
hydraulic or pneumatic means and require system, which activates and/or deactivates
transducers to sense the variables that will the necessary controls, e.g.:
determine the control action needed. Typical yaw (alignment with the wind)
variables and features to be monitored in blade angle regulation
this respect include: activation of the brake system
rotor speed power network connection
wind speed power limitation
vibration shutdown at loss of electrical network
external temperature or electrical load
generator temperature
In addition, it must be possible to stop the turbine to a safe condition and maintain the
wind turbine, e.g. for the purpose of turbine in this condition. It is usually
inspection and repairs, or in case of required that the protection system shall be
emergencies. Monitoring of the control capable of bringing the rotor to rest or to an
system and its functions must be adapted to idling state from any operating condition. In
the actual design of the wind turbine. Design the IEC standard an additional requirement
of a wind turbine control system requires a is that means shall be provided for bringing
background in servo theory, i.e. theory the rotor to a complete stop from a
dealing with control of continuous systems. hazardous idling state in any wind speed less
than the annual extreme wind speed. The
The control system is of particular activation levels for the protection system
importance in areas where weak grids are have to be set in such a way that design
encountered. Weak grids can, for example, limits are not exceeded.
be found in sparsely populated areas where
the capacity of the grids can often be a Situations which call for activation of the
limiting factor for the exploitation of the protection system include, but are not
wind resource in question. Two problems necessarily limited to:
are identified in this context: overspeed
increase of the steady-state voltage level generator overload or fault
of the grid above the limit where power excessive vibration
consumption is low and wind power failure to shut down following network
input is high loss, disconnection from the network, or
voltage fluctuations above the flicker loss of electrical load
limit may result from fluctuating wind abnormal cable twist owing to nacelle
power input caused by fluctuating wind rotation by yawing
and wind turbine cut-ins. The protection system should therefore as a
The solution to the above problems is to use minimum cover monitoring of the
a so-called power control as part of the following:
control system. The power control concept
rotational speed or rotational frequency
implies buffering the wind turbine power in
periods where the voltage limits may be overload of a generator or other energy
violated and releasing it when the voltage is conversion system/load
lower. This method is combined with a extreme vibrations in the nacelle
smoothening of the power output, such that safety-related functioning of the control
fluctuations are removed, in particular those system
that would exceed the flicker limit. As overspeed is by far the most critical
error, rotational speed monitors form a
2.2.2 Protection system crucial element of the protection system.
standard at least two brake systems must be which risk assessment is used to determine
included, and at least one of these must have the interval between inspections.
an aerodynamically operated brake unit. See
Section 2.3.3. Fail-safe is a design philosophy, which
through redundancy or adequacy of design
To ensure immediate machine shutdown in in the structure, ensures that in the event of a
case of personal risk, an emergency stop failure of a component or power source, the
button, which will overrule both the control wind turbine will remain in a non-hazardous
and normal protection system, shall be condition.
provided at all work places.
2.2.3 Brake system
In addition to what is stated above, the
The brake system is the active part of the
protection system is, as a minimum, to be
protection system. Examples of brake
subjected to the following requirements:
systems are:
the protection system must take
mechanical brake
precedence over the control system
aerodynamic brake
the protection system must be fail-safe generator brake
in the event that the power supply fails An aerodynamic brake system usually
structural components in mechanisms of consists of turning the blade tip or, as
the protection system shall be designed commonly seen on active stall- and pitch-
to high safety class controlled turbines, of turning the entire
the protection system must be able to blade 90 about the longitudinal axis of the
register a fault and to bring the wind blade. This results in aerodynamic forces
turbine to a standstill or to controlled that counteract the rotor torque. Also,
freewheeling in all situations in which spoilers and parachutes have been used as
the rotor speed is less than nmax. aerodynamic brakes.
the protection system must be tolerant
towards a single fault in a sensor, in the
electronic and electrical as well as the
hydraulic systems or in active Figure 2-1. Tip brake, from www.windpower.org
mechanical devices, i.e. an undetected (2000), Danish Wind Turbine Manufacturers'
Association.
fault in the system must not prevent the
system from detecting a fault condition
The reliability of a brake system is of the
and carrying out its function
utmost importance to ensure that the system
the reliability of the protection system will serve its purpose adequately. In this
must ensure that situations caused by respect, it is important to be aware of
failures in the protection system, possible dependencies between different
whereby the extreme operating range is brakes or different brake components. For
exceeded, can be neglected example, if all three blades are equipped
The reliability of the protection system may with tip brakes, some dependency between
be ensured by means of either (1) the entire the three tip brakes can be expected, cf. the
protection system being of a fail-safe design, common cause failures that can be foreseen
or (2) redundancy of the parts of the for these brakes. This will influence the
protection system where it cannot be made overall reliability against failure of the
fail-safe, or (3) frequent inspections of the system of the three tip brakes and needs to
functioning of the protection system, in
be taken into account if the failure of the system for determination of possible
probability exceeds 0.0002 per year. failure modes and identification of their
effects on the system. The analysis is based
Brakes or components of brake systems will on a worksheet that systematically lists all
be subject to wear. Thus, current monitoring components in the system, including:
and maintenance are required. component name
function of component
IEC61400-1 requires that the protection possible failure modes
system shall include one or more systems, causes of failure
i.e. mechanical, electrical, or aerodynamic how failures are detected
brakes, capable of bringing the rotor to rest effects of failure on primary system
or to an idling state from any operating function
condition. At least one of these systems shall
effects of failure on other components
act on the low-speed shaft or on the rotor of
necessary preventative/repair measures
the wind turbine. The idea behind this is to
The failure mode and effects analysis can be
have a brake system which ensures that a
supplemented by a criticality analysis, which
fault will not lead to a complete failure of
is a procedure that rates the failure modes
the wind turbine.
according to their frequency or probability
of occurrence and according to their
DS472 is more strict by requiring at least
consequences. The assigned ratings can be
two fail-safe brake systems. If the two
used to rank the components with respect to
systems are not independent, i.e. if they have
their criticality for the safety of the system.
some parts in common, then the turbine shall
An example of a worksheet is given in Table
automatically be brought to a complete stop
2-1.
or to controlled idling in the event of a
failure in the common parts. At least one
A failure mode and effects analysis can be
brake system is required to have an
conducted at various levels. Before
aerodynamic brake unit.
commencing, it is thus important to decide
what level should be adopted as some areas
2.2.4 Failure mode and effects may otherwise be examined in great detail,
analysis while others will be examined at the system
A failure mode and effects analysis is a level only without examination of the
qualitative reliability technique for individual components. If conducted at too
systematic analysis of mechanical or detailed a level, the analysis can be rather
electrical systems, such as a wind turbine time-consuming and tedious, but will
protection system. The analysis includes undoubtedly lead to a thorough under-
examination of each individual component standing of the system.
The failure mode and effects analysis is at the next level. If one of several events
primarily a risk management tool. The causes the higher event, it is joined with an
strength of the failure mode and effects OR gate. If two or more events must occur
analysis is that, if carried out correctly, it in combination, they are joined with an
identifies safety-critical components where a AND gate.
single failure will be critical for the entire
system. It is a weakness, however, that it
depends on the experience of the analyst and
that it cannot easily be applied to cover
multiple failures.
Application
Fault tree analysis has several potential uses
in relation to wind turbine protection
systems:
In frequency analysis, it is common to
quantify the probability of the top event
occurring based on estimates of the
failure rates of each component. The top
event may comprise an individual Figure 2-2. Fault tree symbols.
failure case, or a branch probability in
an event tree. If quantification of the fault tree is the
In risk presentation, it may also be used objective, downward development should
to show how the various risk stop once all branches have been reduced to
contributors combine to produce the events that can be quantified in terms of
overall risk. probabilities or frequencies of occurrence.
In hazard identification, it may be used Various standards for symbols are used
qualitatively to identify combinations of the most typical ones are shown in Figure
basic events that are sufficient to cause 2-2. An example of a fault tree is shown in
the top event, also known as cut sets. Figure 2-3.
protection system. Furthermore, it can be how likely it is that the structure will reach a
used to determine the most critical events limit state and enter a state of failure.
and parts of the protection system.
There are several types of limit states. Two
Beware, however, that it is often hard to find types are common, ultimate limit states and
out whether a fault tree analysis has been serviceability limit states. Ultimate limit
carried out properly. Fault tree analyses states correspond to the limit of the load-
become complicated, time-consuming and carrying capacity of a structure or structural
difficult to follow for large systems, and it component, e.g. plastic yield, brittle fracture,
becomes easy to overlook failure modes and fatigue fracture, instability, buckling, and
common cause failures. It is a weakness that overturning. Serviceability limit states imply
the diagrammatic format discourages that there are deformations in excess of
analysts from stating assumptions and tolerance without exceeding the load-
conditional probabilities for each gate carrying capacity. Examples are cracks,
explicitly. This can be overcome by careful wear, corrosion, permanent deflections and
back-up text documentation. It is a vibrations. Fatigue is sometimes treated as a
limitation that all events are assumed to be separate type of limit state. Other types of
independent. Fault tree analyses lose their limit states are possible, e.g. accidental limit
clarity when applied to systems that do not states and progressive limit states.
fall into simple failed or working states such
as human error, adverse weather conditions, 2.3.2 Failure probability and other
etc. measures of structural reliability
In structural design, the reliability of a
structural component is evaluated with
2.3 Structural safety
respect to one or more failure modes. One
2.3.1 Limit states such failure mode is assumed in the
following. The structural component is
During the lifetime of a structure, the
described by a set of stochastic basic
structure is subjected to loads or actions.
variables grouped into one vector X,
The loads may cause a change of the
including, e.g. its strength, stiffness,
condition or state of the structure from an
geometry, and loading. Each of these
undamaged or intact state to a state of
variables are stochastic in the sense that they
deterioration, damage, or failure. Structural
owing to natural variability and other
malfunction can occur in a number of modes
covering all failure possibilities that can be possible uncertainties may take on random
imagined for the structure. Although the degrees of realisation according to some
transition from an intact state to a state of probability distribution. For the considered
malfunction can indeed be continuous, it is failure mode, the possible realisation of X
common to assume that all states with can be separated into two sets:1) the set for
respect to a particular mode of malfunction which the structural component will be safe,
can be divided into two sets: 1) states that and 2) the set for which it will fail. The
have failed, and 2) states that have not failed surface between the safe set and the failure
or are safe. The boundary between the safe set in the space of basic variables is denoted
states and the failed states is referred to as the limit state surface, and the reliability
the set of limit states. The safety or problem is conveniently described by a so-
reliability of a structure is concerned with called limit state function g(X), which is
defined such that
For the simple example that X consists of A useful by-product of a structural reliability
two variables, the load L and the resistance analysis by these methods is the so-called
R, and the limit state function can be design point x*. This is a point on the limit
specified as g(X) = RL, the failure state surface and is the most likely
probability becomes a simple convolution realisation of the stochastic variables X at
integral failure.
rules in a design code is known as the code The characteristic values are usually taken
format. as specific quantiles in the load and
resistance distributions, respectively. The
The code format most frequently used in characteristic values are often the mean
design codes today is a format which is value for dead load, the 98% quantile in the
expressed in terms of design values of distribution of the annual maxima for
governing load and resistance variables. variable (environmental) load, and the 2% or
These design values are defined as 5% quantile for strength. Reference is made
characteristic values of the load and to Figure 2-4. The code also specifies values
resistance variables, factored by partial to be used for the partial safety factors f and
safety factors. Such a form of code format m .
results from requirements for an easy and
yet economical design and is known as a
design value format. Design according to a fL, fR
design value format is sometimes referred to
as load and resistance factor design (LRFD).
L R
In its simplest form, a code requirement can
be expressed as a design rule in terms of an L, R
inequality
LC RC
LD < RD
Figure 2-4. Probability density functions and
in which LD is the design load effect and RD characteristic values for load L and resistance R.
is the design resistance. The design load
effect is calculated as The design equation is a special case of the
design rule obtained by turning the
LD = f LC inequality into an equality, i.e., RD = LD in
the example.
where LC is the characteristic load effect and
f is a load factor. Similarly, the design 2.3.5 Code calibration
resistance is calculated as Fulfilment of the design rules, as required by
a structural design code, is meant to ensure
RC that a particular prescribed structural safety
RD =
m level is achieved. The purpose of a code
calibration is to determine the set of partial
safety factors to be used with the chosen
where RC is the characteristic resistance and
code format, such that structural designs
m is a material factor. according to the code will meet this
prescribed level of safety.
Usually, a number of different load effects
have to be combined into a resulting load
Structural reliability analysis results,
effect, and often the largest of several
obtained as outlined above, play an
different such load combinations is used for
important role in codified practice and
the load effect in the design rule. An
design. Their application to calibration of
example of such a load combination is the
partial safety factors for use in structural
combination of gravitational loads and wind
design codes is of particular interest.
loads.
Table 2-3. Target annual failure probabilities PFT and corresponding reliability indices T.
Failure consequence
Failure type Less serious Serious Very serious
LOW SAFETY CLASS NORMAL SAFETY CLASS HIGH SAFETY CLASS
(small possibility for (possibilities for (large possibilities for
personal injuries and personal injuries, personal injuries,
pollution, small fatalities, pollution, fatalities, significant
economic and significant pollution, and very
consequences, economic large economic
negligible risk to life) consequences) consequences)
Ductile failure
with reserve PF = 103 PF = 104 PF = 105
capacity T = 3.09 T = 3.72 T = 4.26
(redundant
structure)
Ductile failure
with no reserve PF = 104 PF = 105 PF = 106
capacity T = 3.72 T = 4.26 T = 4.75
(significant
warning before
occurrence of
failure in non-
redundant
structure)
Brittle failure
(no warning PF = 105 PF = 106 PF = 107
before occurrence T = 4.26 T = 4.75 T = 5.20
of failure in non-
redundant
structure)
approximately, when designs are carried out Hauge et al. (1992), Ronold (1999), and
according to the code. Hence, the goal of a Ronold and Christensen (2001).
reliability-based code calibration is to
determine the particular common set of 2.3.6 Example axially loaded steel
partial safety factors that reduces the scatter tower
of the reliabilities, achieved by designs
The example given below deals with design
according to the code, to a minimum over
of an axially loaded steel tower against
the scope. This can be accomplished by
failure in ultimate loading. The probabilistic
means of an optimisation technique, once a
modelling required for representation of
closeness measure for the achieved
load and capacity is presented. A structural
reliability has been defined, e.g. expressed
reliability analysis of the tower is carried
in terms of a penalty function that penalises
out, and a simple calibration of partial safety
deviations from the prescribed target
factors is performed.
reliability. For principles and examples of
such code optimisation, reference is made to
The design of the axially loaded tower is
governed by the maximum axial force Q in a
one-year reference period. The maximum which gives a requirement to the ratio of the
axial force Q follows a Gumbel distribution partial safety factors 1/2 = 1.252. There is
thus an infinite number of pairs (1,2) that
FQ(q) = exp(exp(a(qb))) will lead to the required reliability. This
implies an arbitrariness in selecting the
in which a = 0.4275 and b = 48.65 partial safety factor set (1,2) for the code.
correspond to a mean value E[Q] = 50 MN The reliability analysis gives the design
and a standard deviation D[Q] = 3 MN. The point values q* = 70.89 MN for the force
yield strength of steel F follows a normal and F* = 353.2 MPa for the strength. These
distribution with the mean value E[F] = are the most likely values of the governing
400 MPa and the standard deviation D[F] = variables at failure. A robust choice for the
24 MPa. The cross-sectional area of the partial safety factors (1,2) can be achieved
tower is A. Failure occurs when the axial by designing to the design point values from
force Q exceeds the capacity FA, thus a the reliability analysis. The partial safety
natural format of the design rule is F,DA factors are therefore selected as
qD. The subscript D denotes design value.
The limit state function is correspondingly q* *
1 = = 1.227 and 2 = F = 0.980
chosen as qC F ,C
2F,CA1qC = 2360.50.2007157.78 = 0
2.3.7 Example fatigue of FRP blade deviation = 0.86. The cumulative damage
root in bending is calculated as the Miners sum
The example given here deals with design of
n ( s i )
an FRP blade root against fatigue failure D=
during its design life of 20 years. The i N (si )
probabilistic modelling required for dFS
representation of load and resistance is ds
n tot
presented. A structural reliability analysis of = ds
m
the blade root is carried out, and a simple 0 K exp( ) s
This is based on the assumption that the Note also that with f = 1.0 prescribed, all
long-term stress range distribution is known. levels of uncertainty and variability
The validity of this assumption in the associated with a fatigue problem such as
context of wind turbines is discussed later. It the present, are accounted for by one single
is also assumed that variability in the safety factor, m. This factor is thus applied
individual damage contributions from the as a safety factor on resistance, regardless of
individual stress ranges averages out over whether some of the uncertainty is
the many contributing stress ranges in the associated with load rather than with
long-term distribution. This assumption resistance. This is in accordance with most
would not hold if the cumulative damage standards. Note, however, that in the new
was dominated by damage contributions Danish standard DS409/DS410, a partial
from only one or a very few large stress safety factor, f, on load is introduced which,
ranges, which could be the case for very under certain conditions, is to be taken as a
large m values, say m>10. value greater than 1.0. This applies to
situations where the loads causing fatigue
The characteristic SN curve is taken as the damage are encumbered with uncertainty or
expected SN curve minus two standard ambiguity, such as if they are traffic loads,
deviations. A partial safety factor, m, is or if the various quantiles of the long-term
applied as a divisor on all stress range values stress distribution over the design life are
of the characteristic SN curve. statistically uncertain.
The design damage DD is then obtained as
As regards design of wind turbines against
n x fatigue, the loads causing fatigue damage
D D = tot (m + 1)( m 0 ) m
KC W are dominated by loads generated by the
wind. Whereas the distribution of the 10-
minute mean wind speed on a location may
in which KC = Kexp(2) reflects the
be well-known, the distribution of the
chosen characteristic SN curve. turbulence intensity is usually not well-
Substitution of numbers into the design determined, owing to local conditions and
equation DD = 1.0 leads to the following influence from the presence of the turbine.
requirement to the material factor Nor is the transfer function to stress
response in the wind turbine always clear.
m = 1.149 The distribution of wind-generated loads in
a wind turbine can therefore be expected to
Discussion be known only with some uncertainty, and a
Note that this example is purely tutorial to load factor f greater than 1.0 would thus be
explain a principle. In reality, the resistance required. However, in practice, one would
may be more uncertain than assumed here, account for such uncertainty or ambiguity in
e.g. when there is a limited amount of the load distribution by choosing a load
material data available, such that the distribution on the safe side, a
estimated values of K and m are uncertain. conservative envelope load spectrum, so
Moreover, model uncertainty may be to speak. In the presented example, this
associated with the application of Miners would imply the choice of a conservatively
rule. With such uncertainties properly high value of the Weibull scale parameter x0,
accounted for, the value of the resulting which could then be used in conjunction
material factor, m, will become larger than with f = 1.0.
the one found here.
the rotational speed as is the case with an an alternative part with low impedance, and
internal gear shaft, which will experience a reduction of current through the movable
one complete bending cycle for each components, e.g. by means of electrical
revolution. insulation.
wind turbine manuals and in the procedures contact with any rotating, moving or
for assembly, installation and commission- conducting parts.
ing.
The light in access routes shall have an
2.5.2 Normal operation intensity of at least 25 lux. This shall also
apply when the main switch of the wind
During normal operation of the wind
turbine is turned off.
turbine, the safety of personnel inside and
outside the wind turbine shall be considered.
Working conditions
Normal operation of the wind turbine shall
The wind turbine shall be constructed in
be possible without accessing the nacelle.
such a way that replacing components
subject to service does not entail working
Operational procedures and operation of the
postures or movements which are hazardous
wind turbine shall be described in the user
to health or otherwise dangerous. It shall be
manual, which is furnished to the turbine
possible to block the rotor and yaw system
owner or to the person responsible for the
of the wind turbine in a safe and simple
operation of the wind turbine. It shall appear
manner other than by using the ordinary
from any instructions how personnel safety
brake and yaw system of the turbine. For
has been accounted for.
pitch-controlled turbines, fixation of the
pitch setting shall be possible. Blocking of
2.5.3 Service, maintenance and repair
the rotor shall be done by mechanical
The manufacturer or supplier of the wind fixation of the rotor and shall be capable of
turbine shall provide instructions and keeping the rotor fixed at all wind speeds
procedures, which consider wind speeds and below the defined normal stop wind speed.
other external conditions in such a manner Blocking of the yaw and pitch systems shall
that service, maintenance and repair work on keep the yaw and pitch systems,
the wind turbine can be performed safely. respectively, fixed at all wind speeds below
The wind turbine shall be designed with a the defined normal stop wind speed.
view to ensuring safe access to and safe re-
placement of all components to be serviced. Operation of the blocking mechanisms and
of their area of application shall be
Access described in the user manual of the wind
It shall be made clear by means of locks turbine in order to avoid incorrect use.
and/or signs that it can be dangerous to
ascend the wind turbine. It shall be It shall be possible to illuminate working
prevented that unauthorised persons get areas with a light intensity of at least 50 lux.
access to the control panel and the In addition, the lighting must be designed
machinery of the wind turbine. Operation of such that glare, stroboscopic influences and
the wind turbine and access to its local other disadvantageous lighting conditions
control system shall not require access to are avoided.
electrical circuits with a higher voltage than
50V. It shall be possible to initiate emergency
shutdown close to the working areas in the
Wherever screens and shields are used for wind turbine. As a minimum, it shall be
protection, it shall be ensured that, during possible to initiate emergency shutdown at
normal operation, personnel cannot get in the bottom of the tower, i.e. at the control
panel, and in the nacelle.
where z0 is the terrain roughness parameter Note also that the interpretation of the
which is defined as the extrapolated height limiting value = 1/ln(z/z0) is similar to that
at which the mean wind speed becomes of a turbulence intensity as z approaches the
zero, if the vertical wind profile has a reference height H, cf. the definitions given
32 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
z0 (m) z0 =
g ln( z z )
Plane ice 0.00001 0
3 External Conditions 33
Guidelines for Design of Wind Turbines DNV/Ris
3.1.2 Standard deviation of wind be the case for example if a house is located
speed nearby. Houses and other disturbing
elements will, in general, lead to more
For a given value of U10, the standard
turbulence, i.e. larger values of E[U] and
deviation U of the wind speed exhibits a
D[U], than will normally be found in
natural variability from one 10-minute
smoother terrain. Figure 3-1 and Figure 3-2
period to another. This variability of the
wind speed is known as the turbulence, and give examples of the variation of E[U] and
U is therefore often referred to as the D[U] with U10 for onshore and offshore
standard deviation of the turbulence locations, respectively. The difference
components. Measurements from several between the two Figures is mainly
attributable to the different shape of the
locations show that U conditioned by U10
mean curve. This reflects the effect of the
can often be well-represented by a
increasing roughness length for increasing
lognormal distribution
U10 on the offshore location.
ln b0
F |U 10
( ) = ( )
U
b1 3
mean value
2,5
(m/sec)
in which () denotes the standard Gaussian st. dev.
34 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
(m/sec)
2
mean value
st. dev.
such data may confuse the determination of
an appropriate distribution model for U
E [ U ] 1,5 conditioned by U10.
3 External Conditions 35
Guidelines for Design of Wind Turbines DNV/Ris
36 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
If the wind farm consists of more than five U can be expected. Beware that
rows with more than five turbines in each calculations for changes in the direction of
row, or if the distance between the turbines the wind in complex terrain may come out
in the rows that are located perpendicular to very wrongly, if values for Uy and Uz,
the predominant wind direction is less than which are valid for homogeneous terrain, are
3D, the increase in mean turbulence applied.
intensity shall be taken into account. This is
done by substituting the free flow turbulence Very often, the wind climate at a particular
IT with IT*: location cannot be documented by site-
specific measurements. In such situations,
IT * = I w 2 + IT 2 + IT the distribution of U10 can usually be well-
represented, for example on the basis of
wind speed measurements from a nearby
0.36 location. However, the distribution of U
Iw =
1 + 0.08 sr s f v will usually be harder to obtain as it is
highly dependent on the local roughness
conditions. Thus, it cannot be inferred
sr = xr / D automatically from known wind speed
conditions at nearby locations. On a location
sf = xf / D where wind speed measurements are not
available, determination of the distribution
where xr is the distance within a row, and xf
of the standard deviation U of the wind
is the distance between rows.
speed is often encumbered with ambiguity.
It is thus common practice to account for
3.1.4 Lateral and vertical turbulence
this ambiguity by using conservatively high
The 10-minute mean wind speed, the values for U for design purposes, viz. the
standard deviation of the wind speed, and characteristic values for U given in DS472
the turbulence intensity presented above all and IEC61400-1 and referenced above.
refer to the wind speed in the constant
direction of the mean wind during the 3.1.5 Stochastic turbulence models
considered 10-minute period of stationary
conditions. During this period, in addition to Wind in one direction is considered, i.e. in
the turbulence in the direction of the mean the direction of the 10-minute mean wind
wind, there will be turbulence also laterally speed. The wind speed process U(t) within a
and vertically. The mean lateral wind speed 10-minute period of constant U10 and U is
will be zero, while the lateral standard considered and can be assumed to be
deviation of the wind speed can be taken as stationary. The spectral density of the wind
Uy = 0.75U according to Dyrbye and speed process expresses how the energy of
the wind turbulence is distributed between
Hansen (1997) and as Uy = 0.80U
various frequencies. Several models for the
according to Panofsky and Dutton (1984).
spectral density exist. A commonly used
The mean vertical wind speed will be zero,
model for the spectral density is the Harris
while the vertical standard deviation of the
spectrum
wind speed can be taken as Uz = 0.5U.
These values all refer to homogeneous
terrain. For complex terrain, the wind speed
field will be much more isotropic, and
values for Uy and Uz very near the value of
3 External Conditions 37
Guidelines for Design of Wind Turbines DNV/Ris
3 2fL 2 5 / 6
(1 + ( ) )
2 U 10 The turbulence scale parameter is by
definition the wavelength where the non-
in which f denotes the frequency, and L is a dimensional, longitudinal power spectral
characteristic length, which relates to the density is equal to 0.05.
integral length scale Lu by L = 1.09Lu. A
calibration to full scale data indicates values 10
for L in the range 66-440 m with L 200 m
used to match the high frequency portion of 1
the spectrum. Based on experience, the C, m
m
C
Harris spectrum is not recommended for use 0,1
in the low frequency range, i.e. for
f < 0.01 Hz. 0,01
0,001 0,01 0,1 1 10
38 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
Note that there is some arbitrariness in the a homogeneous terrain. However, for
models for power spectral density. Each turbulence in a complex terrain it is not
model implies idealization and uncommon to see a skewness of 0.1, which
simplification and is usually calibrated to implies that the Gaussian assumption has not
provide a good fit to data within a limited been fulfilled.
frequency range. At low frequencies, in
particular, the models show significant Note that although the short-term wind
differences. In Figure 3-5 three models for speed process will be Gaussian for
power spectral densities are plotted on homogeneous terrain, it will usually not be a
dimensionless scales for comparison narrow-banded Gaussian process. This
comes about as a result of the spectral
Harris density and is of importance for prediction
10
of extreme wind speed values. Such extreme
Kaimal
(Eurocode) values and their probability distributions can
IEC limit be expressed in terms of spectral moments.
1
Reference is made to textbooks on stochastic
process theory.
2
fSu/U
0,1
At any point in time there will be a
variability in the wind speed from one point
to another. The closer together the two
0,01
points are, the higher is the correlation
between their respective wind speeds. The
0,001
wind speed will form a random field in
0,01 0,1 1 10 100 space. A commonly used model for the
fLu/U10 autocorrelation function of the wind speed
field can be derived from the exponential
Figure 3-5. Comparison between the Harris, Kaimal
and IEC power spectral densities. Davenport coherence spectrum
r
Spectral moments are useful for Coh(r , f ) = exp(cf )
representation of the wind speed process u
U(t). The jth spectral moment is defined by where r is the distance between the two
points, u is the average wind speed over the
distance r, f is a frequency, and c is the non-
m j = j S U ( )d dimensional decay constant, which is
0 referred to as the coherence decrement, and
which reflects the correlation length of the
In the short term, such as within a 10-minute wind speed field. The auto-correlation
period, the wind speed process U(t) can function can be found as
usually be represented as a Gaussian
process, conditioned by a particular 10- 1
3 External Conditions 39
Guidelines for Design of Wind Turbines DNV/Ris
vertically. Note that it is a shortcoming of are not considered important for small wind
the Davenport model that it is not turbines, with rotor diameters in the order of
differentiable for r = 0. Note also that, 10 m. Wind shear may be important for
owing to separation, the limiting value (0) large and/or flexible rotors. A number of
will often take on a value somewhat less failures have been attributed to blade loads
than 1.0, whereas the Davenport model induced by wind shear.
always leads to (0) = 1.0.
The wind profile depends heavily on the
Note that the integral length scale Lu, atmospheric stability conditions, see Figure
referenced above as a parameter in the 3-6 for an example. Even within the course
models for the power spectral density, is of 24 hours, the wind profile will change
defined as between day and night, dawn and dusk.
Neutral
neutral conditions. Unstable conditions
1000 Unstable
Strongly unstable
60
height above ground (m)
neutral
50 stable
100 unstable
40
height (m)
30
10 20
10
0
1 6 7 8 9 10 11 12 13
0 5 10 15 wind speed (m/s)
w ind speed (m /s)
40 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
3 External Conditions 41
Guidelines for Design of Wind Turbines DNV/Ris
near such features, and the wind profiles will Though the yaw system of the wind turbine
become altered. Theories exist for will hold the rotor in the direction of the
calculation of such changed wind profiles, mean wind direction, the short-term
see Jensen (1999). An example of effects of fluctuations in the wind direction give rise to
a ridge is given in Figure 3-8. fatigue loading. At high wind speeds sudden
changes in the wind direction during
production can give rise to extreme loads.
100
downwind
Wind rose
The distribution of the wind direction is of
upwind
particular interest with respect to installation
ridge crest
of turbines in wind farms. As can be seen in
Section 3.1.3, wind turbines installed behind
height (m)
42 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
quantity is normalised, thereby allowing the pass and is usually not covered by
maximum to reach the outer circle. The commonly available analysis tools for
corresponding frequency in % for each of simulation of turbulence. Figure 3-11 to 12
the three quantities is given in the small box show examples of the most import transient
below the wind rose. The inner dotted circle wind events.
corresponds to half of the value of the outer
circle.
Figure 3-12. Example of strong wind shear. Notice that Extreme value analysis
the wind speed is inversely proportional to the height, Extreme winds are usually given in terms of
i.e. so-called negative wind shear. Courtesy Vestas. 10-minute mean wind speeds, which occur
with some prescribed recurrence period, e.g.
These are all wind events, which by nature the 50-year wind speed. The 50-year wind
fall outside of what can normally be speed is the 10-minute mean wind speed,
represented by stationary wind conditions. which, on average, is exceeded once every
Note that the simultaneous change in wind fifty years. Determination of the 50-year
direction and wind speed occurs when fronts wind speed requires an extreme value
3 External Conditions 43
Guidelines for Design of Wind Turbines DNV/Ris
analysis of available wind speed data. It has formula is used to implicitly solve a new
proven useful to carry out such an extreme value of u* that corresponds to the desired
value analysis on the friction velocity new reference roughness z0.
pressures derived from the wind speed data,
rather than on the wind speed data When the friction velocities u* have been
themselves. For this purpose, the observed determined from the wind speed data as
wind speeds u are transformed to friction described above, the corresponding velocity
velocities u* by pressure q is calculated from
u 1
u* = q= u * 2
ln( z z 0 ) 2
44 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
where T0 = 1 year. This, in particular, can be When the wind speed with a 50-year
used to find the 50-year velocity pressure for recurrence period is given, i.e. U10,50-yr, the
T = 50 years. wind speed with the recurrence period T
years can be found as
The corresponding 10-minute mean wind
speed with the recurrence period T at height T
z and terrain roughness z0 can be found as U 10 ,T = U 10, 50 yr 0.57 + 0.11 ln
ln(1 p )
1 2q T z
U 10 ,T = ln in which p = exp(nT) is the probability of
0 z0 no exceedance in T years, and n is the
number of exceedances per year. For T = 50
= 0.4 Von Karmans constant years, n = 0.02. Reference is made to
3
0 = 1.225 kg/m density of air DS410.
3 External Conditions 45
Guidelines for Design of Wind Turbines DNV/Ris
Some parts of the world are characterised by intervals of another duration than 10
having other weather phenomena than those minutes, the mean wind speed and standard
considered here. One such weather deviation over these other periods need to be
phenomenon is cyclones. As this is a transformed to values referring to a duration
weather phenomenon totally different from of 10 minutes. The following approximate
the storms dealt with above, one cannot just formula applies to transformation of the
extrapolate results for such storms. mean wind speed UT in the measurement
period T to the 10-minute mean wind speed
3.1.10 Site assessment U10
Before a wind turbine is installed, various
UT
conditions of the specific site have to be U 10
evaluated. It shall be assessed that the T
1 0.047 ln
environmental, electrical and soil properties 10
are more benign than those assumed for the
design of a turbine. If the site conditions are in which T is to be given in units of minutes.
more severe than those assumed, the Reference is made to Gran (1992) and DNV
engineering integrity shall be demonstrated. (1998).
The environmental conditions include:
temperature, icing, humidity, solar radiation, When the measurement period T is less than
corrosion conditions and possible 10 minutes, the standard deviation of the
earthquakes. wind speed measurements will come out
smaller than the sought-after value for U.
The wind conditions shall be assessed from
monitoring measurements made on the site, The following approximate formula applies
long-term records from a nearby to transformation of the standard deviation
meteorological station or from local codes or U,T obtained from wind speed
standards. Where appropriate, the site measurements in the measurement period T
conditions shall be correlated with long-term to the 10-minute standard deviation U
data from local meteorological stations. The
monitoring period shall be sufficient to T 2
U U ,T + U 10 (0.047 ln
2 2
obtain a minimum of six months of reliable )
data. Where seasonal variations contribute 10
significantly to the wind conditions, the
monitoring period shall include these in which T is to be given in units of minutes.
effects. The formula is valid for T < 10 minutes. In
cases where T > 10 minutes, the formula to
The two variables U10 and U are essential be used reads
as parameters in the models available for
representation of the wind speed. Estimation T 2
U U ,T U 10 (0.047 ln
2 2
)
of wind speed measurements constitutes the 10
most common method for determination of
these two parameters. Both U10 and U refer As described in Section 3.1.1, the long-term
to a 10-minute reference period, U10 being distribution of the 10-minute mean wind
the 10-minute mean wind speed and U speed U10 can be represented by a Weibull
being the standard deviation of the wind distribution
speed over the 10 minutes. Note that if wind
speed measurements are obtained over
46 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
3 External Conditions 47
Guidelines for Design of Wind Turbines DNV/Ris
When the prediction site is more rugged is very limited. Insofar as regards wind
than the reference site, the wind speeds at turbine structures, DS472 gives a
the prediction site will be overpredicted by temperature interval (10C, 30C) for
WASP. Similarly, the wind speeds will be operation as an example. However, the
underpredicted when the prediction site is temperature interval for normal operation
less rugged. The difference in RIX numbers should always be chosen in accordance with
between the two sites is a fairly coarse the specified recommendations for the
measure of the significance of the problem respective materials which are being used
and provides estimates of the magnitude and for the construction of the wind turbine. In
sign of the prediction error. As a rule of the this context, the following issues may be
thumb, when there is a difference RIX in critical:
RIX numbers between the prediction site components and connections, which
and the reference site, the approximate involve two or more materials with
magnitude of the relative prediction error in different coefficients of expansion. The
the wind speed will be 2RIX. For example, level of expansion in relation to glass,
if RIXprediction = 20% and RIXreference = 10%, concrete and steel is more or less the
then RIX = 20%10% = 10%, and the same, whereas plastic expands more.
overall relative error in the predicted wind choice of fluids for lubrication and
speeds will be about 210 = 20%. hydraulic systems.
materials whose mechanical properties
Note that WASP engineering can give change when the temperature changes,
accurate results outside its operation limits, e.g. rubber for seals, gaskets and
provided that the difference in RIX numbers dampers becomes brittle at low
between the actual site and the reference site temperatures, and polyester behaves
is small and that the topographical data are like glass when the temperature falls
adequate and reliable. below the so-called glass transition
temperature.
It is important to consider the temperature
3.2 Other external conditions interval for normal operation when materials
are selected for the construction of the wind
3.2.1 Temperatures
turbine and when other conditions of
Operational temperatures importance for the safety of the turbine are
A temperature interval for normal operation to be evaluated.
of the wind turbine is to be chosen. Under
normal functional conditions, the wind Extreme temperatures
turbine is considered to be in operation For structural parts which can be damaged
when the temperature of the air is within this by extreme temperatures, extreme values of
interval. temperatures need to be considered. Extreme
values of high and low temperatures should
Different structural design codes offer be expressed in terms of the most probable
different approaches to how temperatures upper or lower values with their corre-
can be handled. Insofar as regards steel sponding recurrence periods.
structures, a rather general instruction can be
found in DS412 according to which one may Table 3-4 gives values of the daily minimum
set the lowest temperature of operation at temperature and the daily maximum
10C for Danish locations. This is due to the temperature for various recurrence periods
fact that the frequency of low temperatures at two Danish locations. These values are
48 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
3 External Conditions 49
Guidelines for Design of Wind Turbines DNV/Ris
50 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
damage. If no special provisions are made, be such that a sufficient corrosion protection
leakage through covers should be system may be established.
considered. If there is a possibility for
accumulation of snow on the wind turbine, Industrial environments may be rather harsh
the increase of weight due to such to wind turbine structures.
accumulation should be considered. The
density of heavy snow is in the range 100- Offshore wind turbines will be exposed to
150 kg/m3. corrosion owing to their saline and marine
environment. Their foundations will have
In general, caution should be exhibited with structural parts whose locations in the wave
regard to installation of wind turbines in splash zone will be particularly exposed to
areas where hail storms are known to corrosion.
prevail. When designing wind turbines for
areas where hail storms are common, The expected corrosion rate depends on the
possible damage due to impact from hail environment.
stones should be considered. The nacelle
and the coating on the blades may be 3.2.8 Earthquake
especially vulnerable to such damage, in
The effects of earthquakes should be
particular, if the hail stones fall during
considered for wind turbines to be located in
strong winds, i.e. they hit the turbine at an
areas that are considered seismically active
angle different from vertical. The velocity
based on previous records of earthquake
by which a hail stone falls towards the
activity.
ground is governed by drag and is not
influenced by the weight of the hail stone.
For areas where detailed information on
When the speed of the hail stone is known
seismic activity is available, the seismicity
together with the wind speed, the angle by
of the area may be determined from such
which the hail stone will hit a wind turbine
information.
structure can be determined.
For areas where detailed information about
Extreme hail conditions may exist on some
seismic activity is not generally available,
locations. Such conditions can be critical for
the seismicity should preferably be
a wind turbine and may force the turbine to
determined on the basis of detailed
stop. Extreme hail events may in some cases
investigations, including a study of the
govern the design of the leading edges of the
geological history and the seismic events of
rotor blades. The density of hail is usually
the region.
higher than that of snow. NASA (1978)
reports a density of hail of about 240 kg/m3.
If the area is determined to be seismically
active and the wind turbine is deemed to be
3.2.7 Atmospheric corrosion and
affected by a possible earthquake, an
abrasion
evaluation should be made of the regional
Abrasive action by particles transported by and local geology in order to determine the
the wind should be considered for exposed location relative to the alignment of faults,
surfaces. the epicentral and focal distances, the source
mechanism for energy release, and the
The corrosive environment should be re- source-to-site attenuation characteristics.
presented by generally recognised methods. Local soil conditions need to be taken into
The classification of the environment should account to the extent they may affect the
3 External Conditions 51
Guidelines for Design of Wind Turbines DNV/Ris
52 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
3 External Conditions 53
Guidelines for Design of Wind Turbines DNV/Ris
1188(EN), Ris National Laboratory, Wind Mortensen, N.G., and E.L. Petersen,
Energy Department, May 2001. Influence of topographical input data on the
accuracy of wind flow modelling in complex
Gran, S., A Course in Ocean Engineering, terrain, Proceedings, European Wind
Elsevier, Amsterdam, the Netherlands, 1992. Energy Conference, Dublin, Ireland, 1997.
The Internet version, located at
www.dnv.com/ocean/, provides on-line National Aeronautics and Space
calculation facilities within the fields of Administration (NASA), Engineering
ocean waves, wave loads, fatigue analysis, Handbook on the Atmospheric
and statistics. Environmental Guidelines for Use in Wind
Turbine Generator Development, NASA
IEC61400-1, Wind turbine generator Technical Paper 1359, 1978.
systems Part 1: Safety requirements,
International Standard, 2nd edition, 1999. Norwegian Petroleum Directorate, Acts,
regulations and provisions for the petroleum
Jensen, N.O., Atmospheric boundary layers activity, Stavanger, Norway, 1994.
and turbulence, in: Wind engineering into
the 21st century. Proceedings of Tenth Panofsky, H.A., and J.A. Dutton, Atmo-
International Conference on Wind spheric Turbulence, Models and Methods
Engineering, Copenhagen, Denmark, 21-24 for Engineering Applications, John Wiley
June 1999. Larsen, A.; Larose, G.L.; and Sons, New York, N.Y., 1984.
Liversey, F.M. (eds.), A.A. Balkema,
Rotterdam/ Brookfield, Vol. 1, pp. 29-42, Statens Planverks Frfattningssamling,
1999. Svensk Byggnorm, SBN avd. 2A, Brande
konstruktioner, med kommentarer, in
Laursen, E.V., J. Larsen, K. Rajakumar, J. Swedish, LiberFrlag, Stockholm, Sweden,
Cappelen, and T. Schmith, Observed daily 1980.
precipitation and temperature from six
Danish sites, 1874-1998, Technical Report
No. 99-20, Danish Meteorological Institute,
Copenhagen, Denmark, 1999a.
54 3 External Conditions
Guidelines for Design of Wind Turbines DNV/Ris
4 Loads 55
Guidelines for Design of Wind Turbines DNV/Ris
it is the duty of the designer to validate their wake effects wherever the wind turbine
adequacy. When calculating loads it is is to be located behind other turbines,
important to keep in mind a few factors e.g. in wind farms
which may influence the involved wind misalignment of wind flow relative to
conditions and the magnitude of the loads: the rotor axis, e.g. owing to a yaw error
tower shadow and tower stemming, i.e.
the disturbances of the wind flow owing Note that some of the load cases that are
to the presence of the tower listed in Table 4-1 and Table 4-2 are load
cases which refer to site-specific wind
turbines only.
56 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
4 Loads 57
Guidelines for Design of Wind Turbines DNV/Ris
4.2 Load types mean load, i.e. the mean flapwise bending
moment, and in a reduction of the stiffness.
The external loads acting on a wind turbine
are mainly wind loads. As a wind turbine
The load response in a rotor blade is very
consists of slender elements such as blades
dependent on the damping. The total
and tower, inertia loads will be generated in
damping is a combination of aerodynamic
addition to the gravity loads that act on these
damping and structural damping. The
elements. Loads due to operation such as
aerodynamic damping depends on:
centrifugal forces, Coriolis forces and
choice of blade aerodynamic profile in
gyroscopic forces must also be considered.
conjunction with chosen blade twist
In most cases, the loads on a wind turbine operational condition
can thus be classified as follows: wind speed
aerodynamic blade loads rotor frequency
gravity loads on the rotor blades vibration direction of blade cross-
centrifugal forces and Coriolis forces section
due to rotation motion of blade section relative to
gyroscopic loads due to yawing incoming flow
The structural damping depends much on
aerodynamic drag forces on tower and
the blade material. The aerodynamic load
nacelle
response is very much a result of the lift and
gravity loads on tower and nacelle
drag forces in conjunction with the blade
profile properties and damping, and with
Gravity loads on the rotor blades cause
effects of the motion of the rotor structure
bending moments in the blades in the
included.
edgewise direction. For a pitch-controlled
turbine, gravity loads will also cause
The following subsections give a brief
bending moments in the flapwise direction.
introduction to the most important load
Due to the rotation of the blades, the gravity
types encountered for a wind turbine with
load effects in the blades will be cyclically
emphasis on the physics behind them. More
varying bending moments. The larger the
details about loads and how to predict them
rotor diameter is, the greater are the gravity
are given in Sections 4.3 through 4.5.
load effects. Typically, the bending moment
at the blade root will follow a fourth-power
4.2.1 Inertia and gravity loads
law in the rotor diameter. Considering that
the rotor area follows a quadratic power law The inertia and gravity loads on the rotor are
in the rotor diameter, this forms one of the mass dependent loads. The cross-sectional
greater challenges in making wind turbines centrifugal force Fc depends on the angular
larger. rotor speed, the radial position, and the mass
of each blade element. At the blade root, this
Centrifugal forces induced by the rotation of force is
the blades can be utilised in conjunction
with rearwards coning of the blades to n
58 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
In general, gyroscopic loads on the rotor will Note that the quoted formulas for MK and
occur for any flexible rotor support. In MG need to be adjusted for a possible rotor
particular, gyroscopic loads on the rotor will tilt.
occur whenever the turbine is yawing during
operation. This will happen regardless of the 4.2.2 Aerodynamic loads
structural flexibility and will lead to a yaw Blades
moment MK about the vertical axis and a tilt The true wind flow in the vicinity of the
moment MG about a horizontal axis in the wind turbine rotor is rather complex,
rotor plane. because the rotor induces velocities. Hence
it is common practice to use a simplified
For a three-bladed rotor, the net resulting method for calculating rotor loads to be used
yaw moment due to the gyroscopic load for a wind turbine design.
effects is zero, MK = 0, whereas a non-zero
constant tilt moment is produced, MG = The wind velocity conditions at a blade
3M0/2, in which cross-section are illustrated in Figure 4-1.
n
M 0 = 2 K m i ri
2 Rotor plane
i =1
MK = 2 M0 cos(t) sin(t)
Figure 4-1. Diagram of air velocities at a blade cross-
and section.
4 Loads 59
Guidelines for Design of Wind Turbines DNV/Ris
Fd = 0.5 A V02 CD
4.3 Aeroelastic load calculations
CD aerodynamic drag coefficient Load calculations for a wind turbine
A projected area perpendicular to the flow structure are usually performed by means of
a computer program based on an aeroelastic
calculation procedure. The purpose of an
aeroelastic wind turbine analysis is to solve
the equations of motion for a given arbitrary
60 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
set of forces acting on the structure and for The loads that are derived from an
forces generated by the structure itself. aeroelastic model are used for design of the
Often, such a code applies a geometrically various components that constitute the wind
non-linear finite element approach or a turbine structure. In this context, most
modified modal analysis approach. components are subjected to investigations
Reference is made to Figure 4-2. concerning both extreme loading and fatigue
loading.
In any case, it is required that the code must
be able to include and simplify the complex It is inevitable that some engineering
mechanical structure of the wind turbine as judgment will be involved when preparing
well as being able to model arbitrary for aeroelastic load calculations. It is,
deterministic and stochastic forces acting on therefore, important that the modelling as far
the turbine. The general formulation of the as possible is supported by measured data,
differential equations of motion is which may be available from various
sources. These include type testing of the
M&x& + Cx& + Kx = F rotor blades and prototype testing of an
actual turbine, and they are both important
M mass matrix to consider when an aeroelastic model is to
C damping matrix be validated.
K stiffness matrix
F force vector acting on the structure and 4.3.1 Model elements
typically varying with time
Wind field modelling
x and the derivatives of x are unknown
The wind field contains three wind velocity
vectors containing translations and/or
components:
rotations and their derivative
longitudinal wind velocity
transversal wind velocity
vertical wind velocity
The wind field is usually divided into:
a mean wind field with shear and slope
a fluctuating wind field, i.e. turbulence
Wind field simulation is an important part of
a structural wind turbine analysis. For wind
turbines, spatial variations in the turbulence
must be considered, and three-dimensional
wind simulation is required. A prime
purpose of wind field simulation is to
predict time series of the wind speed in a
number of points in space, e.g. a number of
points across the rotor disc of a wind
turbine. Such time series of the wind speed
form useful input to structural analysis
models for wind turbines.
4 Loads 61
Guidelines for Design of Wind Turbines DNV/Ris
62 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
standard deviation of the longitudinal wind turbine blades stall and thereby limit the
velocity by IT = u/U, is usually measured power output and the loads. The overall
by means of a cup anemometer. This three-dimensional flow on a rotor is a very
measurement corresponds to vectorial complex, unsteady flow depending on a
summation of the longitudinal and number of variables, which include:
transversal wind velocity components. wind speed
Hence, the turbulence intensity calculated on wind shear
this basis usually comes out somewhat atmospheric turbulence
higher than the sought-after turbulence yaw angle
intensity for the longitudinal component rotational speed
alone. However, it is often used to represent rotor radius
the longitudinal intensity when no other overall layout of the rotor blade
estimate is available. twist
taper
Data for the relationship between u, v, and
thickness distribution
w are given in Section 3.1. The longitudinal
airfoil properties
length scale Lu is in general dependent on
thickness
the height and the terrain roughness, see
Section 3.1 for details. The transverse and camber
vertical length scales can be assumed to smoothness of surface
relate to the longitudinal length scale by Lv = leading edge thickness
0.3Lu and Lw = 0.1Lu. Typical values of the roughness insensitivity
longitudinal decay factors ci are in the range blunt/sharp trailing edge
2-27.
Sum of pressure forces
4 Loads 63
Guidelines for Design of Wind Turbines DNV/Ris
methods do exist, but they are time- stall has been reached. The stall angle stall
consuming and not considered any better. is somewhat arbitrary.
Consider first a two-dimensional flow past a Lift and drag coefficients CL and CD are
profile. Two-dimensional flow is confined to defined as
a plane. The out-of-plane velocity is thus
zero. In order to obtain a two-dimensional FL
flow, it is necessary to extrude the profile CL =
1
into a blade of infinite span. For a true W 2 c
blade, the shape, the twist and the profile 2
change over the span, and the blade starts at
a hub and ends in a tip. However, for long and
slender blades, the spanwise velocity
component is usually small relative to the FD
CD =
streamwise component, such that 1
W 2 c
aerodynamic solutions for two-dimensional 2
flow have a practical interest for wind-
turbine rotor blades. The resultant force F density of air
from the flow on the blade comes about as c Chord length of the airfoil
the integral of pressure and frictional forces, FL lift forces per unit length
see Figure 4-3. The resultant force F is FD drag forces per unit length
decomposed into one component FL
perpendicular to the direction of the The lift and drag coefficients are functions
resulting relative wind velocity W and one of the inflow angle , as shown in Figure
component FD parallel to this direction. FL is 4-5, of the airfoil shape, and of the Reynolds
the lift and FD is the drag, see Figure 4-4. number Re = cW/, in which is the
Both the lift FL and the drag FD depend on kinematic viscosity. CL increases linearly
the inflow angle . with up to the stall angle, stall, where the
profile stalls. For greater values of , CL
reaches a maximum value, followed by a
decrease in CL for further increases in . For
small , CD is almost constant, but increases
rapidly after stall. The stall phenomenon is
closely related to the separation of the
boundary layer from the upper side of the
airfoil. The manner in which the profile
stalls is dependent on the geometry of the
profile. Thin profiles with a sharp nose, i.e. a
Figure 4-4. Definition of lift and drag on airfoil in 2-D
flow. high curvature around the leading edge, tend
to stall more abruptly than thicker airfoils.
c/4 in Figure 4-4 is referred to as the quarter This is a result of differences in the way the
chord point. boundary layer separates. Values of the
coefficients CL and CD can be looked up in
Stall is a nonlinear phenomenon that results tables, calibrated to wind tunnel
in a dramatic loss of flow attachment and measurements, see Abbott and von
airfoil lift when some limiting inflow angle Doenhoff (1959).
64 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
A1 = B1 / 2
A2 = (C Ls C D max sin( s ) cos( s ))
(sin( s ) / cos 2 ( s ))
Figure 4-5. Lift and drag curve for an airfoil (Bak et al.,
1999a). s inflow angle at stall onset (usually 15o)
CDs drag coefficient at stall onset
An example of the variation of CL and CD CLs lift coefficient at onset of stall
with the inflow angle is given in Figure
4-5. It appears that two-dimensional wind To describe the physics in a mathematical
tunnel measurements are used to obtain the way, the wind turbine rotor can be
coefficient values in the pre-stall region. considered as a disc, which is able to absorb
Computational fluid dynamics with energy from the wind by reduction of the
corrections for three-dimensional effects is wind speed. The wind speed and the
used to obtain these values in the stall pressure conditions around the disc are
region. In the post-stall region, illustrated in Figure 4-6.
measurements are typically used to
determine the coefficient values. Otherwise,
a method by Viterna and Corrigan (1981)
can be used for their prediction. This method
assumes rotors with zero twist angle, and
results by the method therefore need
modification when this assumption is not
fulfilled. By this method, the maximum drag
coefficient at inflow angle = 90o is
Speed
4 Loads 65
Guidelines for Design of Wind Turbines DNV/Ris
wake behind the rotor will be rotating. The The torque on the ring element is
wake is illustrated in Figure 4-7.
dQ = 2 r 2 u C dr
66 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
4 Loads 67
Guidelines for Design of Wind Turbines DNV/Ris
FN = cC N
2 sin 2 Moments of stiffness inertia about the
various axes defined in Figure 4-9 can be
and the tangential force per length unit of calculated by means of standard formulas,
the blade is which can be found in structural engineering
textbooks. The angle between the
reference axis X and the principal axis X
1 V 0 (1 a)r (1 + a ' )
FT = cC T can be calculated as
2 sin cos
1 2[ED X 'Y ' ]
The procedure is repeated for all ring = arctan
elements modelled, i.e. for all radius values
2 [EI Y ' ] [EI X ' ]
r, and the result thus consists of distributions
along the rotor blade of the normal and where
tangential forces per unit length. These
distributions form the basis for calculating [ED ] = EX ' Y ' dA
X 'Y '
stresses, forces and moments in any cross- A
Structural modelling
Rotor blades are slender such that they from are the bending stiffness properties about the
a structural point of view will act like X and Y references axes, respectively. The
beams, and beam theory can be applied. bending stiffness about the principal axes
Reference is made to Section 5.1. For can now be computed as
analysis of a rotor blade by means of beam
68 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
1
P= Au 3 C P ( , )
2
density of air
A rotor area
u wind speed
Figure 4-9. Section of blade showing principal axes, CP rotor efficiency
chord line, tip chord line, and angles to chord line.
CP which is a function of the pitch angle
The most important structural stiffness data and of the tip speed ratio = RR/u. R is
for a rotor blade have been dealt with here. the angular frequency of the rotor, and R is
Note that since present wind turbine blades the rotor radius.
are usually relatively stiff in torsion, the
torsional stiffness has usually not been The pitch angle and the rotor speed R are
considered. Note also that this may change the two parameters which can be used for
in relation to future wind turbine designs. the control of the turbine. The control of the
turbine is usually based on one of the
In structural modelling and analysis, it is following two approaches:
important to be aware of the flutter Optimisation of the power below the
phenomenon, which may result from nominal power, i.e. the rotor is kept as
coupled torsional and flapping motion. Low close as possible to the maximum value
ratios of torsional and flapwise frequencies of CP, which occurs for a particular
and high tip speeds indicate rise of flutter, optimal set (opt, opt) of and . This is
which may be destructive for the rotor blade. achieved by choosing equal to its
Reference is made to Section 4.3.4. optimal value opt, and keeping
constant at its optimal value opt.
Control system modelling
However, cannot be controlled
The control system is to keep the operating
directly, since u is hard to measure with
parameters of the wind turbine within
sufficient accuracy. Therefore, instead,
specified normal limits. This, in turn, is to
information about which power
keep the loads on the wind turbine within
production is optimal for a given value
certain limits. Because the operating
4 Loads 69
Guidelines for Design of Wind Turbines DNV/Ris
70 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
the stall region should still be correctly the lift coefficients are lower at the
modelled, because no pitch regulation blade tip in the stall region when 3-D
mechanism can be considered so perfect that effects are included
the blades will never experience a stalled the lift coefficients are unchanged at a
condition. distance approximately two thirds of the
rotor radius from the rotor axis
When the turbine is stall-regulated or active- the lift coefficients are higher on the
stall-regulated, an appropriate choice of the inner part of the blade, i.e. closer to the
aerodynamic coefficients is more difficult to rotor axis, when 3-D effects are
achieve. A first step in deriving included
representative values for the aerodynamic the drag coefficients are unchanged on
coefficients consists of identifying the the outer part of the blade
profile series used for the rotor blade. When the drag coefficients are slightly lower
the profiles are well-known, it is easy to find on the inner part of the blade for inflow
two-dimensional wind tunnel data to support angles up to approximately 20 when 3-
the choice of coefficient values. Data for so- D effects are included. For higher
called NACA profiles can readily be found inflow angles, the drag coefficients are
in Abbott and von Doenhoff (1959). When higher than indicated by 2-D data.
the profiles are not well-known, it is
recommended to inspect a visualisation of Beware that these guidelines are based on
the profile shape and to find a representative calibration of computational results for one
and well-known profile series for which particular blade, viz. the LM19.1 blade, and
measured aerodynamic two-dimensional cannot necessarily be projected to apply to
data exist. other blades without validation.
The next step is to assign values to the
aerodynamic coefficients with due It is usually not possible to find
consideration to possible three-dimensional aerodynamic data for inflow angle outside
effects. It is recommended to evaluate or the range 20-20. As high inflow angles
calibrate the aerodynamic coefficients from outside this range do occur on wind turbines,
power curve data or thrust curve data. A in particular, during extreme conditions
representative thrust curve can easily be formed by extreme yaw errors or extreme
derived from the tower bottom bending. wind speeds, it is often necessary to
When reliable measured data are available, extrapolate the available aerodynamic data
this will provide a good basis for deriving to values for these high angles. A method
correct data. for such extrapolation can be found in
Eggleston and Stoddard (1987).
In most cases, measurements are not Alternatively, the method by Viterna and
available. The assignment of values to the Corigan (1981), described in Section 4.3.1,
aerodynamic coefficients must be based on a can be used.
general impression and on experience.
Alternatively, it can be considered to apply Note that proper selection of values for the
guidelines for three-dimensional corrections aerodynamic coefficients is a very important
of two-dimensional data according to Bak et step in the design analyses of a wind turbine,
al. (1999b). These guidelines can be since a reliable prediction of the dynamic
summarised as follows: response of the wind turbine is very
dependent on correct choices of aero-
dynamic coefficients.
4 Loads 71
Guidelines for Design of Wind Turbines DNV/Ris
A commonly used model for representation i = 1,p, in which i is the ith damping ratio
of damping is the Rayleigh damping model and i is the angular frequency for the
whose major advantage is a decoupling of vibration mode that i was obtained for.
72 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
The disadvantage of using this generalised and the damping coefficient is easily
damping model is that the damping matrix C determined from a curve fit of the equation
in the general case is a full matrix. This
may cause the simulations in an aeroelastic f ( x) = C exp( f n t )
analysis to become rather time-consuming.
For p = 2 this model is reduced to a t time
Rayleigh damping model. fn excitation frequency
logarithmic decrement
For appreciation of the results presented C calibration constant from the fit
herein, recall that the ith damping ratio is
defined as An example of such a fit is shown in Figure
4-10. The dynamic coefficients in the model
c must be adjusted until the correct
i =
2 mi k i logarithmic decrement results from this
exercise.
mi ith mass
ci ith damping
ki ith stiffness
2 i
i = 2 i
1i
2
4 Loads 73
Guidelines for Design of Wind Turbines DNV/Ris
by 10-minute mean wind speeds of about 15 configuration of the turbine, whereas Figure
m/sec. Such edgewise vibrations are 4-12 shows the vibrations for the new
undesirable from a structural point of view configuration, with increased shaft stiffness.
and need to be avoided. Edgewise vibrations Both figures are extracted from Petersen et
occur whenever the negative aerodynamic al. (1998). Note that flapwise vibrations are
damping exceeds the structural damping. equally important to consider.
The aerodynamic forces supply more energy
to the vibrations than the structural damping Recommendations concerning the design of
can absorb. An example of such vibrations, new blades and modification of old blades,
simulated for a 500 kW turbine, are shown when edgewise vibrations pose a problem,
in Figure 4-11 and Figure 4-12. Figure 4-11 are given by Petersen et al. (1998) and are
shows the vibrations for the basic reproduced in the following.
Figure 4-11. Edgewise blade root moment for the 500 Figure 4-12. Edgewise blade root moment for the 500
kW turbine in basic configuration. kW turbine with an increased shaft stiffness.
For a new turbine design, where a great deal 3. Use a quasi-steady approach (Petersen
of freedom is at hand, the following et al., 1998) to calculate the basic
procedure is recommended: aerodynamic damping characteristics
1. Base the airfoil on well-documented for a single blade, considering both the
airfoil data. fundamental flapwise and edgewise
2. Choose airfoils, blade planform (chord, mode shapes. If not satisfactory, repeat
aerodynamic twist) and blade structural from Step 2.
properties (stiffness, mass, mode shapes 4. Extend the quasi-steady analysis and
i.e. amplitude and direction of include verified models for dynamic
deformation, structural damping, modal stall on both the airfoil lift and drag,
mass, modal stiffness and modal natural still considering a single blade. If not
frequency), considering not only the satisfactory repeat from Step 2.
rotor performance, but also the 5. When the single blade analysis results
aerodynamic damping characteristics. in appropriate aerodynamic damping,
Especially, the blade twist should be continue the analysis by use of the full
considered, not only in relation to aeroelastic model, including verified
power performance, but also in relation models for stall hysteresis on both
to mode shape vibration direction. airfoil lift and drag. The structural
models must have a sufficiently detailed
74 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
4 Loads 75
Guidelines for Design of Wind Turbines DNV/Ris
Figure 4-15. Airfoil data with stall strips. Figure 4-16. Airfoil data with vortex generators.
Use of stall strips and vortex generators lift and drag coefficients. It is important to
Stall strips can be used on the outer section account for dynamic stall when it occurs,
(typically the far one-third) of a rotor blade since lift and drag coefficients are usually
as an emergency action when problems derived from wind tunnel tests with constant
with heavy vibrations of the blades occur. In inflow angle. Various models for dynamic
general, stall strips increase the damping on stall can be found in Petersen et al. (1998).
local sections of the blade by increasing the
drag after the onset of stall.
(t ) = 0 + sin( t )
4.4 Load analysis and synthesis
0 mean inflow angle 4.4.1 Fatigue loads
amplitude
t time Fatigue loads are cyclic loads, which cause
angular frequency cumulative damage in the materials of the
structural components, and which eventually
The phenomenon of delayed onset of stall is lead to structural failure. Fatigue loads are
denoted dynamic stall, and it influences the usually loads well below the load level that
76 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
will cause static failure, and many load defined according to the following rules
cycles are required before a fatigue failure (Wirsching and Shehata, 1977):
will take place. This is commonly referred to 1. A rain flow is started at each peak and
as high-cycle fatigue. However, for some trough.
materials with particularly high S-N curve 2. When a rain-flow path started at a
slopes, such as some epoxy materials, the trough comes to the tip of the roof, the
loads of importance for fatigue are close to flow stops if the opposite trough is more
those that will cause static failure. For such negative than that at the start of the path
materials fatigue becomes an extreme value under consideration (e.g., path [1-8],
problem as far as the loads are concerned path [9-10], etc.). A path started at a
with only a few load cycles required to peak is stopped by a peak which is more
cause fatigue failure. This is commonly positive than that at the start of the rain
referred to as low-cycle fatigue. path (e.g. path [2-3], path [4-5], path [6-
7], etc.).
The cumulative damage is commonly 3. If the rain flowing down a roof
determined using the Palmgren-Miner rule, intercepts flow from a previous path,
as explained in Appendix C. This method the present path is stopped (e.g. path [3-
requires knowledge of the distribution of the 3a], path [5-5a], etc.).
stress ranges, which is dealt with in the 4. A new path is not started until the path
following subsections. under consideration is stopped.
4 Loads 77
Guidelines for Design of Wind Turbines DNV/Ris
Figure 4-18. Illustration of the rain-flow counting method: (a), (b) application to a stress history, from Wirsching and
Shehata (1977); (c) application to a low-cycle strain history, from Madsen et al. (1986).
In the rain-flow method small reversals are three that identifies both slowly varying
treated as interruptions of the larger ranges, stress cycles and more rapid stress reversals
and the rain-flow method also identifies a on top of these. The peak counting method
mean stress for each stress cycle. A will in general assign larger probabilities to
comparison between test results and results larger stress ranges, while the range
predicted by the three mentioned cycle counting method will assign larger
counting methods shows that the rain-flow probabilities to smaller stress ranges.
counting method generally gives the best Compared to the rain-flow counting method,
results. This method is the only one of the the peak counting will therefore result in
78 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
(col. j)
Mean stress
Analytical results for the stress range
distribution obtained through rain-flow
counting are very difficult to obtain, and the (row i) nij
method is generally used with measured or
simulated stress histories only.
Figure 4-19. Matrix representation of rain-flow counted
The rain-flow method identifies a mean stress range distribution.
stress for each stress cycle. An attractive
representation of the resulting stress range This approach is useful for damage
distribution from cycle counting by the rain- predictions for structural materials, such as
flow method is therefore to form a matrix blade materials, for which a non-zero mean
with one row for each mean stress level and stress is of importance. For such materials,
one column for each stress range. Each the dependency on the non-zero mean stress
element of this matrix will then contain the will not be adequately accounted for by a
number of stress cycles associated with a Miners sum model, which is based on only
particular stress range and a particular mean one single S-N curve associated with full
stress. Each row of the matrix will contain a stress reversals about a zero mean stress.
discretised stress range distribution
conditioned on a particular mean stress. Stress range distributions
When S-N curves are available for various To establish the stress range distributions for
ratios R between the compressive stress a structural component in a wind turbine, it
amplitude and the tensile stress amplitude, is important to consider the various con-
this representation will allow for prediction ditions that the turbine can be in. There are
of the partial damage in each element of the various operational conditions, including:
matrix by applying the appropriate S-N production
curve for that element. The total fatigue start at cut-in and start at cut-out
damage D can subsequently be determined stop at cut-in and stop at cut-out
by summing up the partial damage over all idling and standstill
elements in the matrix, yaw misalignment
Start and stop are transient conditions, for
n ij ( S j )
D = which the stress distributions in the
i j N ij ( S j ) considered structural component are not
easily determined. During production,
in which nij denotes the number of stress stationary conditions can be assumed to
cycles in the matrix element corresponding prevail in the short term, e.g. during 10-
to the jth stress range, Sj, and the ith mean minute periods. During a 10-minute period,
stress, and Nij is the number of stress cycles the wind climate parameters such as the 10-
to failure in this element. Reference is made minute mean wind speed U10 and the
to Figure 4-19. turbulence intensity IT at the hub height can
be assumed to be constant. During such
short periods of stationary conditions, the
load response processes that give rise to the
stress ranges in the considered structural
4 Loads 79
Guidelines for Design of Wind Turbines DNV/Ris
component can be taken as stationary logarithmic diagram in Figure 4-20 had been
processes. Under stationary conditions, a straight line, the distribution would have
stress ranges are often seen to have been an exponential distribution, which is a
distributions which are equal to or close to a special case of the Weibull distribution.
Weibull distribution,
FS(s) = 1exp((s/SA)B)
250
Note that when the exponent B equals 1, this 200
distribution turns into an exponential
150
distribution, and when it equals 2 it becomes
S
a Rayleigh distribution. For representation 100
of stress range distributions, which are not 50
quite Weibull distributions, it will often suf-
0
fice to apply a somewhat distorted Weibull
0 2 4 6 8 10
distribution. The simplest distribution model
log10n
among this family of distributions is a three-
parameter Weibull distribution, Figure 4-20. Example of compound stress distribution,
a so-called load spectrum.
FS(s) = 1exp(((sa)/SA)B)
To obtain the distribution of all fatigue loads
Other models for moderate distortions of a in the design life, this compound distribution
parent Weibull distribution to form a para- has to be supplemented by the stress ranges
metric distribution model, which fits data owing to start, stop, standstill, idling, and
well, exist (Ronold et al., 1999). yaw misalignment, to the extent that these
are considered to contribute to the
The coefficients a, B and SA are distribution cumulative fatigue damage.
parameters, and they can often be expressed
as functions of the wind climate parameters For this purpose, information about the duty
U10 and IT. When the short-term stress range cycle of the turbine is essential. The duty
distributions, conditional on U10 and IT have cycle is a repetitive period of operation,
been established, when the long-term which is characterised by a typical
distributions of U10 and IT are known as succession and duration of different modes
outlined in Section 3.1, and when the total of operation. The duty cycle can be specified
number of stress cycles during the by a sequence of parameters, which define
production life of the turbine is assessed, the mode of operation for consecutive 10-
then the compound distribution of all stress minute periods for a representative time
ranges during this production life can be span. Information about the duty cycle
established. This compound distribution can should as a minimum contain the number of
itself often be represented by a Weibull starts and stops during the time span, which
distribution, and it can be represented as the duty cycle covers, together with the wind
shown in Figure 4-20, where n denotes the climate parameters in this time span.
number of stress cycles which exceed a
stress range S during the production life. Starting the wind turbine is considered to be
This distribution forms a significant potentially critical with respect to fatigue.
contribution to the loads that cause fatigue One reason for this is that connection of an
damage. Note that if the curve in the semi- electric generator to the grid can cause high
80 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
transient loads in the drive train and the neq. This is a constant load range S0, which
rotor blades. It is recommended to in neq cycles will lead to the same
distinguish between start at the cut-in wind accumulated damage as the true load
speed and start slightly below the cut-out spectrum that consists of many different
wind speed. load ranges Si and their corresponding cycle
numbers ni. When the equivalent number of
When stopping the wind turbine, it is cycles neq is chosen or specified, the
recommended to distinguish between equivalent load range S0 can be found as
stopping slightly below the cut-in wind
speed and stopping at the cut-out wind n i S im
1m
4 Loads 81
Guidelines for Design of Wind Turbines DNV/Ris
ranges. Time series of load response are Table 4-4. Quantiles of the Students t distribution
simulated for various wind climates. t
1 , n 1
Compound lifetime load spectra are 2
1 n
S0 = S 0i
n i =1
4.4.2 Ultimate loads
Extreme value distributions are of interest
when extreme load responses are needed
The standard deviation of the S0is is
such as for design against failure in ultimate
estimated by
loading. Reference is usually made to a
particular load case, e.g.:
1 n normal operation at a 10-minute mean
s= ( S 0i S 0 ) 2
n 1 i =1 wind speed near the cut-out wind speed
standstill at a rare 10-minute mean wind
A Gaussian assumption can usually be made speed such as the one with a 50-year
for the simulated equivalent load ranges S0i. recurrence period
On this background, a two-sided confidence faulty operation at a high wind speed
interval for the estimated equivalent load due to error in the protection system
range with confidence 1 can be
established as For the considered load case, it is assumed
that a total of n 10-minute time series of the
s load response X has been generated by
S0 t aeroelastic simulations. The following
n 1
2
, n 1
quantities associated with the load response
X can be interpreted from each of the n time
in which t1/2,n1 is the 1/2 quantile in the series:
Students t distribution with n1 degrees of mean value
freedom. Table 4-4 tabulates these quantiles standard deviation
for selected degree-of-freedom values for a skewness 3
couple of common choices for the kurtosis 4
confidence 1. Quantiles of the Students t rate of upcrossings of level
distribution can, in general, be found in most
maximum xm in 10 minutes
published tables of statistical distributions.
The maximum value Xm of the load response
X in the 10 minutes is of interest. The
maximum value will not be a fixed value,
but will have a natural variability, which can
82 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
There are two fundamentally different The mean value and standard deviation of
approaches to predicting the maximum load Xm are estimated by
response and particular quantiles of its
distribution: E
statistical model, which utilises the
m = + and m =
6
information about the maximum load
response obtained from the n simulated
respectively.
time series in terms of n simulated
maximum values xm
The -quantile in the distribution of Xm can
semi-analytical model, which based
be estimated by
on stochastic process theory utilises
the information about the underlying
x m , = m + k m
load response process in terms of the
four statistical moments , , 3 and 4,
and the crossing rate . in which
4 Loads 83
Guidelines for Design of Wind Turbines DNV/Ris
confidence interval for the -quantile with sided confidence interval for xm,95% becomes
confidence 1 becomes 4.673 2.780.413 = 4.673 1.148. This
indicates a rather wide interval about the
x m , t se( x m , ) central estimate. If n is changed from 5 to
1 , n 1
2 100, the interval is narrowed considerably to
4.673 0.183.
in which t1/2,n1 is the 1/2 quantile in the
Students t distribution with n1 degrees of Semi-analytical model
freedom. When a characteristic value with a The semi-analytical model owing to
specified confidence is aimed for, it is Davenport (1961) utilises more information
usually taken as the upper confidence limit, about the n 10-minute time series of the load
i.e. a one-sided confidence interval is response than just the n maximum response
considered. The characteristic value with values xm. The load response X can be
confidence 1 then becomes viewed as a stochastic process during the
10-minute series. The process X can be
x m , + t 1 , n 1 se( x m , ) considered as a quadratic transformation of a
parent standard Gaussian process U,
Note that a high number of simulations n
may be necessary to achieve a sufficiently X = + (U + U 2 ), << 1
accurate estimate of xm, or m.
A first-order approximation gives the
Example following expressions for the coefficients
Consider a load response process X whose
extreme value Xmax in 10 minutes has been 3
estimated on the basis of n = 5 simulated 10- =
6
minute time series. The following estimates =
pertaining to the extreme value distribution
=
have resulted:
from which it appears that use is made of the
= 3.69, = 3.87, m = 4.02, m = 0.35. mean value , the standard deviation and
the skewness 3 of the load response process
An estimate of the = 95% quantile of Xmax X.
with 1 = 95% confidence is sought. This
gives k = 1.866. The central estimate of the The mean value and standard deviation of
95% quantile of Xmax is Xm are estimated by
84 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
4 Loads 85
Guidelines for Design of Wind Turbines DNV/Ris
words, higher-order moment estimates based The upper bounds of the mean value m of
on available simulated time series of load the largest extreme Xmax of X in all sectors
response can usually not be trusted. during the time T are
86 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
a driving torque Me,nom = Pnom/(2nr), reduced by 33%. For design of the main
where Pnom is the nominal power of the shaft, a torque from the mechanical brake of
wind turbine, nr is the rotor frequency, two times the value of MY is to be assumed.
and is the nominal efficiency, usually
0.9. The design loads given here are compatible
the weight of the rotor mg, where m is with the Danish codes. Thus, when applied
the rotor mass and g is the acceleration for design purposes, they need to be checked
of gravity against design capacities calculated
according to the Danish codes.
The rotor loads are expressed in terms of
these three quantities as summarised in 4.5.3 Quasi-static method
Table 4-5. Moments are based on the The quasi-static method presented in the
assumption that the airflow load F0 is following gives simple expressions for four
applied with an eccentricity e = R/6. different design loads as derived for a still-
standing wind turbine in rather severe wind
For calculation of blade loads, it is assumed conditions: the blade load is calculated for a
that the airflow load is distributed evenly vertical blade above the hub. The root
between the NV blades, such that the moment is used as the design tilt moment.
flapwise airflow load on one blade becomes The axial force is determined by summing
the wind load over the blades. The yaw
Table 4-5. Design rotor loads by simplified method.
moment is determined as the root moment
Load component Sym- Static Dynamic
bol load load from a horizontal blade.
amplitude
Horizontal force FX 0 0 By the quasi-static method, the load per unit
in rotor plane length of the blade is calculated by
Moment about MX eF0 0.25eF0
horizontal axis in
rotor plane 1
p(r ) = U 10 D(r )C
2
Horizontal force FY F0 0.25F0
along rotor axis
2
Moment about MY 1.3Me,nom 0.251.3Me,nom
rotor axis in which C is the maximum value of the lift
Vertical force FZ mg 0 coefficient CL or the drag coefficient CD.
Moment about MZ eF0 0.25eF0 Typical maximum values of CL and CD are
vertical axis
in the range 1.3-1.5. U10 is the 10-minute
mean wind speed with a recurrence period
PV = F0/NV of fifty years at a height h. The height h and
the coefficient C are given in Table 4-6,
This load is assumed to be the resultant of a depending on which design load is to be
triangular-shaped flapwise line load along calculated. is the density of air, and D(r) is
the blade with the maximum value occurring the chord length of the blade at distance r
at the blade tip. from the hub.
For calculation of blade loads in the rotor
The quasi-static gust factor in the load
plane, i.e. edgewise blade loads for the
expression is to be calculated as
individual blades, the gravity loads need to
be considered. For calculation of effects on
machinery and tower, the dynamic load
amplitudes given in Table 4-5 may be
4 Loads 87
Guidelines for Design of Wind Turbines DNV/Ris
z
2
Moreover, the background turbulence effect whose value depends on which design load
kb in the expression for is approximated is being considered
by
L
0.9 2.5 l for blade load
kb =
L
0.75 3 for axial force 1
l for blade load
nL
1+ 3
U 10
and the resonance effect kr in the same 1
expression is given by F ( n) = for axial force
1 + 12 nL
n0 l U 10
nL
U 10 2 2.7
kr = F (n 0 ) U for rotor
nl 2 10
(1 + 1.5 0 ) 5 3
nL nL 2 moments
U 10 1 + 4.4 + 21.8( )
U 10 U 10
Here, = 6.8Lu, in which Lu is the integral
length scale of the Kaimal spectrum, see 4.5.4 Peak factor approach for
Section 3.1.5. = 2(0+a) is the extreme loads
logarithmic increment of damping expressed Extreme loads are often specified in terms of
in terms of the structural damping ratio 0 the load which has a certain recurrence
and the aerodynamic damping ratio a. F(n) period, such as fifty years, i.e. the load that
is the aerodynamic admittance function on average will be exceeded once every fifty
years. Two approaches are presented here.
88 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
1
Peak-over-threshold method p0 = W 2 cC L
2
By the so-called peak-over-threshold
method, the load with a recurrence period T
4 2
can be estimated by W2 =( n r R) 2 + V0
3
qT = q0+ln(T)
density of air
in which q0 is some chosen threshold for the c characteristic chord length of the blade
load q, is the mean rate of exceedances of at a distance r = 2/3R
the threshold q0 by the load q, and is the CL lift coefficient at a distance r = 2/3R
mean value of the exceedances q = qq0. W resulting wind speed
nr rotor frequency
Periodical maximum method R rotor radius
The distribution of the maximum load in a V0 nominal stall wind speed at the height
period of specified duration such as one year of the hub
can be assumed to be a Gumbel distribution
V0 is defined as the minor of the following
two wind speeds:
F(q) = exp(exp((q)))
the nominal 10-minute mean wind
speed Vnom at which the turbine reaches
When the distribution parameters and
its nominal power Pnom
refer to the distribution of the maximum
load in one year, the load with the the 10-minute mean wind speed at
recurrence period T years can be estimated which stall just extends to the entire
by blade for airflow parallel to the rotor
shaft
1 T 1
qT = ln(ln ) + ln T The load distribution along the blade is
T 1 represented as a triangular line load whose
value is 0 at the hub and p0 at the blade tip a
4.5.5 Parametrised load spectra distance R away from the hub, such that the
Parametrised load spectra are useful for value of the line load at a distance r from the
calculations where not only the extreme load hub can be calculated as
is of interest, but also the entire load
distribution, such as for design against p0 r
p=
fatigue failure. Parametrised load spectra are R
simplified and idealised load distributions
expressed in terms of a number of and such that the resulting bending moment
characteristic parameters. at the blade root becomes
4 Loads 89
Guidelines for Design of Wind Turbines DNV/Ris
with the additional condition F*(N)2k where p may denote, for example, line load
for blades. p is then the mean line load,
90 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
and pC and pS are load ranges for cosinus- The characteristic frequency nC to be used is
oidal and sinusoidal load components, also indicated and expressed in terms of the
respectively. rotor frequency nR.
4 Loads 91
Guidelines for Design of Wind Turbines DNV/Ris
illicit
0,4
frequency 4.6 Site-specific design loads
0,2 range
As part of a scheme to reduce the cost of
0
0 1 2 3 4 5 6 7 8
produced energy, site-specific characteristics
n 0/n R can be included directly in the design
process of wind turbines. The idea is to
Figure 4-22. Amplification factor kR vs. frequency ratio optimise the wind turbine design by
n0/nR. minimising the cost of produced energy,
given the characteristics at a particular site.
Table 4-8 gives expressions for determin- Extreme loads and fatigue loads on a wind
istic and stochastic load components for turbine are site-dependent. Site-specific
rotor loads. The expressions are based on the design loads will therefore form part of such
blade line load p0, the nominal torque Mnom, a cost-optimal design process.
the rotor mass M, and the standardised load
92 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
Two approaches to site-specific designs are foundation structure will be exposed to wave
envisaged: loads, current loads and ice loads. The
adaptation of existing wind turbine prediction of wave loads is dealt with in the
designs to specific site by minor following. A brief introduction to current
adjustments loads and ice loads is also given.
design from scratch
4.7.1 Wave loads
Site-specific design has a potential Wave climate
whenever wind turbines are to be installed in Usually, the wave climate at a location can
complex terrain such as in mountainous be considered stationary within periods of
areas, where special conditions may prevail, typically three hours duration. The wave
or for large offshore wind farms where a climate is represented by the significant
large number of identical turbines are to be wave height HS and the peak period TP. The
installed at the same site. It can be used to significant wave height HS is a measure of
verify that all relevant load cases have been the intensity of the wave climate and is
considered. defined as four times the standard deviation
of the sea elevation process . Some sources
Using site-specific design loads and carrying
define the significant wave height as the
out site-specific wind turbine designs is
average of the highest one third of the wave
somewhat in contrast with the current trend
heights. For a narrow-banded Gaussian sea
within the wind turbine industry. In order to
elevation process, the two definitions
keep down manufacturing costs, the current
converge. The peak period TP is related to
trend is not to site-optimise wind turbines,
the mean zero-crossing period of the sea
but rather to produce a selection of standard
elevation process. The significant wave
wind turbines. The task is then to choose a
height and the peak period can be taken as
standard wind turbine from this selection
constant within each three-hour period.
and verify that it is suitable for a given
location. The tower and the foundation may
Long-term distributions of HS and TP are
still be site-optimised if desirable, and site-
site-dependent. The long-term distribution
specific loads will be required for this
of HS can often be represented well by a
purpose. The foundation design will always
Weibull distribution, whereas the
have to be site-specific in that it needs to be
distribution of TP conditioned on HS is
designed for the prevailing local soil
usually well-represented by a lognormal
conditions.
distribution whose distribution parameters
are functions of HS. Examples can be found
Reference is made to Section 3.1.10 on the
in Bitner-Gregersen and Hagen (2000).
subject of site assessment.
Wave spectrum
The frequency content of the sea elevation
4.7 Loads from other sources than
process can be represented by the power
wind
spectral density. The spectral density of the
Installation of wind turbines in shallow sea elevation process can be represented by
waters of up to 15 m water depth can be the JONSWAP spectrum
foreseen. A support structure or foundation
structure is used for transfer of loads from
the wind turbine and its tower to the
supporting soils at the seabed. The
4 Loads 93
Guidelines for Design of Wind Turbines DNV/Ris
5 T
5
The maximum wave height Hmax during
S ( ) = H S T P P
2
some time span TL is often of interest for
32 2 design. Let N denote the number of zero-
4
5 T upcrossings of the sea elevation process in
exp( P ) this period of time, i.e. N = TL/TZ. The
4 2 distribution of Hmax can then be
1 T P
2
exp
2 2 2
1
approximated by
C ( )
2h 2
C() = 10.287ln FH max (h) exp( N exp( ))
(1 2 ) H S 2
= 0.07 for 0 < < 2/TP
= 0.09 for > 2/TP and the expected value of the maximum
wave height can be approximated by
The peak-enhancement factor can be taken
as (1 2 ) ln N
E [H max ] H S
2
TP TP
= exp(5.75 1.15 ) ; 3.6 5
HS HS A first-order approximation yields the
following value for the spectral width
in which HS is in metres and TP in seconds. parameter, = 0.43. With this value of ,
The following approximate relationship the following relationship between the
exists between the peak period TP and the maximum wave height Hmax and the
zero-crossing period TZ maximum wave crest Zmax holds
Hmax 1.8Zmax
5+
TZ = TP
11 + In shallow waters, shoaling effects imply
that wave crests become more peaked while
Reference is made to Gran (1992) and DNV wave troughs become flatter and not quite as
(2000). deep. The sea elevation process has become
somewhat skewed and will not quite be
Wave heights Gaussian, and the individual wave heights
In deep waters, the sea elevation process will not quite be Rayleigh distributed.
is a Gaussian process, and the individual Techniques are available to account for the
wave heights H, measured form trough to skewness introduced by the shoaling.
crest, will follow a Rayleigh distribution Reference is made to Winterstein et al.
when HS is given (1991) and to U.S. Army Coastal
Engineering Research Center (1973).
2h 2 Note that in shallow waters, the wave
FH (h) = 1 exp( )
(1 2 ) H S 2 heights will be limited by the water depth, d.
The maximum possible wave height at a
spectral width parameter water depth d is approximately equal to the
water depth
94 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
H max,lim d
4 2 2kd
h0 = H max (1 + ) tanh[kd ]
and the Rayleigh distribution of the wave HS sinh[2kd ]
heights will become distorted in the upper
tail to approach this limit asymptotically. in which k is the wave number, which
results implicitly as the solution to
Wave periods
Once the significant wave height HS is
given, the zero-upcrossing period TZ is
2 = gk tanh[kd ]
usually well-represented by a shifted
lognormal distribution where L = 2/k and = 2/T
4 Loads 95
Guidelines for Design of Wind Turbines DNV/Ris
According to first-order linear wave theory, when a nodal line at the stillwater level
the horizontal wave-induced velocity is passes the structure. The drag force FD, on
the other hand, reaches its maximum when
cosh[k ( z + d )] the crest passes the structure, and if this
x& = AW sin(t )
sinh [kd ] force is dominating, a significant error can
be introduced by ignoring the contribution
from the wave crest.
and the acceleration is
Note also that Morisons equation is only
cosh[k ( z + d ]) valid when the dimension of the structure is
&x& = AW 2 cos(t )
sinh [kd ] small relative to the wave length, i.e. when
D < 0.2L, and that it is only valid for
AW wave amplitude nonbreaking waves. In deep water, waves
break when H/L exceeds about 0.14.
The resulting horizontal force F on the The inertia coefficient depends on the cross-
cylinder can be found by integration of sectional shape of the structure and of the
Morisons equation for values of z from d orientation of the body. Typically, CM is in
to 0 the range 1.6-2.5. For a vertical cylinder,
CM = 2.0. For a cylinder with in-service
F = FM + F D marine roughness, e.g. owing to marine
0 growth, CM should normally not be less than
D2 H 2
= (C M cos(t ) 1.8. The drag coefficient CD is never less
d 4 2 than 0.6, and for a smooth cylinder CD = 1.0.
cosh[k ( z + d ])
)dz
sinh[kd ] An example of application of Morisons
0
equation is given in Figure 4-24 for a
D H2 2 cylinder with 4 m diameter in 10 m water
+ (C D sin(t ) sin(t )
d 2 4 depth. The upper part of this figure shows
cosh 2 [k ( z + d )] the sea elevation process, the middle part
)dz shows the horizontal water particle velocity
sinh 2 [kd ]
and acceleration at the stillwater level, and
the lower part of the figure shows the
Note that the integration from d to 0 resulting horizontal wave force and
ignores contributions to the force from the overturning moment at the seabed. The
wave crest above the stillwater level at z = 0. example is based on CM = 2 and CD = 1.2.
However, this is a minor problem for the
inertia force FM, since this has its maximum
96 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
FM = AM cos(t )
FD = AD sin(t ) sin(t )
H C sinh 2 [kd ]
= M
D C D ((sinh[2kd ] / 4 + kd / 2) A
4 Loads 97
Guidelines for Design of Wind Turbines DNV/Ris
With reference to Figure 4-23, structures amplitude A, this theory gives the following
located above the curve for A = 1 experience maximum horizontal wave force
loads which are dominated by the drag term
in Morisons equation. Structures located 4 gA sinh [k (d + A sin )]
F X , max =
below the curve for A = 1 are dominated by k2 tanh[kd ]
inertia loads. The two dashed asymptotes in
Figure 4-23 are valid for shallow water
whose vertical arm measured from the sea
waves (d/L < 1/20) and deep water waves
floor is
(d/L > 0.5), respectively. The asymptotes are
derived from asymptotic results for small-
amplitude wave kinematics. kd sinh [kd ] cosh[kd ] + 1
hF = d
kd sinh[kd ]
Example
A cylinder with diameter D = 5 m is The coefficients and are given in Table
considered subjected to the wave climates at 4-10, extracted from Gran (1992).
the Middelgrunden and Rdsand sites in
Danish waters. Water depths and wave data Table 4-10. Coefficients and .
for a number of cases at these two locations
are tabulated in Table 4-9 with the resulting
ratios H/D, d/L and A. It appears that for
H/D, d/L and A. It appears that for both sites
A values occur, which indicate that the
inertia force dominates the loading.
98 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
gravity-based structures. Note, however, that For ice loads in general, reference is made to
the formulas may lead to erroneous results if API (1995).
the structural geometry deviates much from
the assumed cylindrical shape, such as when 4.7.4 Earthquake loads
a conical structural component is present in
For prediction of earthquake loads, reference
the wavesplash zone to absorb or reduce ice
is made to Section 3.2.8.
loads.
4 Loads 99
Guidelines for Design of Wind Turbines DNV/Ris
of wind and wave climates of high that occurs for this climate over its duration.
intensities in design. In a probabilistic Practically, one may for example consider
analysis, this can practically be done by the wave climate for the significant wave
modelling one of the climate variables as a height with a 50-year recurrence period in
so-called independent variable by means of combination with a wind climate
its marginal cumulative distribution conditioned on this wave climate, e.g. the
function, and then model the other variable expected value of U10 conditional on the 50-
as a dependent variable by means of a year significant wave height, or some higher
distribution conditioned on the independent quantile of U10. With such a rare
variable. For the considered wind and wave characteristic wave and wind climate in
example, one could represent the significant mind, the following paragraphs outline how
wave height HS by its marginal long-term the combined load response, e.g. the
distribution, typically a Weibull distribution, horizontal force in some section of the
and then model the 10-minute mean wind foundation structure, during this climate can
speed U10 conditional on HS. The be found.
distribution of U10 conditioned on HS will
typically be a lognormal distribution For linear load combinations, Turkstras rule
plays a central role. The rule states that the
ln u b1 maximum value of the sum of two
FU10 | H S (u ) = ( ) independent random processes occurs when
b2
one of the processes has its maximum value.
Application of Turkstras rule to the
standard normal distribution combination of two load processes, e.g.
function wave load and wind load, implies that the
b 1, b 2 functions of the significant wave combined load will have its maximum either
height HS, i.e. b1 = b1(HS) and b2 = when the wave load has its maximum or
b2(HS when the wind load has its maximum. Let
Q1 and Q2 denote the two load processes.
In some cases, other generic distribution Mathematically, the maximum combined
types than the lognormal distribution may load Qmax over a time span T will be
provide the best representation of U10
conditional on HS, e.g. a Weibull
distribution. max Q1 (t ) + Q2 (t )
Qmax = max 0t T
Q1 (t ) + max
0 t T
Q2 (t )
Once the wind and wave climates are given
in terms of a specific set of concurrent U10
Turkstras rule indicates that a natural code
and HS values, the wind speed process format for a combination of two loads for
conditioned on U10 and the wave process use in deterministic design is
conditioned on HS can be considered
independent. It is therefore, for example, not
reasonable to expect that the maximum wind q + 2 2 q 2 k
q design = max 1 1k
speed will occur at the same time as the 1 1 q1k + 2 q 2 k
maximum wave height.
in which q1k and q2k are characteristic values
For design, it is reasonable to consider some of Q1 and Q2, 1 and 2 are associated partial
relatively rare combination of wave and
safety factors, and the -factors are load
wind climate as the characteristic climate combination factors.
and then to find the maximum load response
100 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
For a wind turbine structure, loads do not Ris R-1065(EN), Ris National
always combine linearly to give the sought- Laboratory, 1999a.
after maximum load response. Aeroelastic
wind load calculations may be nonlinear, Bak, C., H.A. Madsen, and N.N. Srensen,
and the combined wave and wind load Profilkoefficienter til LM19.1 vingen
response may not necessarily come out as bestemt ud fra 3D CFD, 1999b.
the linear combination of the separately
calculated wave load response and the Bathe, K.J., Finite Element Procedures in
separately calculated wind load response. Engineering Analysis, Prentice-Hall, 1982
For such nonlinear cases, the combined
response is to be calculated from some Baumgart, A., I. Carln, M. Hansen, G.
appropriate structural analysis for the Larsen, S.M. Petersen, Experimental Modal
concurrent characteristic wave and wind Analysis of a LM 19.1 m blade, unpublished
load processes without applying any partial work.
load factors. The resulting maximum load
response from this analysis can be Bitner-Gregersen, E., and . Hagen, Aspects
interpreted as a characteristic load response, of Joint Distribution for Metocean
which reflects the combined wave and wind Phenomena at the Norwegian Continental
loading. A common partial safety factor is Shelf, ASME Paper No. OMAE-2000-6021,
then to be applied to this characteristic load Proceedings, International Conference on
response to give the design value qdesign, i.e. Offshore Mechanics and Arctic Engineering,
it will no longer be possible to distinguish 2000.
between different partial safety factors for
wave loads and wind loads. Davenport, A.G., The Application of
Statistical Concepts to the Wind Loading of
Structures, Proc. Inst. of Civil Engineers,
REFERENCES Vol. 19, 1961.
4 Loads 101
Guidelines for Design of Wind Turbines DNV/Ris
Hallam, M.G., N.J. Heaf, and L.R. Petersen, J.T., Geometric nonlinear finite
Whootton, Dynamics of Marine Structures: element model for a horizontal axis wind
Methods of Calculating the Dynamic turbine, Ris National Laboratory, 1990.
Response of Fixed Structures Subject to
Waves and Current Action, CIRIA Petersen, J.T., H.A. Madsen, A. Bjrk, P.
Underwater Engineering Group, 6 Storeys Enevoldsen, S. ye, H. Ganander, D.
Gate, London SW1P 3AU, Report UR8, Winkelaar, Prediction of Dynamic Loads
1978. and Induced Vibration in Stall, Ris-R-
1045, Ris National Laboratory, 1998
Haver, S., On a Possible Lower Limit for the
Spectral Peak Period, Statoil Report No. Ralston, T.D., Ice Force Design
F&U-MT 90009, Stavanger, Norway, 1990. Considerations for Conical Offshore
Structures, Proceedings, Fourth POAC
IEA, Expert Group Study on Recommended Conference, Vol. 2, pp. 741-752, St. Johns,
Practices for Wind Turbine Testing and Nfld., Canada, 1977.
Evaluation, 3. Fatigue Loads, 2nd edition,
1990. Ronold, K.O, J. Wedel-Heinen, and C.J.
Christensen, Reliability-based fatigue design
Larsen, G., and P. Srensen, Design Basis of wind-turbine rotor blades, Engineering
2, Proceedings, IEA Symposium State-of- Structures, Elsevier Science Ltd., Vol. 21,
the-Art of Aeroelastic Codes for Wind No. 12, pp. 1101-1114, 1999.
Turbine Calculations, pp. 137-145,
Lyngby, Denmark, 1996. Stiesdal, H., Rotor Loadings on the
BONUS 450 kW Turbine, Proceedings,
Madsen, H.O., S. Krenk, and N.C. Lind, EWEC91, Amsterdam, the Netherlands,
Methods of Structural Safety, Prentice-Hall 1991.
Inc., Englewood Cliffs, N.J., 1986.
U.S. Army Coastal Engineering Research
Madsen, P.H., K. Pierce, and M. Buhl, The Center, Shore Protection Manual, Vols. I-
use of aeroelastic wind turbine response III, Washington, D.C., 1973.
simulations for prediction of ultimate design
loads, Proceedings, 3rd ASME/JSME Joint Veers, P.S., Three-Dimensional Wind
Fluids Engineering Conference, Paper No. Simulation, Report No. SAND88-0152,
FEDSM99-S295-10, San Francisco, Cal., Sandia National Laboratories, Albuquerque,
1999. N.M., 1988.
Mann, J., Wind field simulation, Viterna, L.A. and R.D. Corrigan, Fixed pitch
Probabilistic Engineering Mechanics, rotor performance of large horizontal axis
Elsevier Science Ltd., Vol. 13, No. 4, pp. wind turbines, DOE/NASA Workshop on
269-282, 1998. Large Horizontal Axis Wind Turbines,
Cleveland, Ohio, July 28-30, 1984.
Mann, J., The spatial structure of neutral
atmospheric surface-layer turbulence, Winterstein, S.R., Nonlinear Vibration
Journal of Fluid Mechanics, No. 273, pp. Modes for Extremes and Fatigue, Journal
141-168, 1994. of Engineering Mechanics, ASCE, Vol. 114,
No. 10, pp. 1772-1790, 1988.
102 4 Loads
Guidelines for Design of Wind Turbines DNV/Ris
4 Loads 103
Guidelines for Design of Wind Turbines DNV/Ris
104 5 Rotor
Guidelines for Design of Wind Turbines DNV/Ris
the twist of a blade. Note in this context that direction of rotation at the root than at the
the pitch angle, which is referred to tip. Relative to the rotor axis, the force at the
throughout this document, usually refers to root has a smaller torque arm than the force
the collective rotation of the entire blade at the tip and will therefore give about the
relative to the rotor plane. same contribution to the starting torque as
the force at the tip.
For a blade, four non-coincident trajectories
can be defined: During operation, the wind approaching the
mass axis, the spanwise locus of section blade profile constitutes the vectorial sum of
mass centres the farfield wind speed perpendicular to the
elastic axis, the spanwise locus of points rotor plane and the head wind due to the
about which no section is exposed to rotational movement of the blade through
bending deflection the air. Smaller aerodynamic forces are
control axis, the axis of mechanical produced near the root than at the tip.
feathering, which is determined by the However, the forces produced near the root
blade retention and pitching mechanism are more aligned with the direction of
aerodynamic axis, the blade section rotation than the forces near the tip. The
quarterchord for a conventional airfoil change in magnitude and direction of the
shape within linear performance limits forces along the blade contributes to
These four trajectories are not axes in the determine the shape and design of the blade,
true sense, since they owing to the including the width, thickness and twist of
geometry of the blade are not straight the blade.
lines. However, for calculations for a
particular section of the blade, four axes, The design loads for a blade can be
perpendicular to the section, can be defined determined by means of blade momentum
as the tangents to the trajectories at their theory and aeroelasticity. For details,
respective intersections with the section. reference is made to Chapter 4.
In principle, the same airflow conditions Wind turbine blades are made of lightweight
would apply at all sections along a blade as materials to minimise the loads from
long as the profile stays the same, while the rotating mass.
magnitude and direction of the forces would
change depending on the distance to the tip. A rotor blade is built up of the following
However, in practice, the profile and the elements:
blade thickness vary along the blade and external panels - form the aerodynamic
thereby make the airflow conditions more shape and carry a part of the bending
complex. load
internal longitudinal spars/webs carry
At stand-still, the wind pressure force will shear load and a part of the bending
depending on the load case be somewhat load, restrain the cross section against
larger at the root than at the tip. This is due deformation and the panels against
to the fact that the blade is wider at the root. buckling.
The force is acting roughly at a right angle inserts like bushings - transfer the loads
to the flat side of the blade profile. As the from the panels and spars into the steel
blade is more twisted at the root, a larger hub.
component of the force will act in the
5 Rotor 105
Guidelines for Design of Wind Turbines DNV/Ris
lightning protection carries a lightning Carbon fibre is used for aerodynamic brake
hitting the blade tip to the root shafts.
aerodynamic brake for some types of
turbines with fixed pitch an Glass and carbon fibres come in different
aerodynamic brake is part of the types having different chemical com-
protection system. The aerodynamic positions. The most important type of glass
brake is typically the tip turning on a is E-glass.
shaft.
The mechanical properties, modulus and
Fibre reinforced plastics (FRP) is the tensile strength, may vary significantly
favoured group of materials for external between different types of fibres and can
panels, internal spars/webs and shafts for also vary considerably within each type of
aerodynamic brakes. fibre. There are for example low, medium
and high strength carbon fibres. The
FRP is materials where fibres are used for selection and qualification of fibres is part of
transferring the global loads and a polymeric the blade design.
resin is used to distribute the load between
the fibres and to restrain the fibres against It is important to realise that the final
local relative displacements. properties of a FRP laminate not only
depend on the properties of the fibres. The
FRP is often used on both sides of a cellular properties of the matrix, the structural
core in a sandwich structure. The core in a interaction between fibres and matrix and
sandwich structure is used for increasing the the volume fraction of fibres are also
local bending stiffness of the panel or web. important.
An increased local bending stiffness may be
required to avoid buckling when subjected Sizing is deposited on the surface of the
to compressive or shear loading. fibre during production. The purpose of the
sizing is to increase the strength of the bond
Gelcoats and topcoats are used to protect the between the fibre and the resin. Sizing can
FRP against abrasion, UV radiation and also be applied to enable easy handling
moisture and to give the right colour. without inducing defects in the fibres during
blade manufacturing. A particular sizing that
Wood is a natural material with the same has been developed for one type of
type of structure as FRP. Wood is used as an application or one type of resin is not
alternative to FRP in design of blade panels necessarily adequate for another application
and spars/webs. or another resin. It is important that the
supplier of the fibre material provides
Further information about FRP can be found adequate information about this and that the
in Mayer (1992), in DNV (1999) or on supplier's recommendations are followed
www.marinecomposites.com. during blade manufacturing.
106 5 Rotor
Guidelines for Design of Wind Turbines DNV/Ris
Fabrics are defined, apart form the type and Foams are based on thermoplastics, e.g.
stacking sequence, also by the total weight PVC, and are delivered in a range of
of reinforcement per unit area, usually densities. The most important mechanical
expressed as g/m2. properties are the shear modulus, shear
strength and ductility or yielding behaviour.
Note that carbon fibres used as Stiffness and strength increases with the
reinforcement of the rotor blades will cause density. Different types and grades of foam
galvanic corrosion of any steel parts they get can have significantly different properties.
in touch with, unless stainless steel is used Detailed information on the properties is
for the making of these parts. given in the manufacturer's product
specification. PVC foams may be supplied
in a "heat stabilised" condition, which
5 Rotor 107
Guidelines for Design of Wind Turbines DNV/Ris
108 5 Rotor
Guidelines for Design of Wind Turbines DNV/Ris
lamination and the new bond shall be treated The advantage of this technique is the good
as a new bond. The surface preparation for a control of, and the relatively high, volume
new bond is critical since there will be no fraction of fibres one achieves and thus
effective crosslinking between the consistency in the properties of the
laminates. The new applied resin will in this laminates and a lower laminate weight.
case act as an adhesive.
It is important that the gel time is
Wet lay-up may be carried out with vacuum sufficiently long such that the resin has time
bagging where an airtight (plastic) mem- to infuse the whole mould before it gels.
brane is applied over the hand laid-up lami-
nate or core. The air under the membrane is Filament winding
sucked out by means of a pump such that an Shafts for aerodynamic brakes are typically
overpressure (from the atmosphere) acts manufactured by filament winding where
over the entire laminate. fibres or tapes are pre-impregnated with a
resin and wound on a rotating mandrel.
The vacuum bagging technique is used to
improve the compacting of the laminate and Internal spars can also be fabricated by
thus to increase the volume fraction of filament winding
reinforcement and to improve the
consistency of the finished laminates. 5.1.5 Quality assurance for blade
Normally a porous mat is applied on top of design and manufacture
the laminate and under the membrane to
The blade manufacturer, not the supplier of
facilitate even resin distribution and the
the raw materials, controls the final
escape of the air. Vent holes may be
properties of the laminates.
required through the core for the same
purpose.
The blade designer shall consider the
qualification of FRP materials. Properties to
Pre-pregs
be considered in this context include, but are
Plies are supplied as fabric saturated with
not limited to:
resin that will cure when heated. The plies
are stored cold. The complete stack of plies stiffness, ultimate and fatigue strength
is laid-up in the mould. The laminates in the at relevant temperatures
mould is vacuum bagged and cured by toughness (at low temperatures if
heating the mould. The applied vacuum appropriate)
consolidates the laminate and removes creep
entrapped air. ageing characteristics (considering
humidity and temperature)
Vacuum-assisted resin transfer moulding resistance of wood to rot and fungus
(VARTM)
The technique has similarities to vacuum- Note that for some materials, some of these
bagging. The complete set of reinforcement properties may not be relevant. Guaranteed
plies is applied to the mould, but without property values may be given in terms of:
adding the resin. The airtight membrane is manufacturers nominal value,
put on top and vacuum drawn. The resin is manufacturers specified value, or
then transferred via piping/hoses from a manufacturers specified minimum
container of premixed resin by the suction value.
created by the vacuum.
5 Rotor 109
Guidelines for Design of Wind Turbines DNV/Ris
A type approval of a given material, e.g. in For each load case, a set of design loads is
accordance with DNV standards, can be established by multiplying the relevant
used as part of the quality control and design characteristic loads by partial safety factors
documentation, but is in itself not sufficient for load. The standards applied for this
for approval of the material for its intended purpose should be quoted when the design
use. loads are documented. In principle, each
load case can be defined in terms of six load
Full-scale tests of a sample blade are components and their variation over the
required to verify the strength of the blade, blade span. The resolution used to specify
statically as well as in fatigue as not all this variation must be fine enough to allow
aspects can be covered by the material for sufficiently accurate calculations in all
qualification. points of interest along the blade, especially
in all critical areas, e.g. wherever changes in
FRP material specimens for testing shall be geometry or material occur.
manufactured with a curing cycle that is
representative for the blade and have a Environmental conditions, which affect the
representative fibre to resin ratio. material behaviour, should be considered
and taken into account. In particular, such
Test specimens should be wide enough to conditions include humidity and tempera-
cover at least four repetitions of the structure ture, which may both lead to degradation of
of the weave/fabric/mat. strength and stiffness, and their design
effects calculated by multiplying charac-
Ultimate strength of FRP and wood shall be teristic effects by appropriate partial safety
investigated in both tension and factors should be applied in the strength
compression. Fatigue testing shall cover analysis.
both the effects of stress width and the mean
stress. Loads on critical components such as tip
brakes are often different in character from
The fibre alignment is important for the load the general loads on the blades and may thus
carrying capacity of FRP in compression. need extra attention.
The work instructions for blade manu-
facturing shall control the fibre alignment. As blades are getting larger, large
unsupported panels will be present between
5.1.6 Strength analyses the webs and the leading and trailing edges.
This may have an impact on the stability of
Structural analyses of the rotor blades must
the blades. Therefore, the buckling capacity
be carried out for all relevant load cases in
of a blade must be verified by a separate
order to verify that the strength of the blades
calculation, in addition to the full-scale test.
is sufficient to withstand the loads that these
For this calculation, a FEM analysis will
load cases exert on the blades. By the
normally be required. Furthermore, buckling
strength calculations in these analyses, it
of the webs may also have to be considered.
must be verified that both the ultimate
strength and the fatigue strength, for a given
As rotor blades become longer, evaluation
design life, are sufficient. For structural
of the stability against buckling becomes
parts in compression, stability against
still more important because of the large
buckling must also be considered.
unstiffened panel segments, which are
usually involved.
110 5 Rotor
Guidelines for Design of Wind Turbines DNV/Ris
5 Rotor 111
Guidelines for Design of Wind Turbines DNV/Ris
lower tail of the distribution, e.g. the 2% or components should be given in all points of
5% quantile. Note that different standards relevance along the blade, including phase
may define the characteristic value and frequency information. This
differently, i.e. it is not always defined as requirement is automatically fulfilled when
the same percentile in the different the six load components are given as time
standards. Accordingly, different standards series. From the time series of the six load
may prescribe different partial safety factors components, long-term stress distributions
to be used with their respective can be established in all points of relevance.
characteristic values for design. As an In principle, this includes distributions of the
example, DS472 defines the characteristic mean stress as well as distributions of the
value of an FRP material as the 5% quantile stress range that represents the variation
and requires a materials factor of 1.7, while about the mean stress. Rain-flow counting is
DS456 defines the characteristic value as the a commonly used method for this purpose.
10% quantile and requires a materials factor The total number of stress cycles in the
of 1.8. In design, it is essential that the design life can also be extracted from the
characteristic value of a strength-property in time series and can be used to transform the
question is combined with the correct partial stress range distribution into a design
safety factor. It is not licit to combine the lifetime histogram of stress ranges. A
characteristic value of one standard with the sufficiently fine discretisation of the stress
partial safety factor of another as this can range axis needs to be chosen for this
lead to erroneous results and unsafe designs. purpose.
112 5 Rotor
Guidelines for Design of Wind Turbines DNV/Ris
5 Rotor 113
Guidelines for Design of Wind Turbines DNV/Ris
case peeling at the back edge. Lightning blades, i.e. blades longer than about 20 m, it
protection of rotor blades can, in principle, may be necessary to secure the blades
be designed in two different ways: against damage from lightning that strikes
the lightning can be prevented from the blade in other locations than the blade
penetrating into the blade by diverting tip. Note also that it is necessary to account
the current along a prepared current for the limited conductivity of the carbon
path, such as a conductive tape, on or in fibre materials that are used in tip shafts.
the surface of the blade. From the aircraft industry, methods exist for
the lightning can be directed from the lightning protection of glass fibre and
point of stroke on the blade through the carbon fibre materials by which these
blade by way of a conductor cable. materials are made electrically conductive
by means of metallic sheets, nets, or threads,
rather than by mounting metallic conductors
on the material surfaces. Reference is made
to DEFU (1999). Examples are shown in
Figure 5-4.
Figure 5-4. Examples of how lightning protection of For every new type of blade to be manu-
blades can be arranged From DEFU (1999). factured, one blade is to be tested statically
and dynamically to verify the strength of the
The latter approach is the most common blade design.
approach. By this approach, a lightning
conductor is mounted in the blade tip, and The production of a test blade is to be
through a commutator lug in the interior of inspected in order to verify that the blade is
the blade, the current is directed to the blade representative of the blade design.
root. The effectiveness of this method is Alternatively, the blade may be sampled at
expected to be highly dependent on the size random from the blade production. How-
of the blade and on the amount of metals or ever, this alternative approach is usually not
carbon fibres used in the blade. Note that a feasible, because it is often desirable to carry
lightning protection like the ones presented out the blade test in the early phases of a
here cannot be expected to work in all cases. production. Therefore, the test blade is often
A blade may be hit by many lightning selected on a deterministic basis as one of
strokes over its design life, and there may be the first blades in the production series.
cases for which the protection will fail, e.g.
if the conductor cable fuses. For large
114 5 Rotor
Guidelines for Design of Wind Turbines DNV/Ris
The test blade is to be equipped with strain Furthermore, the lowest natural frequencies
gauges and displacement transducers. The and corresponding damping ratios of the
measurement results from the strain gauges blade are measured as these are important
are continuously monitored on computers. input parameters for the load calculations.
Nonlinear variations in the pattern of This is achieved by excitation of the blade at
bending may reveal a damage in the rotor different frequencies and in different
blade structure. directions. It is essential that the natural
frequencies of the blade do not coincide
The blade is to be tested statically by a static with the rotational frequencies of the wind
load in two opposite directions flapwise and turbine. The purpose of this part of the test
in two opposite directions edgewise. The is to make sure that the natural frequencies
static load must at least equal the extreme of the blade differ from the rotational
design load. Tests in two opposite directions frequency with sufficient margin. Note that
instead of just in one direction are necessary the aerodynamic damping is also measured,
due to lack of symmetry in the blade. in particular, in the flapwise direction.
Measurements by means of the strain gauges
and displacement transducers are used to In case the natural damping of the blade is
monitor and verify that the strains and insufficient to avoid vibrations, e.g.
displacements stay within the design limits edgewise vibrations, a damper may be built
and design calculations during the entire into the blade. The damper may consist of
test. Nonlinearity in the measured strains materials with high internal damping, which
may indicate buckling or damage to the is built into the blade, or it may be a
blade. mechanical damper. It is essential that this
damper is also included in the test blade,
Dynamic testing is also carried out for since the design of the connection between
loading in both flapwise and edgewise the damper and the blade may introduce
directions. Normally, the blade is tested for weak areas or large stiffness changes.
loading in one direction at a time, but a load
with simultaneous components in both A rotor blade can also be tested for its
directions may also be applied. The dynamic residual static strength (and thus its ability to
testing is carried out as an accelerated test, withstand extreme loads in the long term) by
i.e. at a load level, which causes the same being bent once by a very large force. The
damage to the blade as the true load magnitude of this force is usually
spectrum, including a test factor of 1.3. determined by the load level of the static test
Normally, the test is carried out at a fixed described above. The residual strength test is
load level and for a number of load cycles of usually performed after the blade has been
between 2106 and 5106 cycles. subject to fatigue testing, and the purpose is
to verify that the static strength of a blade,
Infrared cameras can be used to reveal local which has been in operation for a substantial
build-up of heat in the blade. This may amount of time, is sufficient and that the
either indicate an area with structural stiffness although it may have changed
damping, i.e. an area where the blade is still acceptable.
designer has deliberately laid out fibres
which convert the bending energy into heat A test specification for the blade test should
in order to stabilise the blade, or it may be worked out. The specification should
indicate a zone of delamination, or a zone specify the loads and the measurement set-
where the fibres are close to failure. up, including means of avoiding overload of
5 Rotor 115
Guidelines for Design of Wind Turbines DNV/Ris
116 5 Rotor
Guidelines for Design of Wind Turbines DNV/Ris
5 Rotor 117
Guidelines for Design of Wind Turbines DNV/Ris
The layout of a wind turbine hub often treatment of treads, e.g. waxing
makes it difficult to determine which section torque to be applied
is the structurally most critical section of the The relation between bolt torque and bolt
hub. In this context, the Finite Element tension may be subject to test.
Method (FEM) forms a suitable tool for
strength analysis of the hub and can be used 5.2.4 Hub enclosure
in conjunction with state-of-the-art fatigue
analysis techniques to determine the fatigue The hub enclosure, which is sometimes
life and to optimise the design with respect referred to as the nose cone, is usually made
to strength and cost. It is advisable to qualify of glass fibre reinforced polyester. In cases
the loading assumptions for such FEM where the hub enclosure is large, it is
analyses by measurements. recommended to consider the wind load it
will be exposed to.
The FEM analyses can, in particular, be
used to document that a satisfactory strength 5.2.5 Materials
is available in critical sections, such as at Spheroidal graphite cast iron, also known as
stress concentrations and stiffness nodular cast iron, is the preferred material
transitions, as well as at the shaft-hub for the hub. Cast iron is classified according
connection and the interfaces to the blades. to its mechanical properties, such as strength
and hardness, in EN1563. Cast hubs are
5.2.3 Analysis of bolt connections usually tested by non-destructive testing
The blades are usually bolted to the hub. (NDT) for verification of the mechanical
There are two techniques for mounting the properties and for detection of possible
bolts in the blades: defects and internal discontinuities. The
a flange is established at the blade root following NDT methods are available:
by moulding the glass fibre reinforced ultrasonic inspection
plastics to form a ring, in which steel magnetic particle inspection
bushes for the bolts are embedded. visual inspection
treaded steel bushes are mounted hardness measurements
directly into the blade root and fixed to Ultrasonic inspection can be carried out as
the blade by glue. point testing or, more thoroughly, as
In both cases, the bolts from the blade pass complete scanning. For ultrasonic
through a flange on the cast hub. The bolt inspection, it is common to assign different
holes in this flange can be made somewhat acceptance criteria, e.g. in terms of different
elongated to enable adjustment of the tip allowable defect sizes, to different areas of
angle. the cast hub. These areas and the assigned
acceptance criteria should be indicated on
For these bolt connections, a bolt tension the drawing of the casting. Usually, a strict
procedure is usually required. Such a acceptance criterion is assigned to an area
procedure should usually specify: with high stresses, whereas a more lax
bolt, nut, washer type, dimension and acceptance criterion is assigned to an area
quality with low stresses. The stricter the
flatness tolerances for surfaces acceptance criterion is in a particular area,
roughness of surface the more thorough is the ultrasonic
inspection in that area, and the stricter are
surface treatment and protection
the requirements to the allowable size of
bolt tensioning sequence and method
detected discontinuities.
118 5 Rotor
Guidelines for Design of Wind Turbines DNV/Ris
5.2.6 Standards
EN1563 Founding Spheroidal graphite
cast irons. CEN, 1997.
REFERENCES
5 Rotor 119
Guidelines for Design of Wind Turbines DNV/Ris
120 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
stresses as negative.
0,95
structural
Technological size effect 0,9
steel
Reduction due to the technological size
Alloy steels
effect is based on the fact that specimens of K1 0,85
identical dimensions, made of materials of
the same kind but of different dimensions, 0,8
have different fatigue properties. Test
specimens are made with relatively small 0,75
dimensions (typical diameter d = 5-10 mm)
and have had their mechanical properties 0,7
1 10 100 1000
improved as a result of the reduction of the d [mm]
cross-section by forging or rolling. The
technological size effect can be accounted Figure 6-2. Technological size effect (d denotes
for by means of the influence factor K1, diameter of considered part of shaft), from Roloff and
applied to the fatigue strength. K1 can be Matek (1994).
determined from Figure 6-2.
6 Nacelle 121
Guidelines for Design of Wind Turbines DNV/Ris
Stress concentrations
Figure 6-6. Shaft lay-out in vicinity of bearings, from
Stress concentrations will occur at local SKF (1989).
changes in geometry, e.g. where cross-
sections alter. This is also the case at shrink
122 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
Whenever a notch is encountered, the notch The geometrical stress concentration factor
sensitivity factor q can be used to convert is denoted and can be determined from
the stress concentration factor into a notch Figure 6-7 and Figure 6-8.
factor to be used instead
Stress concentration factor, torsion
= 1+q(-1)
2,1
A commonly encountered notch is found by h/r=4
the recess, which is to transfer the axial load 2
from a main bearing. A common lay-out of a h/r=3
shaft in the vicinity of bearings is shown in 1,9
Figure 6-6, and data for the involved
geometrical quantities are given in Table 1,8
h/r=2
6-1.
1,7
Table 6-1.
rs ba ha rc 1,6
Mm Mm mm mm
1 2 0.2 1.3 1,5
1.1 2.4 0.3 1.5 0,005 0,01 0,015 0,02
1.5 3.2 0.4 2 r/D
2 4 0.5 2.5
2.1 4 0.5 2.5
Figure 6-8. Stress concentration factor, from Roloff and
3 4.7 0.5 3
Matek (1994).
4 5.9 0.5 4
5 7.4 0.6 5
6 8.6 0.6 6 The following symbols are used in the
7.5 10 0.6 7 figures:
D Larger diameter of shaft
d Smaller diameter of shaft
Stress concentration factor, bending
h (D-d)/2
r radius of notch
3
h/r=5
2,9
Stress concentrations due to tight bearing
2,8 h/r=4 fits, corresponding to the ISO 286 shaft
2,7 tolerances from m to r, can be accounted for
2,6 h/r=3 by a stress concentration factor = 1.1-1.2.
2,5
6 Nacelle 123
Guidelines for Design of Wind Turbines DNV/Ris
necessary to reduce the maximum allowable use the following estimate, which refers to
stress range. However, for main shafts the the mean S-N curve with 50% failure
stress ratio often assumes values near to 1 probability
such that it may not be necessary to consider
any such reduction. In more accurate D50% = (0.436Re+77)
calculations, one may take into account the
influence of the mean stress by means of Re yield strength in MPa
methods available for this purpose, e.g. a
Haig diagram. Reference is made to D50% in units of MPa applies to polished
Gudehus and Zenner (1999), Bergmann and test specimens of small dimensions, 7-10
Thumser (1999) and VDI. mm, made of mild steel or low-alloy steel.
124 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
in which and are the actual characteristic For main shafts, both bending and torsion
normal stress and torsional stress, occur at the same time. One can take this
respectively, in the observed cross-section. into account by using the reference stress
D2.3% is the characteristic value for the
endurance limit, calculated from D50% as i ,ref = i + 3 i
2 2
specified above.
in which i is the normal stress, and i is the
For the notch factor , a somewhat better
torsional stress. As the bending stress is
approximation can be achieved by
oscillating with a stress ratio near 1, while
accounting for the smaller sensitivity of the
the shear stress is oscillating with a positive
material to notches at lower numbers of
stress ratio between 0 and approximately
cycles. At N = 10 one can set = 1 and then
0.8, use of the reference stress i,ref implies
let increase linearly with logN to =
an approximation. A more accurate
1+q(1) at N = 106 as shown in Figure 6-9. approach can be found in Gudehus and
Zenner (1999).
Notch factor
6.1.4 Ultimate strength
= 1 + q ( - 1 ) Stress calculation is to be carried out
according to standard mechanical
engineering methods and is usually straight-
forward. For materials with a high ultimate
strength, it may be necessary to account for
stress concentration in extreme load cases.
The influence of the ultimate strength can be
accounted for by the following factor
extreme = 1+(1)(Rm/1000)2
1
0 1 2 3 4 5 6 7 8
Rm ultimate strength in Mpa
Log10(N)
stress concentration factor taken from
Figure 6-9. Notch factor. Figure 6-7 and Figure 6-8
Fatigue loads are usually specified in terms The following criterion is to be fulfilled in
of a distribution of stress amplitudes, i.e. a design
so-called load spectrum, which on
discretised form gives the number of stress f < Re/(mextreme)
cycles in each interval of a suitably
discretised stress amplitude axis. The stress actual stress in the observed section
cycles in each stress amplitude interval Re yield strength
contribute to the total fatigue damage. The
total fatigue damage can be calculated by For stress ratios > 0 and dissimilar stress
Palmgren-Miners method and has to be less distribution, a plastic strain up to 0.2% may
than 1.0 in order to ensure a sufficiently low be allowed for single extreme load cycles.
failure probability in the design life of the Reference is made to Gudehus and Zenner
shaft. Reference is also made to Appendix (1999). The amount of plastic strain may be
C. estimated from Neubers correction, see
6 Nacelle 125
Guidelines for Design of Wind Turbines DNV/Ris
Gudehus and Zenner (1999) and Sundstrm can also be difficult. Usually, some kind of
(1998). For repetitive extreme load cycles shrink fit coupling is used, e.g. in
with stress ratios < 0, a low-cycle fatigue conjunction with a shrink disk.
calculation should be conducted, see
Gudehus and Zenner (1999) and Sundstrm Shrink disks
(1998). A correct functioning of clamp couplings
requires careful dimensioning and assembly.
Partial safety factors If the coupling, in addition to torque, also is
The requested safety level is normally to transfer bending moments, it is important
achieved by multiplying the characteristic to include this in the calculations. When
loads by a load safety factor f and reducing applying catalogue values for transferred
the characteristic material properties with a torque, it is recommended to apply a safety
material safety factor m. factor of 1.5 to the peak torque in order to be
on the safe side.
The load safety factor accounts for the
random nature of the load components (e.g. To achieve the prescribed level of safety, it
fatigue loads, extreme loads, gravitational is important to make sure that the roughness
loads, etc.) and the uncertainty in the conditions of the surface and the tolerances
calculation method. The material safety for both parts are in accordance with the
factor accounts for the scatter in the material recommendations of the manufacturer.
properties and for the level of quality control During assembly, correct bolt pretensions
of the material. shall be applied, and the required cleanness
of the frictional surfaces shall be accounted
The values of these safety factors are for.
normally given in the relevant wind turbine
regulations, e.g. DS472, and in the Couplings
associated material standards. For machine In certain cases where the gear is rigidly
components, the certifying bodies or the mounted on the machine foundation, a
owner sometimes define their own values coupling is applied which can only transfer
for m. torque and which is flexible with respect to
bending moments. In these cases, the hub
6.1.5 Main shaft-gear connection parts of the couplings are usually shrunk
both on main shaft and on gear shaft. Insofar
In cases where the main shaft does not
as regards the design of the coupling itself,
constitute an integral part of the gear and in
reference is made to special literature.
cases where it forms the generator shaft in
multi-polar generators, the main shaft should
6.1.6 Materials
somehow be connected to the transmission
input shaft. Formerly, hollow-axis gears In most cases, non- or low-alloyed
were used by which the torsional moment machinery steels are used, i.e. steel with a
was transferred from the main shaft to the carbon content of 0.3-0.7% and with less
shaft bushing of the gear by means of a than 5% alloy of metals such as Mn, Cr, Mo,
keyway connection. This principle for Ni and V.
torsional moment transfer has now been
abandoned, except for small simple wind This classification covers steel with rather
turbines, as keyway connections are only different mechanical properties, ranging
poorly suited to transfer varying and from non-alloyed steel with an ultimate
reversing loads. Assembly and dismantling strength of about 500 MPa and failure
126 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
6 Nacelle 127
Guidelines for Design of Wind Turbines DNV/Ris
Figure 6-11. Bearing frame with main shaft and bearing arrangement for wind turbine with two main bearings;
courtesy Vestas.
128 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
yg
ys
yr
MzR
Fs
FzR Fa
FyR Frh
Figure 6-12. Nacelle with transmission system for FxR Frv
which the main bearing is integrated in the gearbox.
MyR
Courtesy NEG Micon. MxR
6 Nacelle 129
Guidelines for Design of Wind Turbines DNV/Ris
130 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
range, the ability of protection against Lithium soap base grease has the quality of
corrosion and the load carrying ability. being waterproof and usable at a wide of
range temperatures 35 C to +130 C.
Water in the lubrication leads to corrosion,
degradation of the lubrication, formation of Other possibilities are sodium grease or
aggressive substances together with oil calcium soap base grease.
additives, and it affects the formation of a
load carrying lubricating film. Sodium grease absorbs large quantities of
water and is useful in environments with
Since the rotational speed varies from zero condensation. It may, however, soften to
to the nominal speed, a boundary lubrication such an extent that it flows out of the
condition will consist a considerable part of bearing.
the operational time. This will normally call
for a lubricant with EP (extreme pressure) Calcium soap base greases of penetration
additives and as high a viscosity as class 3 do not absorb any water. This is
practically possible. In this respect, it is advantageous in situations where bearing
essential to consider possible load cases seals are exposed to splash water.
where the radial load is low or even zero
combined with low temperatures (high The stiffness of the grease is determined by
viscosity), because it may cause sliding in the consistency class. A stiff grease, which
the bearing. belongs to consistency class 3 or higher, can
contribute to the sealing of the bearing and
Grease lubrication keep out contaminants by lying in a
The most commonly used lubrication in the labyrinth seal or in the contact area of a
main bearings is grease. rubbing seal.
Grease has the advantage of being easily However, for high P/C load ratios, greases
retained in the bearing arrangement, it of consistency class 1-2 should be selected.
contributes to sealing the bearing In a dusty environment, a stiff grease of
arrangement from contamination, and the penetration class 3 should be used.
use of an expensive circulation system is
avoided. Rolling bearing greases are The relubrication interval corresponds to the
standardised in DIN 51825. minimum grease life F10 of standard greases
in accordance with DIN 51 825. The grease
Grease consists of a base oil with thickeners service life is dependent on the type and
and possibly additives added. The following amount of grease, bearing type and size,
grease types are distinguished: loading, speed, temperature and mounting
mineral oil with metal soaps as conditions. For fairly large bearings (> 300
thickener. mm) the relubrication interval is
mineral oil with non-soap thickener recommended to be more frequent than F10.
synthetic oils with non-soap thickener In some cases, continuous lubrication is
established.
A possible choice of grease lubrication
would be a lithium soap base grease of Great care should be exercised if the grease
penetration class 2-3 with EP additives and type is to be changed. If incompatible
maybe corrosion and oxidation inhibitors. greases are mixed, their structure can change
6 Nacelle 131
Guidelines for Design of Wind Turbines DNV/Ris
drastically, and the greases may even soften Before going into operation, oil must be
considerably. supplied to the bearing. In case of
circulating oil lubrication, the oil pump
As regards the amount of grease to be used, should be started before the turbine goes
a rule of the thumb is to fill the bearing into operation. In other cases, the bearing
completely with lubrication while the must be manually lubricated before first
housing is half filled (SKF, 1989). start-up, and total drain of the bearing during
service must be avoided.
Oil lubrication
Since temperatures are normally relatively 6.2.5 Rating life calculations
low, the lubricant does not need to function
The fatigue load carrying capacity of the
as heat dissipator. In that case, there is no
bearing is characterised by the basic
need for circulation of the lubricant, which
dynamic load rating C. This quantity can be
simplifies the design. On the other hand,
calculated according to ISO 281 (1990).
when the lubrication is not circulated, there
is no possibility of filtering, and
The static load carrying capacity of the
relubrication is therefore necessary.
bearing is characterised by the static load
rating C0. This quantity can be calculated
Oil lubrication makes it necessary to
according to ISO 76 (1990).
monitor the lubrication system because of
the risk of leakage. A disadvantage
The general methodology for selecting and
compared to grease lubrication is that it
calculating rolling bearings is given in
demands a better sealing and a circulation
Section 6.3 for both dynamic and static
system.
loads. In general, for a design life of 20
years, the required basic rating life L10h
Straight oils and preferably corrosion- and
should equal or exceed 300,000 hours. The
deterioration-inhibited oils can be used. If
modified rating life L10mh should at the same
the recommended viscosity values are not
time reach 175,000 hours. In the context of
maintained, oils with suitable EP additives
grease lubrication and oil lubrication
and anti-wear additives should be selected.
without filter, the contamination factor C
If the bearings are heavily loaded (i.e. the
should not be chosen higher than 0.2 if
load ratio P/C > 0.1), or if the operating
special precautions are not taken to obtain
viscosity is smaller than the rated viscosity
and verify higher values. For oil lubrication
1, oils with anti-wear additives should be with off-line filtering, a contamination factor
used. EP additives reduce the harmful C of 0.5-0.7 may be obtained. The C/P ratio
effects of metal-to-metal contact, which should not exceed 2.5 for the largest load
occurs in some places. The suitability of EP interval.
additives varies and usually depends largely
on the temperature. Their effectiveness can The above-mentioned figures apply to
only be evaluated by means of tests in medium size wind turbines (600-1000 kW).
rolling bearings. (FAG WL-81 115/4.) It should be borne in mind that large
bearings are relatively less sensitive to
The intervals between oil changes depend on contamination particles of a certain size than
the specific lubrication system and smaller bearings.
circulation, contamination and ageing of the
oil.
132 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
The static safety factor C0/P0 based on the 6.2.7 Bearing housing
extreme load cases should not be lower than
The bearing must be firmly supported by the
4.
whole circumference to achieve a proper
load transmission.
6.2.6 Connection to main shaft
The thrust from the rotor should be For bearings with normal tolerances, the
considered to be taken up by shoulder or by dimensional accuracy of the cylindrical
friction, or a combination of both. One seating in the housing should be at least IT
should then be aware of the stress concen- grade 7, and on the shaft at least IT grade 6.
tration in the main shaft at the shoulder.
The tolerance for cylindrical form should be
In cases where there is a possibility of the at least one IT grade better than the
wind turbine being in a situation where the dimensional tolerance.
wind is coming from the back, one should
consider a stop ring as shown in Figure Since cast bearing houses will often have a
6-15. complex geometry, an obvious method to be
used for verification would be a finite
back wind element analysis, see Appendix D.
secure ring
6.2.8 Connection to machine frame
The connection between the bearing housing
and the machine frame will most likely be a
bolt connection. The connection should be
capable of transferring the combination of
Figure 6-15. Back wind stop ring. axial and radial forces from the bearing to
the main frame by friction or by shear in
A tight fit, i.e. complete support of the bolts by tight fit, depending on the geometry
bearing ring over its entire circumference, of the connection (see Appendix A).
makes a full utilisation of the bearing's load
carrying capacity possible. 6.2.9 Standards
The bearing clearance should be as small as ISO 76 Roller bearings Static load ratings.
possible to ensure accurate guidance,
however, large enough in order not to get ISO 281 Roller bearings Dynamic load
stuck in any situation. When considering the ratings and rating life + Amendment 1 and
bearing clearance things to take into account 2.
will be a possible difference in temperature
and expansion of inner ring and compression IEC 61400-1 Wind generator systems, Part
of outer ring during mounting, respectively. 1. Safety requirements.
6 Nacelle 133
Guidelines for Design of Wind Turbines DNV/Ris
Other types of gears, rarely or never used in Figure 6-17. Planetary gear principle, outer fixed
wind turbine main gears, include the annulus with three revolving planets and one rotating
following, which are listed here merely for planet carrier in the middle, from Niemann and Winter
(1985).
completeness:
bevel gears consist of a pair of toothed
Tooth form
conical wheels whose working surfaces
The tooth forms used practically universally
are inclined to nonparallel intersecting
in spur, helical, bevel and worm gears are
axes. Bevel gear wheels can be
so-called involute teeth. This tooth form
designed with spur teeth, helical teeth,
implies that rotation of the base circle at a
and curved spiral teeth. Spiral bevel
uniform rate is associated with uniform
gears result when the teeth are helical or
displacement. The path of contact is a
curved. See Figure 6-18.
straight line, which coincides with the line
worm gears consist of a worm-thread of action.
wheel, or endless screw, working in
conjunction with a cylindrical toothed
134 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
6 Nacelle 135
Guidelines for Design of Wind Turbines DNV/Ris
Pi power in the ith defined load interval Table 6-2. Reliability factor a1.
hi number of hours of operation in the ith Failure Reliability Reliability
interval probability 1n factor a1
ht total number of hours of operation n (%) (%)
10 90 1.00
Rolling bearings 5 95 0.62
The design of roller bearings for a gear is 4 96 0.53
made using a load duration spectrum. The 3 97 0.44
load duration spectrum may be transformed 2 98 0.33
to an equivalent cubic mean load, Peq. 1 99 0.21
According to ISO281/1, the basic rating life
is defined as Bearing manufacturers recommend values
for the factor product a23 = a2 a3. These
p
C dyn values may vary from manufacturer to
L10 = 10 6 revolutions manufacturer, and a23 will usually depend
P
eq on the viscosity ratio
136 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
6 Nacelle 137
Guidelines for Design of Wind Turbines DNV/Ris
The nominal tangential load is given as load, and accounts for the
maldistribution of load in multiple-
2T path transmissions
FT = KV Internal dynamic factor to account for
d
internally generated dynamic loads in
T applied torque the gear
d reference diameter KH face load factor of contact stress and
scuffing
Surface durability KH transverse load distribution factor of
Contact stress. The design contact stress H contact stress and scuffing
is derived from the nominal tangential load
as follows The factor Z is defined as the product
Z = ZBDZHZEZZ
FT (u + 1)
H = Z K
d 1bu ZBD zone factor for inner point of single
pair contact for pinion or wheel
u gear ratio per stage ZH zone factor for pitch point
b face width ZE elasticity factor that accounts for
d1 reference diameter of the pinion influence of modulus of elasticity and
Youngs modulus. This is a factor
The factors K and Z are compound influence whose squared value is in units of
factors to account for various effects. The stresses
factor K is defined as the product Z contact ratio factor that accounts for
influence of transverse contact ratio
K = KAKKVKHKH and overlap ratio
Z helix angle factor
KA application factor, defined as the ratio
between maximum repetitive torque Note that all factors, except ZE, are
and nominal torque, and accounts for dimensionless. Formulas and details for
dynamic overloads external to the calculation of the various factors can be
gearing found in ISO6336, DIN3990, and DNV
K load-sharing factor, defined as the ratio CN41.2
between the maximum load through
the actual path and the evenly shared
138 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
6 Nacelle 139
Guidelines for Design of Wind Turbines DNV/Ris
Two criteria are to be fulfilled to ensure a Formulas for calculation of S, MB, MB,
sufficient level of safety against scuffing flamax, and flaint can be found in DNV
failure. Both criteria are formulated in terms CN41.2.
of criteria on temperature, i.e. the local
contact temperature may not exceed some Note that gray staining may occur under the
permissible temperature. The one criterion is same conditions that may lead to scuffing,
a so-called flash temperature criterion, based and that this may even happen without or
on contact temperatures, which vary along before the occurrence of a literal scuffing
the path of contact. The other criterion is an failure.
integral temperature criterion, based on the
140 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
6 Nacelle 141
Guidelines for Design of Wind Turbines DNV/Ris
142 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
strength values for both pitting and bending N41.2 is based partly on ISO6336. The
fatigue are dependent on the quality level. material and testing requirements that the
Values are given in DIN3990 T5 and Classification Note 41.2 is based on are
ISO6336-5. The DNV Classification Note given below.
6 Nacelle 143
Guidelines for Design of Wind Turbines DNV/Ris
potential controls and continuously It is required that the depths to 550 HV, 400
recorded HV and 300 HV and the core hardness are
documented to be within the approved
The entire case-hardening process is specification. Further, the hardness at any
checked at regular intervals with regard to: point below the surface is not to exceed the
The case microstructure: to be surface hardness (before grinding) with
martensite with allowance for up to more than 30 HV. The core impact energy is
15% retained austenite and fine not to be less than the approved
dispersed individual carbides. (Higher specification, and in no case less than 30 J.
percentage of retained austenite may be
accepted provided increased safety As mentioned above, the case-hardening
factor against scuffing). process (the coupons) is normally to be
Decarburisation: not to be visible at a documented for each hardening batch and
magnification of 500. each material charge. The use of coupons
The core microstructure: to be marten- made of the same material type but not same
sitic/bainitic with no free ferrite in charge may be accepted provided that the
critical tooth root area. manufacturer has a quality assurance system
which ensures sufficient reproducibility. In
The case-hardening process is normally to particular, the limits of elements in the
be documented for each hardening batch and chemical composition combined with the
each material charge with a certificate, respective heat treatment processes must
which includes the following: ensure that the required core properties are
hardness profile obtained. This part of the quality assurance
core impact energy (KV) system has to undergo special evaluation. A
If no alternative procedure is approved, the reduced extent of impact testing may also be
certificate is to be based on a coupon test. considered.
The coupon is normally to be made of ma-
terial from the same charge as the actual Case-hardened gears are to have a minimum
gear and heat-treated along with this charge. tooth root space hardness of 58 HRC over
The coupon is to be sampled in the the entire face width. Otherwise, a reduction
tangential direction and is not to be of permissible tooth root stresses applies,
separately forged. If it is not possible to see Classification Note 41.2 Part 3, Section
sample coupons tangentially, longitudinal or 7.
radial samples may be accepted.
Depending on the specific material type, this
The coupon diameter is not to be less than 2 may be difficult to obtain for large gears and
times the normal modulus, minimum 20 control testing may be required. Therefore,
mm. Further, the size is to be sufficient for manufacturers may carry out special
making 2 test pieces for impact energy (KV) procedure tests in order to document the
of the core. permissible sizes for their various material
types. Components of smaller sizes than
The hardness profile (hardness as a function those tested need no documentation of tooth
of depth) is to be determined by hardness root hardness.
measurements with a load of 10-50 N. The
measurements are to be made from the sur- If a component exceeds the tested size, or if
face to the core. The expected amount of the manufacturer has not carried out a
grinding is to be subtracted. procedure test, the tooth root space hardness
144 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
As mentioned above, the nitriding process Coupling bolts are to be prestressed so that a
(the coupons) is normally to be documented suitable amount of prestress remains even
for each hardening batch and each material under the most severe running conditions, in
charge. The use of coupons made of the particular, with regard to bending moments.
same material type but not from the same The level of safety is to be demonstrated in
charge may be accepted provided that the both the ultimate limit state and the fatigue
manufacturer has a quality assurance limit state. The same minimum safety
system, which ensures sufficient reproduci- factors as those for shaft design apply, see
bility. Section 6.1.
6 Nacelle 145
Guidelines for Design of Wind Turbines DNV/Ris
If the torque transmission is based only on The yield strength to be applied in checks
friction between the mating surfaces of according to these two criteria is not to
flange couplings, the friction torque exceed 2/3 of the tensile strength of the key,
(including the influence of axial forces and and it is not to exceed twice the yield strength
bending moments) is not to be less than 1.5 of the shaft or the hub, whichever is involved.
times the characteristic peak torque.
In principle, there is to be no clearance
The torque transmission may also be based between the hub and the shaft, however, a
on a combination of shear bolts and friction certain amount of minimum interference fit is
between the mating flange surfaces. A basic required, e.g. approximately 0.02% of the
principle is that both the friction alone shaft diameter.
(including the influence of axial forces and
bending moments) and the shear bolts alone 6.4.4 Torsionally elastic couplings.
should be able to transmit the characteristic
peak torque. Rubber couplings are to be designed such
that a failure of a rubber element does not
6.4.2 Shrink fit couplings cause loss of the connection between the
rotor and the brake.
The friction connection is to be able to
transmit at least 1.5 times the characteristic 6.4.5 Tooth couplings
peak torque without slipping. Bending
moment influence is to be considered. Tooth couplings are to have a reasonable
degree of safety with respect to surface
For tapered mating surfaces where a durability and tooth strength. This is subject
slippage due to torque and/or axial force to special consideration.
may cause a relative axial movement
between the tapered members, the axial
movement is to be prevented by a nut or 6.5 Mechanical brake
similar. When a nut is required, the pre- Mechanical brakes are usually used as a
stress is to be of the same magnitude as the backup system for the aerodynamic braking
axial force component from the tape. system of the wind turbine and/or as a
parking brake, once the turbine is stopped,
The permissible material stress depends on e.g. for service purposes. Mechanical brakes
the relative wall thickness, material type, are sometimes also used as part of the yaw
and whether the coupling is demountable or system. In a mechanical brake, brake
not, and the usual range of permissible callipers, brake discs and brake pads form
equivalent stress (von Mises) is 70% to crucial parts. A hydraulic system is usually
110% of the yield strength of the hub. used for the actuation and release of the
brake.
6.4.3 Key connections
The connection is to be able to transmit the 6.5.1 Types of brakes
characteristic peak torque. Mechanical brakes can be active or passive,
depending on how the hydraulic system of
The shear stress in the key is not to exceed the brake is applied:
50% of the yield strength in shear. The active brake: the pressure of the
pressure on the side of the keyway is not to hydraulic system actively pushes the
exceed 85% of the yield strength of the key. brake pads against the brake disc.
146 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
6 Nacelle 147
Guidelines for Design of Wind Turbines DNV/Ris
passive brake: the pressure of the In general, the thicker the disc is, the better
hydraulic system keeps a spring tight. is its ability to absorb temperature loading.
Once the pressure is released, the spring
is also released and will push the brake The possible variation in the frictional
pads against the brake disc. coefficient poses a problem, which must be
In either case, the hydraulic pressure of the given due consideration when the brake
hydraulic system is crucial in order to be system is being dimensioned. If the
able to operate the brake as intended. The frictional coefficient is too big, the brake
hydraulic pressure is usually provided by force will become too large. If the frictional
means of an accumulator. For active coefficient is too small, the brake system
systems, it is particularly important to make will be unable to brake.
sure that the pressure in the accumulator is
always available, and it is important to have 6.5.3 Brake torque sequence
redundancy in this respect, i.e. an extra
The rise time from zero to maximum brake
pressure source is necessary for backup.
torque will influence the dynamic response
of the turbine heavily. Since the turbine is
The type of spring used in a mechanical
rotating when the brake is actuated, full
brake to keep up a pressure is often a coil
brake force will be mobilised immediately
spring of the disc spring type. This type of
upon the actuation of the brake.
spring is nonlinear and has the advantage
that it is capable of maintaining an
approximately constant spring pressure over
a considerable range of deflection.
148 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
the torque raises and develops as shown in hydraulic fluid is not to have a flash point
the figure. lower than 150C and is to be suitable for
operation at all temperatures that the system
may normally be exposed to. Means for
6.6 Hydraulic systems filtration and cooling of the fluid are to be
incorporated in the system wherever
In a hydraulic system, power is transmitted
necessary.
and controlled through a liquid under
pressure within an enclosed circuit.
6.6.2 Accumulators
Hydraulic systems are used in wind turbines,
for example in terms of a hydraulic For gas and hydraulic fluid type
accumulator for pitching of the rotor blades. accumulators, the two media are to be
A hydraulic system must be protected suitably separated if their mixture would be
against exceeding the maximum admissible dangerous or would result in the
pressure. A pressure release valve can be contamination of the hydraulic fluid and/or
used for this purpose and will prevent loss of gas through absorption.
explosion in the event of fire. All
components of the hydraulic system must be Each accumulator is to be protected on both
easily accessible for assembly, adjustment its gas side and its hydraulic fluid side by a
and maintenance. safety device such as a relief valve, a fuse
plug or a rupture disc to prevent excessive
Pressure shocks should be kept to a pressure if overheated. When the
minimum. Pressure shocks or a large accumulator forms an integral part of a
pressure drop must not lead to a dangerous system with such a safety device, the
condition. A safe condition must be accumulator itself need not be supplied with
guaranteed in the event of power supply such a safety device.
failure and in the subsequent event of
restoration of the power supply. 6.6.3 Valves
The following external factors must not Valves are used in hydraulic systems to
affect the operation of a hydraulic system: control the hydraulic effect between a pump
and an engine, cylinder, or actuator. The
salt and other corrosive substances
purpose of valves is to govern the direction
sand and dust
and amount of the volume flow rate or to
moisture block the volume flow rate, and it is also to
external magnetic, electromagnetic and limit or control the pressure of the fluid. A
electric fields distinction can be made between four major
sunlight types of valves, viz.
vibrations shut-off valves, which block flow in one
direction and allow for partly or full
When a hydraulic system forms part of the flow in the opposite direction
protection system, grid failures and extreme directional control valves, which control
temperatures must not compromise the the direction of the volume flow rate,
operation of the system. and which can block the volume flow
rate or adjust the amount of volume
6.6.1 Arrangement flow rate
Hydraulic systems should have no pressure valves, which limit or control
connections with other piping systems. The the hydraulic pressure
6 Nacelle 149
Guidelines for Design of Wind Turbines DNV/Ris
flow control valves, which adjust the the return pipes must be made
volume flow rate to the hydraulic sufficiently strong by their design
actuator such that the desired speed of and construction, or by protection,
the hydraulic actuator is achieved. such that they cannot be closed off
by external damage.
6.6.4 Application in protection filters must be installed on the
systems pressure side of the pump and
should be avoided in the return
If a hydraulic system forms part of a
pipe. If filters are installed in the
protection system, the design and
return pipe, a bypass is required.
construction of the hydraulic system must
comply with the requirements to the no components must be installed in
protection system, and it must be designed, the return pipe which can lead to
constructed and used as a fail-safe or blockage of the pipe as the result of
redundant system. Hydraulic systems, which maloperation.
form parts of protection systems, can be 2. For actively released brakes, which are
divided into three categories as follows: being actuated hydraulically or
pneumatically, the actuation must be
1. Systems in which the brake is actively accomplished by means of a pressure
released by a hydraulic or pneumatic accumulator, and the following
pressure medium. additional requirements apply:
2. Systems in which the brake is actively the connecting pipe between the
released (mechanically, hydraulically, accumulator and the actuated
pneumatically or electrically), but component, e.g. the blade adjust-
actuated hydraulically or pneumatically. ment mechanism, must be as short
The active release thus takes place as possible.
against a passive oil or air pressure. no other components, such as
3. Systems in which the brake is released valves, couplings and rotating
in the neutral state (passively released) elements, are allowed in this
and is actuated hydraulically or connection.
pneumatically. the pressure in the accumulator
must be monitored at a level which
It is recommended to comply with the is sufficiently high to guarantee
following requirements to these three types independent braking action.
of hydraulic systems: the other protection system of the
1. When brakes (mechanically or wind turbine must be actuated
aerodynamically) are actively released, mechanically, and the actuating
the pressure medium shall be able to element shall be designed and
flow away in a reliable manner during a constructed safe-life, e.g. a
braking action. In the hydraulic system, mechanical spring.
the following features shall be present 3. The use of a passively released braking
as a minimum: system, by which the braking action
two valves must be placed parallel takes place by means of build-up of
to one another as a switching pressure, is only permissible when the
element. An incorrect switch following conditions are met:
position of a valve must lead to a the monitoring, control and
safe situation. It must be possible to actuation systems are designed and
test each valve individually. constructed with redundancy, i.e.
150 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
In addition, for Categories 2 and 3, an This section deals with the safety aspects of
actively operated hydraulic or pneumatic the generator and its interaction with the
installation shall not be used to keep a wind rotor and the transmission system. The
turbine in a safe state for a long period after functioning of the generator and its
a protection system has been actuated. interaction with the grid is dealt with in
Chapter 9. An example of a generator is
6.6.5 Additional provisions shown in Figure 6-24.
It is recommended to take the necessary
steps to ensure that failure of a redundant
system can be detected. Long-term standby
redundancy should be avoided. In the case
of oil leaks in hydraulic systems, other wind
turbine components or other systems must
not be affected.
ISO, Hydraulic fluid power General rules The rotational speed of the generator is
for the application of equipment to dependant on the grid frequency and the
transmission and control systems, number of poles
International Standard, ISO4413, 1st edition,
1979. f
ns = 60
p
ISO, Hydraulic fluid power Gas-loaded
accumulators with separators Range of f grid frequency in Hz
pressures and volumes, characteristic p number of pole pairs
quantities and identification, International ns synchronous rotational speed in rpm.
Standard, ISO5596, 1st edition, 1982.
6 Nacelle 151
Guidelines for Design of Wind Turbines DNV/Ris
The produced alternating current, which is insignificant. When a larger slip is allowed,
transmitted to the electrical grid, must match say within 10%, and can be varied
the frequency of the grid. The required electronically, e.g. by a rotor current
rotational speed of the generators rotor is controller, it is referred to as variable slip.
achieved by means of the gearbox of the The pitch or stall control of the wind turbine
wind turbine, since the wind turbine rotor is meant to ensure that the allowable slip of
itself is not allowed to rotate at this high the generator is not exceeded.
speed for physical reasons. Rotation of the
wind turbine rotor at the high rotational The advantage of the variable slip comes
speed of the generator rotor would cause about when the wind turbine is operated at
aerodynamic problems and supersonic its rated power. At the rated power, power
speeds. Noise would also be a problem, and fluctuations caused by changes in the wind
excessive centrifugal forces would be speed are undesirable. When a wind gust
generated. A multi-pole generator has the hits the wind turbine rotor, the slip enables
advantage of reducing the mechanical the generator speed to increase a little in
complexity of the generator and makes it response to the gust without causing a
possible to reduce the gearbox and corresponding increase in the generated
sometimes even to omit it. power output. Thus, the slip ensures a
smooth power output and at the same time
There are two major types of generators: contributes to keeping the loads on blades,
synchronous generators main shaft and gearbox down.
asynchronous generators
The variation of the operating speed with the
A synchronous generator operates at a applied torque for an asynchronous
constant speed, dictated by the frequency of generator is beneficial because it implies a
the connected grid, regardless of the smaller peak torque and less wear and tear
magnitude of the applied torque. The speed on the gearbox than for a synchronous
dictated by the frequency of the grid is also generator. This is one of the most important
known as the synchronous speed. reasons for using an asynchronous generator
rather than a synchronous generator in a
An asynchronous generator is a generator, wind turbine, which is connected directly to
which allows slip, i.e. deviations from the the electrical grid. Another reason for using
rotational speed dictated by the frequency of asynchronous generators is that the involved
the connected grid. In other words, the slip is beneficial when there is flexibility in
rotational speed is allowed to vary the structural system.
somewhat with the applied torque. This is
the most common generator type used in Traditionally, the active materials in a
wind turbines. A variant with coiled rotors is generator consist of magnetically conducting
prevalent. The slip is defined as the iron and electrically conducting thread
difference between the rotational speed of arranged in a coil. Permanent magnets are
the generator and the rotational speed becoming increasingly common, and
dictated by the frequency of the grid. The electrical components, such as temperature
slip is sometimes expressed in percent of the sensors, are becoming integral parts of the
latter. When a slip of up to about 1% is generator.
possible, the operation mode for the
asynchronous generator is referred to as With a view to switches and power failure,
constant speed, given that this slip is rather the generator must be able to withstand
152 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
switching on in phase opposition at 100% humidity in the external ambient air of 95%
residual voltage, or it must be secured within the entire temperature range as
against the occurrence of this situation by specified above.
means of a special arrangement in the
control system. The influence of the humidity of air on the
electrical components is always dependent
Generators are to be constructed in such a on other climatic parameters, in particular,
way that when running at any working on the temperature and on changes in the
speed, all revolving parts are well-balanced. temperature. Creation of condensation can
Suitable fixed terminal connectors are to be be remedied by means of heating when the
provided in an accessible position with generator is not running, and by heating of
sufficient space for convenient connection closets when the self-heating is insufficient
of the external cables. to avoid damaging condensation of water.
The electrical components of the turbine The generator forms one of several links in
shall be fully functional at a relative the transmission between the rotating system
6 Nacelle 153
Guidelines for Design of Wind Turbines DNV/Ris
and the electrical system of a wind turbine, interaction between these features and the
i.e. between the blades and the grid. Failure generator which ensures that the turbine is
of any link in this chain implies a risk. This kept in a safe condition. If the generator is
risk is absorbed by the protection system, disconnected due to an error or due to
which brings the wind turbine to a safe intervention from the protection system, or
condition in which it remains until normal if an error occurs in the blade pitch system
operation can be resumed. The number of or the speed control, at least one of the two
failures that demand activation of the fail-safe brake systems, which form part of
protection system should be minimised in the protection system, shall begin working.
order to reduce the burden on the protection These are usually pitchable blades and
system. The probability of breakdown mechanical brakes. Note that for turbines
results from the probability of failure of the with pitchable blades, one of the two brake
protection system combined with the systems the pitchable blades forms part
probability of a critical error that requires of both the protection system and the control
the intervention of the protection system. system. Errors which have common root
The probability of breakdown shall be less causes need to be given special attention.
than 0.0002 per machine year. The target for Neither blade pitch nor mechanical brake
the reliability of the protection system is a will have the required effect if the blades are
number, which is large enough to keep the locked at a position of, for example, +15.
probability of breakdown below this level.
For such concepts, the target reliability
The grid-connected asynchronous generator quoted above still applies. Components
with short-circuited cage winding in the which form part of both control and
rotor has been the basis for the safety protection functions are to be fail-safe
considerations behind the Danish Approval designed, or their probability of failure is to
Scheme and for the associated recommen- be minimised.
dations.
The generator is to be designed such that it
For a stall-controlled wind turbine with can produce a sufficiently large torque to
asynchronous generator, it is the generator keep the turbine within its defined range of
and its relatively simple control that most of operation, see Section 2.2.1. Disconnection
the time constitute the system that keeps the (switching off) of the generator should be
turbine in a safe condition. If the generator based on reverse power, on zero-power, or
is disconnected due to an error or due to on a signal from the relay protection against
intervention by the protection system, at electrical errors. The purpose is to utilise the
least one of the two fail-safe brake systems, braking power of the generator in all
which are part of the protection system, shall situations.
begin working. These are usually pitchable
blade tips and mechanical brakes. Fail- When a frequency converter is used, it needs
safe is defined as a design philosophy by to be considered together with the generator
which the safety of the turbine is maintained with respect to safety and frequency of
even during component failure or grid errors.
failure.
The generator is to be designed for the
Insofar as regards a turbine with variable mechanical impacts it will be exposed to. At
speed and pitchable blades, and thus a start-up of a turbine with fixed unpitchable
relatively complex control, it is the blades in high wind speeds, the cut-in of the
154 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
6 Nacelle 155
Guidelines for Design of Wind Turbines DNV/Ris
156 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
6 Nacelle 157
Guidelines for Design of Wind Turbines DNV/Ris
158 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
terms of cable twisting if the turbine keeps positioning of the turbine relative to the
yawing in the same direction for a long time. wind is also referred to as forced yaw. Most
An active yaw system employs a mechanism large horizontal axis wind turbines use
of hydraulic or electrically driven motors forced yaw to align the rotor axis with the
and gearboxes to yaw the turbine and keep it wind. An example of an active yaw system
turned against the wind. Such active is given in Figure 6-25.
Yaw drive
Mashin frame (nacelle)
Slewing
bearing Yaw gear wheel
Yaw brake
Tower top
The mechanism used for an active yaw system, including the yaw ring, the yaw
system usually consists of a number of drive with the yaw motors, the yaw bearing,
electrically operated motors in conjunction and the yaw brakes.
with a gear that actuates a large toothed yaw
ring in the tower circumference. Together The yaw error is usually measured by means
with yaw brakes and yaw bearings, these of direction sensors such as one or more
components are most often delivered as wind vanes. The wind vanes are usually
standard components from a supplier, who placed on top of the nacelle. Whenever the
also provides the pertaining design wind turbine is operating, an electronic
documentation. When used in a yaw system, controller checks the orientation of the wind
it should be noted that these components vanes and activates the yaw mechanism
may be exposed to conditions which have accordingly. In addition to this automated
not been taken into account by the supplier. yaw of the wind turbine, it should be
The following sections deal with the possible to yaw the nacelle manually.
different components of an active yaw Manual yaw is needed during start-up,
6 Nacelle 159
Guidelines for Design of Wind Turbines DNV/Ris
service of the yaw system, and testing of the place. When the duration of yaw is not
turbine. known for design, DS472 specifies that yaw
can be assumed to take place during 10% of
6.10.1 Determination of design loads the time for all wind speeds that occur.
Assuming yaw in 10% of the time is,
Yaw is characterised by the maximum
however, quite conservative for most sites.
angular yaw velocity k and the fraction of
the design life during which yaw takes
M zR
FN
yN
F zR
M yR F yR yR
F xR zN
zR Yaw bearing
M xR Rotor centre
Figure 6-26. Loads acting on the rotor and the yaw bearing.
160 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
Figure 6-27. Relation between wind speed, yaw error and yaw moment.
Figure 6-27 shows a computer simulation of for gyro effects to become negligible.
the yaw moment as a function of the yaw Reference is made to Section 4.2.1.
error and wind speed for a three-bladed
wind turbine. It appears that the yaw error A passive yaw system will normally require
oscillates with frequencies equal to the rotor some sort of damping arrangement to reduce
frequency and three times the rotor the yaw speed.
frequency. It also appears that the sign of the
yaw moment changes when the yaw For large turbines with active yaw systems,
direction changes, and not surprisingly the yaw speed is usually lower than 1/sec,
the yaw moment increases when the yaw which is small enough for the gyro effects to
error and the wind speed increase. be ignored. To achieve such a low yaw
speed, the yaw motors needs to be connected
6.10.2 Yaw drive through a gearbox.
The yaw drive is the system of components
Such drives can be delivered as standard
used to cause the yaw motion. A large yaw
equipment from manufacturers of
speed will produce gyro effects that will
electrically operated motors.
result in large loads on the wind turbine. The
yaw speed must therefore be small enough
The yaw drives must have sufficient power
to overcome the largest mean yaw moment
6 Nacelle 161
Guidelines for Design of Wind Turbines DNV/Ris
occurring plus friction in the yaw bearing. yaw moment might be transmitted through
Installation, lubrication and service should the yaw drive.
be undertaken in accordance with the
specification from the manufacturer. 6.10.4 Yaw brake
During power production, turbulence, wind
Pinion shaft, pinion and mounting must be
shear and fairly small inevitable yaw errors
designed to withstand the maximum yaw
will give rise to a torque moment about the
moment, including partial safety factors.
tower axis. To keep the nacelle in position
and to spare the gears, it is common to
6.10.3 Yaw ring
mount a brake disc in connection with the
The yaw drive is normally mounted on the yaw bearing. Braking can then be performed
nacelle in such a way that it is in gear with a by means of hydraulic activated callipers.
toothed yaw ring mounted on the top of the Passive hydraulic callipers, i.e. spring-
tower. Yaw rings are normally toothed on applied brake callipers, are preferable
the inner surface as this provides a better because they can ensure braking also in the
protection from the surroundings and case of a leakage in the hydraulic system.
implies a slightly better mesh.
The brake calliper manufacturer will provide
The gears connected to the drive and the surface tolerances and geometrical tolerances
yaw ring must be designed in such a manner for the brake disc. These tolerances must be
that tooth failure for the maximum peak yaw complied with.
moment is prevented.
The yaw brake callipers, discs and mounting
Bending stress in a tooth can be calculated bolts shall be designed to withstand the
according to DIN 3990, Part 3 from maximum occurring yaw moments. For
turbines with yaw brake systems designed
Fl
F with limited safety (S = 1.15), it shall be
bm Y f Y SF
F =
proven that the turbine will withstand free
yaw operation, i.e. inertia forces due to
Yf tooth shape factor acceleration of the nacelle.
Y load reduction factor
F tooth force The brake is to be protected against dust,
b tooth width corrosion, oil and any other influence that
m module might alter the friction. The brake and
Fl tooth yield stress control system shall be designed such that a
SF safety factor situation where pads are worn out is avoided.
Scuffing might occur in gears with hardened In addition to using the ordinary yaw brake,
pinions whose tip edges act as scrapers. it must be possible to block the yaw
Damage due to wear and fatigue should be mechanism to enable service and
considered. maintenance adjacent to the yaw mechanism
without any personal risk.
It should be noted that the yaw ring is not
only loaded during yawing. Even though Blocking of the yaw system can be done by
yaw brakes might be applied a part of the mechanical fixation of the yaw ring or the
yaw motor shaft. Blocking of the yaw
system must not be dependent on external
162 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
power supply. Only in the case of negative Tolerances for the mounting surfaces should
callipers, the motor is allowed to be be in accordance with specifications from
electrically braked. A fixation of the ring is the bearing manufacturer. To prevent the
preferable because a possible backlash in the bearing from becoming distorted, the contact
yaw gear can be large enough to cause the surfaces must be carefully machined and
nacelle to jerk during service. attention must be drawn to the stiffness of
the surrounding structure. Application of a
The locking mechanism must be designed plastic grouting may compensate for
for wind speeds up to the defined stop wind irregularities in the contact surfaces.
speed, above which maintenance, which
demands locking of the yaw mechanism, is Slewing bearings used in yaw systems are
not permissible. different from normal bearings in that they
are exposed to oscillatory motion. This
6.10.5 Yaw bearing oscillatory motion and the low rotational
speed imply that slewing bearings have a
The yaw bearing is the bearing that supports
tendency to exhibit a relatively low ratio
the nacelle in a horizontal axis wind turbine.
between the lubrication film thickness and
It is located between the rotating nacelle and
the surface roughness.
the stationary tower and transmits wind
loads from the nacelle to the tower. As
Calculation of the load rating for bearings in
regards the yaw bearing, it has been
oscillatory motion is described in NWTC. It
common in the past to choose between two
is quite complex and involves the variation
different solutions slide plates or rolling
of load and the bearing rotational position as
bearings.
a function of time. In practice, it is common
that the bearing manufacturer provides the
Rolling bearings will often be designed as
design calculation or means for how to
slewing bearings, which are capable of
verify the design.
accommodating combinations of axial,
radial and moment loads. Yaw motion is
Unlike regular rolling bearings, whose
generated by gears mounted in the nacelle
strengths are represented by the load rating
and being in gear with the toothed bearing.
C, the strength of a slewing bearing will
Slewing bearings, as seen in Figure 6-28, are
normally be represented in terms of a curve
mounted by means of bolting to the seating
that gives the relation between the allowable
surfaces. Usually, bolts of quality 10.9 are
equivalent tilt moment and the equivalent
recommended, see Appendix A.
axial load. Reference is made to Figure 6-29
for an example.
6 Nacelle 163
Guidelines for Design of Wind Turbines DNV/Ris
will be provided by the manufacturer. KA is As for any other component in the wind
an application factor and is recommended to turbine, the dynamic loading of the bearing
be in the range 1.7-2.0 for yaw bearings. can be represented by one or more load
The safety factor KS equals the partial safety spectra, or on a simpler form by a
factor for the load of the relevant load case. number of load cases that are assumed to
adequately represent the loading that the
The tilt moment as calculated in Section bearing will experience over the design life
6.10.1 is denoted Mt, and the equivalent tilt of the turbine. In the latter case, at least four
moment Meq is calculated as load cases should be modelled. The
equivalent load for a specified load spectrum
Meq = Mt KA KS or a set of representative load cases can be
calculated according to the formula
in which KA and KS are to be taken as for the
axial load.
P O
i
p
i
P= i
,
O
p
Feq i
164 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
Myaw = (aMt + bFaDr + cFrDr), number of plates and claws that engage the
nacelle to the tower, see Figure 6-30.
Myaw friction moment
friction coefficient The materials used for slide bearings are
Mt tilting moment cast polyamide plates or similar materials
Fa axial force (gravity) which are relatively strong, with good
Fr radial force (thrust + side) sliding and hard-wearing properties.
Dr raceway diameter
Table 6-7 gives properties for a few types of
For a four-point ball bearing the coefficients relevant materials. The mechanical
can be set to properties listed apply to the unreinforced
material. The data given are only to be
= 0.006, a = 2.2, b = 0.5 and c = 1.9 considered as guidance, since data from
different manufacturers vary considerably.
Proper lubrication of the yaw bearing is of
great importance. It is advised to follow the Polyurethane (PUR) has a high ultimate
bearing manufacturers instructions strength combined with a large elongation at
regarding type and amount of grease, and failure, i.e. it exhibits a ductile behaviour. It
relubrication intervals. has good wearing properties and exhibits
constancy towards oil and grease. Different
One way of improving the lubrication in a types of PUR are categorised according to
bearing is periodically to let the bearing their Shore hardness. The Youngs
rotate through an angle equal to at least one modulus of elasticity of polyurethanes
bearing segment. grows exponentially with the Shore
hardness.
Bearing seals should be checked every six
months. Polyamide (PA) is characterised by a
combination of mechanical strength and
DNV (1992) recommends that the safety chemical resistance, good sliding properties
against failure for the slew ring toothing is to and high fatigue strength. PA is categorised
be 1.6. according to the number of C-atoms in its
molecules.
6 Nacelle 165
Guidelines for Design of Wind Turbines DNV/Ris
Plates used for slide bearings will normally A major difference between rolling bearings
need to be greased to obtain a sufficiently and slide bearings is that slide bearings
low friction torque and for corrosion involve larger frictional resistance. Slide
protection of the steel parts. bearings therefore require larger yaw motors
but less brake capacity than rolling bearings.
6.10.6 Yaw error and control for a long time. The wind turbine must,
therefore, be equipped with a cable twist
Yaw errors have a significant impact on the
sensor, which monitors the number of
blade loads, the rotor loads and the power
revolutions and informs the controller when
production. Hence, it is important that the
it is time to untwist the cables, usually after
yaw mechanism is efficient and reliable.
2 to 4 revolutions. The sensor can be
Even small yaw errors give rise to increased
designed quite simply as a switch on the
fatigue loads in the blades due to the
yaw drive which is activated once per
variation in the inflow angle during the
revolution.
rotation of the rotor that they cause.
With regard to redundancy in the cable twist
If the yaw error becomes too big, i.e. if it
monitoring system, the wind turbine can
falls outside the specified or assumed range
also be equipped with a pull switch which
of operation, the wind turbine must be
becomes activated when the cables become
stopped.
too twisted.
Special arrangements to prevent long-term
To emphasise the importance of preventing
yaw errors can be made. One such
cable twisting, it suffices to note that the
arrangement is formed by using a set of
turbine can easily experience 50-100
independent wind vanes. In case one of the
rotations in the same direction during one
vanes gets stuck, the vanes will misalign. An
year, if it is allowed to rotate unrestricted.
error message will then be produced, and the
wind turbine will be stopped.
6.10.8 Special design considerations
6.10.7 Cable twist Due to yaw moments being cyclic as
mentioned in Section 6.10.1, loose fits
In large power producing turbines, cables
should be avoided.
are needed to conduct the current from the
wind turbine generator down through the
tower. The cables will become twisted if the
turbine keeps yawing in the same direction
166 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
Bergmann, J., and R. Thumser, Forschung IEC, Wind turbine generator systems Part
fr die Praxis P 249, Synthetische 1: Safety requirements, International
Whlerlinien fr Eisenwerkstoffe, Standard, IEC61400-1, 2nd edition, 1999.
Studiengesellschaft Stahlanwendung e.V.,
Verlag und Vertriebsgesellschaft mbH, ISO, Rolling bearings Dynamic load
Dsseldorf, Germany, 1999. ratings and rating life, International
Standard, ISO281, 1st edition, 1990.
Bonus, Bonus Info, Special Issue, the Wind
Turbine Components and Operation, ISO, Rolling bearings - Static load ratings,
Autumn, Brande, Denmark, 1999. International Standard, ISO 76, 2nd edition,
1987.
The Danish Energy Agency, Technical
Criteria for the Danish Approval Scheme for Niemann, G., and H. Winter, Maschinenele-
Wind Turbines, Copenhagen, Denmark, mente, Band II, Getriebe allgemein, Zahn-
April 2000. radgetriebe Grundlagen, Stirnradgetriebe,
Springer-Verlag, Berlin, Germany, 1985.
Dansk Ingenirforening, Last og sikkerhed
for vindmllekonstruktioner, (in Danish), NWTC, Guideline DG03, Wind Turbine
DS472, 1st edition, Copenhagen, Denmark, Design, Yaw & Pitch Rolling Bearing life,
1992. National Renewable Energy Laboratory,
NWTC Certification Team.
Det Norske Veritas, Calculation of Gear
Rating for Marine Transmissions, DNV Peterson, R.E., Stress Concentration
Classification Note No. 41.2, Hvik, Factors, John Wiley and Sons, New York,
Norway, 1993. N.Y., 1974.
6 Nacelle 167
Guidelines for Design of Wind Turbines DNV/Ris
168 6 Nacelle
Guidelines for Design of Wind Turbines DNV/Ris
Tubular steel tower Tubular concrete Lattice tower Three-legged tower Guy-wired pole tower
tower
Figure 7-1. Various tower structures. From www.windpower.org, Danish Wind Turbine Manufacturers
Association.
7 Tower 169
Guidelines for Design of Wind Turbines DNV/Ris
Denmark is 4.2 m. An advantage of tubular critical loads. Such special cases should be
towers compared to other towers is that they investigated as appropriate.
are safer and more comfortable for service
personnel and others that have to enter and
climb the towers. 7.2 Design loads
As described in Section 4.3, the design loads
Lattice towers
are most often determined on the basis of an
Lattice towers are manufactured by means
of using welded steel profiles or L-section aeroelastic analysis or less frequently by
steel profiles. Since a lattice tower requires a simplified calculation, see Section 4.5.
only about half as much material as a freely Regardless of which analysis approach is
standing tubular tower with a similar used for these calculations, the loads should
stiffness, the basic advantage of lattice be calculated from a model, in which the
towers is reduced cost. It also gives less tower properties (geometry, materials,
wind shade than a massive tower. The major stiffness) are in agreement with the ones
disadvantage of lattice towers is their visual used in the final design. Thus, the design of
appearance, although this is a debatable the tower may demand an iterative
issue. Nevertheless, for aesthetic reasons procedure to get from an initial design to the
lattice towers have almost disappeared from final design with the correct stiffness.
use for large, modern wind turbines.
The design loads for fatigue are to be
Guyed pole towers determined by calculations, which are to be
Many small wind turbines are built with supplemented and verified by actual
narrow pole towers supported by guy wires. measurements from a prototype turbine.
The advantage is weight savings and thereby Since load measurements cannot be made
reduced costs. The disadvantages include until the turbine has been designed and
difficult access around the towers, which constructed, it is recommended to apply an
make them less suitable in farm areas. additional partial safety factor of 1.2 on
Finally, this type of tower is more prone to fatigue loads until measurements are carried
vandalism, thus compromising the overall out and become available. The use of an
safety. additional safety factor as an extra
precaution is meant to avoid a major
Other types redesign in the event of increased design
Some towers are designed as hybrids of the loads.
above types, for example the three-legged
tower for a 95 kW turbine in Figure 7-1. The extreme design loads can only be
determined by calculations, because these
loads cannot be measured due to the long
7.1 Load cases recurrence period between events.
The load cases to be investigated for tower When designing a turbine with fixed speed,
design are described in Section 4.1. the frequency of the rotor revolution is of
However, some special load cases may the utmost importance. This frequency,
apply particularly to the tower, for example often referred to as 1P, may induce
those that need to be considered when a long increased dynamic loads, e.g. due to rotor
tower is to be transported. In addition, the unbalances, wind shear and tower shadow.
erection of the tower on site may involve In addition, the higher Ps are of
importance, e.g. the 2P and the 3P, which
170 7 Tower
Guidelines for Design of Wind Turbines DNV/Ris
are the frequencies of blades passing the i.e. the 1P and 3P frequencies, respectively.
tower on a two- and three-bladed turbine, If it is confirmed that the tower frequency is
respectively. When designing a turbine with kept outside ranges defined as the rotor
variable speed, one must verify that the rotor frequency 10% and the blade passing
speed of the turbine does not operate in or frequency 10%, respectively, then there
near the first natural frequency of the tower, will normally not be any problems due to
see Section 7.3.1. load amplification arising from vibrations at
or near the natural frequency.
7.3 General verifications for towers For turbines with two generators or two
7.3.1 Dynamic response and generator speeds, this investigation should
resonance be performed for both corresponding blade-
passing frequencies. Special attention should
The first natural frequency of the tower be given to variable-speed turbines, in which
should always be measured in connection cases the turbine should not be allowed to
with the erection of a prototype turbine. operate in a frequency interval defined as
the eigenfrequency of the tower 10%.
It should be verified that the first natural
frequency of the tower does not coincide
with the rotor and blade-passing frequencies,
7 Tower 171
Guidelines for Design of Wind Turbines DNV/Ris
172 7 Tower
Guidelines for Design of Wind Turbines DNV/Ris
7 Tower 173
Guidelines for Design of Wind Turbines DNV/Ris
External loads, here denoted by index T, are Alternatively, loads must be combined by
assumed to include the dynamic effect or taking the maximum of each load
gust factor (referring to a quasi-static component from the particular load case
approach). where the most dominant load has its
maximum, or more conservatively they can
In MxT, it is particularly important to include be combined by combining the maxima of
the contribution from the possible eccen- the various load components regardless of
tricity of the nacelle. which load case they actually appear in.
The section force Fw(h) and the moment 7.4.3 Fatigue loads
Mw(h) from the wind load on the tower can
Combining fatigue loads is an even more
be calculated as
complicated task. When using the rain-flow
H
method as described in Section 4.4.1, the
load spectra for the different load
F w ( h ) = V ( z ) 2 D ( z ) C ( z ) dz components are normally not directly
h combinable.
174 7 Tower
Guidelines for Design of Wind Turbines DNV/Ris
DS410, according to which the critical wind vibrations, is the aerodynamic damping of
speed vr can be calculated as the blades and the nacelle.
7 Tower 175
Guidelines for Design of Wind Turbines DNV/Ris
Note also that a weld between two shells of The fatigue damage can be calculated using
different thickness as shown in Figure 7-6c the Palmgren-Miners rule as described in
is symmetrically tapered to avoid stress Appendix C.
concentrations. The slope of the tapering
should not be greater than 1:4. Whenever the weld is perpendicular to the
direction of loading and the material
In case of single-sided tapering, as shown in thickness t is greater than 25 mm, the fatigue
Figure 7-7, a stress concentration is strength fatd should be reduced in
introduced. accordance with the following formula
red,fatd = fatd(25/t)0.25
t2
In cases where it is not possible to achieve
the required design lifetime for a detail in an
analysis that includes partial safety factors,
inspection of the detail is required.
e However, it is still necessary to meet the
requirements to the nominal lifetime when
calculations are made without partial safety
t1 factors on the material properties included.
Note that this approach is only allowed in
Danish standards.
Figure 7-7. Single-sided plate tapering in tubular tower
The period of time until the first inspection
According to DNV (1987), the stress is carried out should at the most be set equal
concentration factor for the single-sided to the calculated design lifetime in the event
plate tapering can be calculated as that partial safety factors are included. Once
one or more inspections have been carried
e , out, subsequent inspection intervals should
SCFtaper = 1 + 6
t 2.5
be chosen depending on the results of the
t 2 1 + 1 previous inspections.
t2
7.4.6 Stress concentrations near
in which t1 and t2 are the plate thickness of hatches and doors
the lower and upper part of the tower shell,
The doors and hatches induce stress
respectively, and the eccentricity e is given
concentrations near the openings. This stress
by
concentration is traditionally represented by
a stress concentration factor (SCF), which
e = (t1 t 2 ) expresses the stress ratio between a sample
with and a sample without the opening. The
When using quality levels poorer than B SCF could be determined from parametric
according to DS/ISO 25817, it is equations, from a finite element analysis or
recommended to apply detail categories a by model experiments. A study by
corresponding number of levels lower. Jrgensen (1990) found a SCF of about 1.8
for a door opening. However, it is re-
commended that an individual SCF analysis
176 7 Tower
Guidelines for Design of Wind Turbines DNV/Ris
7 Tower 177
Guidelines for Design of Wind Turbines DNV/Ris
cr = f yd If
2
However, if the tower height H does not e = (e H)
1000
exceed 1.42 R R / t , then
is to be added to e.
cr = f yd
Finally, the following inequality must be
From theory of elasticity, the Euler force for fulfilled
a cantilever beam is given by
Nd N el M +N e
+ d 2 d < cr
1 2 2Rt N el N d R t
E d R 3 t
N el = 4
H2 Nd design axial force
Md design bending moment
The relative slenderness ratio for global R tower radius
stability is t tower shell thickness
H tower height
cr Ed design modulus of elasticity
r = Poisson's ratio
N el
fyd design yield stress
2Rt
7.4.8 Flange connections
The core radius k of a tube is given by
The tower sections and the connection to the
foundation are often adjoined with L or T
R
k= flange connections.
2
Figure 7-8 shows an L flange connection
For cold-formed welded towers, the along with the deformed shape of the upper
equivalent geometrical imperfection can side of the connection. Using the model in
now be calculated as Figure 7-8, the tension force Z in the tower
shell and the bolt force F are calculated from
e = 0.49( r 0.2)k
4 M
For welded towers, it can be calculated as Z=
Dn
e = 0.34( r 0.2)k and
178 7 Tower
Guidelines for Design of Wind Turbines DNV/Ris
R
Figure 7-8. L flange connections.
Figure 7-9. Tower shell dent tolerances.
The bolt stress, disregarding pretension, in
the L flange connection is given by When using the buckling strength analysis
approach, as described in Section 7.4.7, the
F assumed maximum dent tolerances are as
L = follows with reference to Figure 7-9:
Ast
Between circumferential seams
Md design bending moment measured along an arbitrary production
D diameter of tower shell
with a straight bar of length lr = 4 rt ,
n number of bolts
Ast bolt stress area however, not longer than 95 % of the
distance between adjoining
Since tower flanges often have a circumferential seams, the maximum
considerable thickness, the risk of brittle deviation from w should fulfil: w/lr <
fracture should be considered. This can be 0.01 (Figure 7-9 a).
done according to Eurocode 3, Annex C or measured along an arbitrary
DS412, Annex A. circumference with a shape with
curvature equal to the nominal outer
7.4.9 Corrosion protection radius of the cylinder and length lr =
4 rt , the maximum deviation from w
Corrosion protection of the tower should be
should fulfil w/lr < 0.01 (Figure 7-9 b).
in accordance with DS/R 454 and for
offshore turbines according to DS/R 464 or
similar recognised standards. The most
7 Tower 179
Guidelines for Design of Wind Turbines DNV/Ris
Elements of large thickness, such as tower 7.6 Example of tower load cal-
flanges, should be checked for stratification. culation
Further, the flanges must satisfy certain
7.6.1 Loads and responses
tolerances regarding straightness and
circularity to enable correct tightening of the The behaviour of the tower during extreme
connection. The stresses in the flange should loading is illustrated through an aeroelastic
be based on the smallest flange thickness, cf. calculation for a reference turbine with a
the tolerance specifications. rated power of 1800 kW. The main characte-
ristics of this turbine are given in Table 7-2,
and its dynamics are given in Table 7-3.
7.5 Access and working environ-
ment No. of blades 3
When designing the tower it must be Hub height 71.5 m
verified that proper access to the nacelle is Rotor diameter 66 m
possible. It must be assured that the access Rotational speed 15.9 rpm
complies with the requirements for Rated power 1800 kW
personnel safety and maintenance. In Tower tubular
addition, the working environment during Table 7-2. Main characteristics of reference wind
turbine.
blade inspection using an inspection hatch in
Figure 7-10. Tower response from analysis of a 1.8 MW turbine (left = alongwind
response, right = transverse wind response). Operational loads for 24 m/s wind speed.
180 7 Tower
Guidelines for Design of Wind Turbines DNV/Ris
An aeroelastic response calculation for an tions. These processes constitute the wind
operating turbine showing the bending speed process, the electric power and the
moments is shown as a function of time in horizontal force and moment at the
Figure 7-10. The response is calculated for a foundation level. Considering the variation
10-minute mean wind speed U10 = 24 m/s, of the processes and the time of operation, it
which is near the cut-out wind speed, and a is obvious that a proper selection of the
turbulence intensity IT = 11%. From Figure characteristic value must be rooted in
7-10 it appears that the response is highly probabilistic methods as described in
fluctuating with time, thus the design Section 4.3.1.
responses are to be determined from the pro-
bability distribution of the extreme response, Table 7-4 provides simulated estimates of
see Section 4.4. This fluctuation is due to the the first four central moments of the moment
dynamics of the turbine in conjunction with and horizontal force responses, viz. the
the stochastic turbulence field. mean, the standard deviation, the skewness
and the kurtosis of these two responses. The
7.6.2 Occurrence of extreme loads bottom lines in the table give estimates of
during normal power production the maximum moment and the simultaneous
value of the force, and of the maximum
The results of a response analysis based on
force and the simultaneous value of the
49 realisations (calculations with different
moment. The central estimates are denoted
seeds) of the turbulence field are shown in
by in the table. Statistical uncertainty in
Table 7-4. The load case is analysed for
the estimates is also included in terms of the
U10 = 24 m/s and IT = 11%, and the
standard deviations of the estimates, denoted
simulation time for each realisation is 10
minutes. by in the table. It appears that the
maximum of the moment response process
Table 7-4 shows results for four important and the maximum of the force response
processes as interpreted from the 49 calcula- process do not occur at the same time.
Configuration of turbine
Blades: Blades:
In normal position 90o pitched
Mode shape Freq. [Hz] Damp. (%) Freq. [Hz]
1st tower transversal 0.418 6.0 0.417
1st tower longitudinal 0.419 6.0 0.420
1st rotor torsion 0.805 5.0 0.704
1st rotor torsion 0.979 1.002
1st asymmetric rotor (yaw) 1.000 1.064
1st symmetric rotor (flap) 1.067 3.1 1.769
1st edgewise mode 1.857 3.1 1.032
2nd edgewise mode 1.045
Table 7-3. Dynamic properties of reference wind turbine.
7 Tower 181
Guidelines for Design of Wind Turbines DNV/Ris
etc.). For each load case at least five configuration. The wind is acting
different seeds should be used. perpendicular to the nacelle, i.e. with a full
drag on the nacelle, and the blades are
Run Wind Power Moment Horis. pitched to give a full drag also on the blades.
statistics M force, F
Mean 24.09 1810 14323 210 The extreme loads calculated from 60
0.00 5.5 28.8 0.413 simulations of the response processes are
Std. dev. 2.63 119.6 2046 30.9 shown in Table 7-5. The mean wind speed is
34 m/s and the turbulence intensity is 10%.
0.01 4.2 152 2.3
The central estimates are denoted by in the
Skewness -0.052 -0.187 -0.028 -0.055
table. Statistical uncertainty in the estimates
0.218 0.085 0.127 0.239
is also included in terms of the standard
Kurtosis 2.953 6.650 2.950 2.953
deviations of the estimates, denoted by in
0.377 0.550 0.256 0.239 the table. From the data it appears that the
Upcross. 6.703 1.714 2.520 1.678 statistical variations of the various responses
period 0.881 0.052 0.748 0.896 under consideration are significant also in
max[M] 28.2 1986.7 21018 312 the parked condition.
and simul. 2.9 90.2 777 12
max[H] 27.4 1978.0 20907 314 Run Wind Moment Horis.
and simul. 1.9 95.9 812 12 statistics M force, F
Table 7-4. Extreme loads during power production Mean 33.51 12293.7 234.2
(simulation time: 10 minutes). 0.001 23.3 0.45
Std. dev. 3.333 2555.2 45.79
2. Calculate the statistics (as exemplified
in Table 7-4) for each executed simu-
0.005 79.3 1.09
lation (at least 10 minutes). Skewness -0.027 0.208 0.206
3. Use the procedures in Section 4.4.2 to 0.205 0.203 0.210
project the responses calculated from a Kurtosis 2.939 3.091 3.055
10-minute simulation to a longer period 0.248 0.266 0.233
of operation (e.g. more days max[U] 43.74 17515.3 331.5
calculated from the mean wind speed and simul. 1.03 2362.6 26.4
distribution see Chapter 3). max[M] 40.30 21490.8 395.9
4. Choose the characteristic value of the and simul. 1.72 1052.0 17.4
response as described in the code. (Not max[H] 40.38 21449.9 396.5
many codes deal with this issue, and and simul. 1.67 1056.4 17.0
those that do, do not necessarily define Table 7-5. Extreme loads for parked turbine (simulation
the characteristic value in the same time: 10 minutes)
manner).
5. Apply a safety factor as described in the The characteristic extreme response should
codes (the revised edition of the Danish be chosen equal to the expected maximum
code of practice DS472 specifies f = response in a 10-minute period. This
1.5) to calculate the design value of the approach is justified only if the coefficient
response. of variation on the maximum value is
sufficiently low, see Mrk. An approach by
7.6.3 Extreme loads parked turbine Davenport can be used for this purpose, see
Section 4.4.2. The corresponding design
An extreme load analysis is also performed
value is found by applying the appropriate
with the reference turbine in a parked
182 7 Tower
Guidelines for Design of Wind Turbines DNV/Ris
partial safety factor (the current edition of given in Table 7-7, calculated for 107
DS410 and the revised edition of DS472 equivalent load cycles. Note that the
specify f = 1.5). uncertainty in the lifetime fatigue loading is
still significant with a coefficient of variance
7.6.4 Fatigue loading of 6%. This may call for a safety factor for
fatigue loading different from 1.0.
The fatigue loads are also fluctuating loads,
thus one must assure that a sufficient
Wind Statistical Equivalent
number of simulations are performed before
speed parameter load
the design of the tower can be carried out. A
[m/s] [kNm]
1.5 MW stall-controlled turbine with a
tubular tower as reported in Thomsen (1998) 7 828.7
is considered as an example, and the fatigue COV 17%
loads are as shown in Table 7-6. The table is 10 1511.1
calculated with the Whler curve exponent COV 10%
m = 3, which is the value traditionally 15 3151.1
selected for steel, and the equivalent number COV 13%
of cycles is 600. The turbulence intensity is 20 6059.4
set to 15%. COV 15%
24 7703.1
Table 7-6 shows the variation in the COV 15%
equivalent load as a function of the mean Table 7-6. Equivalent load for tower bending in a 10-
wind speed. The table is based on a large minute simulation.
number of simulations for each wind speed,
and both the mean and the coefficient of Quantity Stat. Eq. load [kNm]
variation of the load estimates are included, Par.
denoted by and COV, respectively. The Req 10292.42
coefficient of variation appears to be rather COV 6%
large (greater than 10%) in this example, neq, L 107
despite the large number of simulations used Table 7-7. Equivalent load for tower bending in a 10-
for the estimations. This indicates the minute simulation.
importance of carrying out a sufficient
number of simulations. It is recommended The dependency of the damage-equivalent
not to use less than five simulations, see load on the wind speed and on the
Danish Energy Agency (2000), however, turbulence intensity is shown in Figure 7-11.
five simulations may in many cases be The upper half of the figure shows the
insufficient. equivalent load in a 10-minute simulation
period as a function of the wind speed. Also,
The equivalent load as shown in Table 7-6 three levels of turbulence intensity are
must be corrected with respect to the proper analysed. From the figure, it is clearly seen
operation time at different wind speeds that the fatigue damage increases with the
(adjusting for the distribution of the mean wind speed and turbulence intensity.
wind speed resulting in the lifetime Moreover, the uncertainty in the damage
equivalent load). The mean wind velocity is increases with increasing wind speed and
assumed to follow a Weibull distribution turbulence. The error bars in Figure 7-11
with scale parameter A = 10 m/s and slope illustrate the mean of the fatigue load
parameter k = 2.0. The lifetime fatigue plus/minus one standard deviation.
equivalent load for a 20-year lifetime is
7 Tower 183
Guidelines for Design of Wind Turbines DNV/Ris
184 7 Tower
Guidelines for Design of Wind Turbines DNV/Ris
13
17
21
25
29
33
19 22 +10 18 5,9 Load combination number
20 24 +10 18 5,3
21 Low Start 0,0
LC's: starts/stops
7 Tower 185
Guidelines for Design of Wind Turbines DNV/Ris
186 7 Tower
Guidelines for Design of Wind Turbines DNV/Ris
8 Foundations 187
Guidelines for Design of Wind Turbines DNV/Ris
borings and in-situ testing in order to get an interpret parameters such as the undrained
understanding of the soil stratification with- shear strength of clay. The extent to which
in a given area, and for offshore locations the various types of in-situ tests and
of the seabed topography within that area. laboratory tests are required depends much
Such a survey can provide guidelines for on the foundation type in question, for
selection of a suitable foundation site within example, whether it is a piled foundation or
the area, if not already decided. Geophysical a gravity-based foundation.
surveys are carried out by means of seismic
methods. The recommendations of the Danish
Technical Criteria for Type Approval of
A geotechnical investigation consists of: Wind Turbines (Danish Energy Agency,
soil sampling for laboratory testing 1998) specify that a geotechnical report
in-situ testing of soil from the geological and geotechnical
Soil investigations should be tailored to the surveys shall be prepared. This geotechnical
geotechnical design methods used. The field report should contain sufficient information
and laboratory investigations should about the site and its soils, e.g. in terms of
establish the detailed soil stratigraphy across soil strength and deformation properties, to
the site, thus providing the following types allow for design of the foundation with
of geotechnical data for all important soil respect to:
layers: bearing capacity
data for classification and description of stability against sliding
the soil, such as settlements
unit weight of sample foundation stiffness
unit weight of solid particle need for and possibility of drainage
water content static and dynamic coefficients of
liquid and plastic limits compressibility
grain size distribution sensitivity to dynamic loading
parameters required for a detailed and The geotechnical report is further required to
complete foundation design, such as contain identification of soil type at foun-
permeability tests dation level, classification of environment,
consolidation tests and estimation of highest possible water
static tests for determination of shear table. For further details, reference is made
strength parameters such as friction to Danish Energy Agency (1998).
angle for sand and undrained shear
strength cu for clay (triaxial tests and 8.1.2 Recommendations for gravity
direct simple shear tests) based foundations
cyclic tests for determination of strength Insofar as regards a gravity-based
and stiffness parameters (triaxial tests, foundation, an extensive investigation of the
direct simple shear tests and resonant shallow soil deposits should be undertaken.
column tests) This investigation should cover soil deposits
to a depth, which is deeper than the depth of
Sampling can be carried out with and any possible critical shear surface. Further,
without drilling. The cone penetrometer test all soil layers influenced by the structure
(CPT) and various vane tests form the most from a settlement point of view should be
commonly used in-situ testing methods. thoroughly investigated. This also holds for
Results from such in-situ tests can be used to all soil layers contributing to the foundation
188 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
8 Foundations 189
Guidelines for Design of Wind Turbines DNV/Ris
trial failure surfaces will have to be analysed vertical force V relative to the centre line of
in order to find the most critical one with the foundation. Reference is made to Figure
respect to stability. 8-1, and the eccentricity is calculated as
2
2Mz 2Mz
H'= + H2 +
LC l eff l eff
V
in which leff is the length of the effective
H
f [kN/m2]
e [m]
Effective foundation area
For use in bearing capacity analysis an
rupture 2 rupture 1 effective foundation area Aeff is needed. The
effective foundation area is constructed such
that its geometrical centre coincides with the
Figure 8-1. Loading under idealised conditions. load centre, and such that it follows as
closely as possible the nearest contour of the
In the following, it is assumed that H and V true area of the foundation base. For a
are design forces, i.e. they are characteristic quadratic area of width b, the effective area
forces that have been multiplied by their Aeff can be defined as
relevant partial load factor f. This is
indicated by index d in the bearing capacity Aeff = beff l eff
formulas, hence Hd and Vd. The load centre,
denoted LC, is the point where the resultant
in which the effective dimensions beff and leff
of H and V intersects the foundation-soil
depend on which of two idealised loading
interface, and implies an eccentricity e of the
190 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
scenarios leads to the most critical bearing Reference is made to Figure 8-2. The
capacity for the actual foundation. effective area representation that leads to the
poorest or most critical result for the bearing
e Aeff capacity of the foundation is the effective
area representation to be chosen.
leff
e
Aeff = 2 R 2 arccos( ) e R 2 e2
R
Aeff
be = 2(R e )
e
and
2
b
le = 2 R 1 1
2R
LC2
leff
Aeff
beff
R
Figure 8-2. Quadratic footing with two approaches to
how to make up the effective foundation area. LC leff
le
Scenario 1 corresponds to load eccentricity
with respect to one of the two symmetry
axes of the foundation. By this scenario, the
following effective dimensions are used:
beff
beff = b 2 e , l eff = b be
Figure 8-3. Circular and octangular footings with
effective foundation area marked out.
Scenario 2 corresponds to load eccentricity
with respect to both symmetry axes of the
foundation. By this scenario, the following The effective foundation area Aeff can now
effective dimensions are used: be represented by a rectangle with the
following dimensions
beff = leff = b e 2
8 Foundations 191
Guidelines for Design of Wind Turbines DNV/Ris
192 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
Shape factors s: H
iq = ic = 1 +
V + Aeff c cot
beff
s = 1 0.4
l eff i = iq
2
beff
s q = s c = 1 + 0.2
l eff H
ic0 = 0.5 + 0.5 1 +
Aeff cud
Inclination factors i:
2 The bearing capacity is to be taken as the
Hd
i q = i c = 1 smallest of the values for qd resulting from
V + A c cot the calculations for Rupture 1 and Rupture
d eff d d
2 2.
i = iq
Sliding resistance of soil
Undrained conditions, = 0: Foundations subjected to horizontal loading
must also be investigated for sufficient
N c0 = + 2 sliding resistance. The following criterion
applies in the case of drained conditions:
s c0 = s c
H < Aeff c + V tan
8 Foundations 193
Guidelines for Design of Wind Turbines DNV/Ris
when the pile, subjected to its loading, is resistance. Note that the local axial
displaced relative to the soil. resistance and the torsional resistance,
respectively, are interdependent, since they
For design of piles, it is common to both arise from the skin friction against the
disregard a possible interaction between the pile surface. The axial, lateral and torsional
axial pile resistance and the lateral pile pile displacements and corresponding soil
resistance locally at any point along the pile reactions for a single pile subjected to
and to treat these two forms of resistance as external loading at its head are illustrated in
being independent of each other. The Figure 8-4.
argument for this is that the soil near the soil
surface principally determines the lateral Note that for design of piles, it is important
resistance without contributing much to the to consider effects of the installation
axial resistance, while the soil further down procedure. For example, the stress history
along the pile toward the pile tip principally during pile driving contributes significantly
determines the axial resistance without to fatigue loading and needs to be con-
contributing much to the lateral capacity. sidered for design against fatigue failure in
The axial and lateral resistance models the pile wall.
presented in the following conform to this
assumption of independence between local 8.3.1 Pile groups
axial resistance and local lateral resistance.
For foundations consisting of pile groups,
i.e. clusters of two or more piles spaced
closely together, pile group effects need to
be considered when the axial and lateral
resistance of the piles is to be evaluated.
There are two types of group effects:
the total capacity of the pile group is
less than the sum of the capacities of the
individual piles in the group, because of
overlap between plastified soil zones
around the individual piles. A lower
limit for the axial pile group capacity is
the axial capacity of the envelope pier
that encloses all the piles in the group
and the soil between them.
larger pile displacements of a given
Figure 8-4. Single loaded pile with displacements and load result for an individual pile when it
reactions, from Reese et al. (1996)
is located in a pile group than when it is
an only pile, because its supporting soils
Note, however, that on a global level the
will have displacements caused by loads
lateral pile capacity and the axial pile
transferred to the soil from adjacent
capacity, resulting from local resistance
piles in the group. This type of group
integrated along the length of the pile, may
effect is also known as pile-soil-pile
interact, because second-order effects may
interaction. For practical purposes, such
cause the axial loading to influence the
pile-soil-pile interaction can often be
lateral behaviour of the pile. When the pile
reasonably well represented by means
is subjected to torque, torsional resistance
of Mindlins point force solutions for an
will be set up in addition to axial and lateral
elastic halfspace.
194 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
8.3.2 Axial pile resistance in which values in the range 0.1-0.25 are
suggested for pile lengths exceeding 15 m.
Axial pile resistance is composed of two
parts
(3) semi-empirical method, by which the
accumulated skin resistance
soil deposit is taken as one single layer, for
tip resistance which the average skin friction is calculated
as
For a pile in a stratified soil deposit of N soil
layers, the pile resistance R can be expressed
f S = ( p 0 m '+2c um )
as
N p0m' average effective overburden pressure
R = R S + RT = f Si ASi + q T AT between the pile head and the pile tip
i =1
cum average undrained shear strength along
the pile shaft
fSi average unit skin friction along the pile dimensionless coefficient, which
shaft in layer i depends on the pile length as shown in
ASi shaft area of the pile in layer i Figure 8-5
qT unit end resistance
AT gross tip area of the pile Hence, by this method, the total shaft
resistance becomes RS = fS AS, where AS is
Clay. For piles in mainly cohesive soils, the the pile shaft area.
average unit skin friction fS may be
calculated according to For long flexible piles, failure between pile
and soil may occur close to the seabed even
(1) total stress methods, e.g. the method, before the soil resistance near the pile tip has
which yields been mobilized at all. This is a result of the
flexibility of the pile and the associated
8 Foundations 195
Guidelines for Design of Wind Turbines DNV/Ris
0,45
qp = Nq p0 ql
0,4
0,35
Nq can be taken from Table 8-1
0,3
ql limiting tip resistance, see Table 8-1
0,25 for guidance
0,2
0,15
0,1
The unit tip resistance of piles in cohesive
0,05
soils can be calculated as
0
0 10 20 30 40 50 60 70 qp = Nc cu
Pile length (m)
Nc = 9
Figure 8-5. Coefficient vs. pile length
cu undrained shear strength of the soil at
the pile tip
For deformation and stress analysis of an
axially loaded flexible pile, the pile can be
modelled as a number of consecutive
column elements supported by nonlinear
springs applied at the nodal points between
the elements. The nonlinear springs are
denoted t-z curves and represent the axial
load-displacement relationship between the
pile and the soil. The stress t is the inte-
grated axial skin friction per unit area of pile
surface and z is the relative axial pile-soil
displacement necessary to mobilize this skin
friction.
196 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
method, a nonlinear relation exists between For clay, the initial shear modulus of the soil
the point of origin and the point where the to be used for generation of t-z curves can be
maximum skin resistance tmax is reached, taken as
t G0 = 2600 cu
z IF r f
R t max
z=t ln for 0ttmax However, Eide and Andersen (1984) suggest
G0 t
1 rf a somewhat softer value according to the
t max
formula
8 Foundations 197
Guidelines for Design of Wind Turbines DNV/Ris
d4y d2y
EI + Q A p( y) + q = 0
dx 4 dx 2
with
d3y dy d2y
EI 3
+ QA = Q L and EI =M
dx dx dx 2
198 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
Several methods are available for For cyclic loading and X > XR, the p-y curve
representation of the p-y curves that are can be generated according to
essential in solving the differential equation
for a laterally loaded pile. For construction pu y 1 / 3
of p-y curves, the type of soil, the type of ( ) for y 3yc
p = 2 yc
loading, the remoulding due to pile 0.72 pu for y > 3yc
installation and the effect of scour should be
considered. The most commonly applied
procedures for construction of p-y curves are For cyclic loading and X XR, the p-y curve
those given by DNV (1992) and API (1987). can be generated according to
For piles in clay, the procedure according to
DNV is presented below. For sand, the pu y 1 / 3
2 (y ) for y 3y c
procedures according to DNV and API are c
For static loading, the p-y curve can be The p-y curve can be generated according to
generated according to
kX
p = Ap u tanh( y)
Ap u
pu y 1 / 3
( ) for y 8yc
p = 2 yc
pu for y > 8yc
8 Foundations 199
Guidelines for Design of Wind Turbines DNV/Ris
200 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
8 Foundations 201
Guidelines for Design of Wind Turbines DNV/Ris
G 0 = 1000 K 0 '
202 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
8 Foundations 203
Guidelines for Design of Wind Turbines DNV/Ris
ES is the value of E at a depth z equal to the analyses. As a rule of thumb, the natural
pile diameter D, and EP is the Youngs frequency of the tower will be reduced by
modulus of the pile material. 0% to 5%, when the assumption of a rigid
foundation (fixed-ended tower) is replaced
The formulas given for foundation by a realistic finite foundation stiffness.
stiffnesses in the tables in this section can be Under special conditions this error may
used to calculate spring stiffnesses to however be up to 20%.
support the tower in aeroelastic wind turbine
Table 8-5. Circular footing on stratum over bedrock or on stratum over half-space.
On stratum over bedrock On stratum over half-space
204 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
8 Foundations 205
Guidelines for Design of Wind Turbines DNV/Ris
206 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
s < 70 N/mm2
Acef=2b(h-hef)
If the stress variation does not fulfil this
equation, more refined fatigue verification
must be carried out, i.e. using the Palmgren-
Miner cumulative damage law. The relation-
ship between the characteristic fatigue
hef
strength fat and the characteristic fatigue
h
value nfat is given by the expression
1
k 1012 k 1012 m
n fat = fat =
( fat )m n fat
b
The Whler curves are characteristic values Figure 8-13 Active concrete area in tension zone for
calculation of crack-width
and should be reduced by partial safety
factors according to the chosen safety level.
The external loads at which the crack widths
The Whler constants m and k, according to
are determined are usually 60% of the wind
DEA (1998) can be used.
turbines maximum operational loads.
8.5.2 Crack-width
The stress in the reinforcement s [N/mm2]
Crack widths appear in RFC from the due to external load can be found by means
influence of the loads and as a result of of the elasticity theory assuming that the
deformations due to temperature or surrounding concrete stress (in tension) is 0.
shrinkage. The size of the crack width w can The external force is the sectional moment
be found acc. to DS411 (DS 411, 1999) as a M only:
function of the reinforcement stress as
follows M
Concrete stress c max =
b b hef2
w = 5 10 5 s a w , [mm]
Reinforcement stress s max = c max
The crack parameter aw [mm] is calculated
as the ratio between the active concrete area M sectional bending moment
Acef and the sum of the crack width decisive As reinforcement area
diameters dw of the reinforcement bars in the b and hef see Figure 8-13
tension zone
Parameters b and to be extracted by
Acef iteration from formulas below:
aw =
d w
As
=
Acef is to be calculated as the maximum b hef
concrete area of which the centre of gravity
COG is coincident with the COG of the 2
tension reinforcement. = ( + 1 1)
8 Foundations 207
Guidelines for Design of Wind Turbines DNV/Ris
208 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
of horizontal earth pressures against the deeper below the sea floor, and it will
piles at some depth. The gravity base contribute to prevent scour from occurring
foundation primarily transfers such forces around and in particular underneath the
directly to horizontal shear forces in the soil foundation, especially in sand. Gravity-
at the foundation-soil interface. The based foundations may often be feasible
monopile also transfers bending moments from a technical point of view, but may not
about the seabed as horizontal earth always come out economically feasible.
pressures against the pile. The tripod They form a well-tested concept, which is,
primarily decomposes such overturning however, also a time and resource deman-
moments into axial forces in the three legs, ding concept. With these pros and cons of
and these forces are then transferred to the the various foundation types in mind, it
soil as axial shear forces at the pile-soil should be noted that the type of foundation
interfaces. The gravity base foundation to be chosen also depends heavily on the
transfers overturning moments about the actual soil profile and soil properties on the
seabed by a varying vertical contact pressure site.
over the foundation-soil interface. During
design, it is important to make sure that this Two of the three distinct foundation struc-
contact pressure in extreme load situations tures for offshore wind turbine installations
never takes on negative values. Any are dealt with in the following sections, viz.
significant, negative contact pressure would the monopile and the tripod structures.
correspond to a tension, which can hardly be
transferred unless purely undrained 8.6.2 Monopile
conditions with development of a suction
General Description
can be counted on, and which may lead to an
Monopiles have traditionally been used as
undesirable separation between foundation
the preferred foundation solution for some
and soil. However, a limited lift of the
types of marine structures. Piles with
foundation of limited duration during
diameters up to about 1 m have been used to
extreme loading is allowable.
support lighthouses and moorings.
Equipment to install piles of diameters up to
Each of the three offshore foundation
3-4 m to large penetration depths is avai-
concepts have their pros and cons. The
lable and makes monopiles a feasible foun-
monopile is usually an attractive foundation
dation alternative for offshore wind turbines.
solution for wind turbines to be placed in
An example of a monopile foundation is
shallow waters and smooth seas. This
depicted in Figure 8-14.
concept is attractive also because it offers a
fast installation. The tripod is preferred over
Advantages of a monopile foundation
the monopile in harsher climates and less
include:
shallow waters, however, the solution with
suction buckets under the three legs is only a fast and highly automated installation
feasible for foundations in clay and will not with no prior preparation of the seabed
work in sand or other permeable soils. The simple fabrication
tripod concept is a light-weight structure, The monopile foundation of a wind turbine
which is usually cost-efficient. It is an consists of three basic parts: the bare pile, a
attractive feature of gravity-based foun- conical transition to the tower that it
dations that they can be equipped with more supports, and a boat landing. By alteration
or less shallow skirts under their bottoms. of the conical transition, the foundation can
This will move the depth of load transfer easily be adjusted to fit different tower
diameters.
8 Foundations 209
Guidelines for Design of Wind Turbines DNV/Ris
210 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
8 Foundations 211
Guidelines for Design of Wind Turbines DNV/Ris
when the length of the pile is increased. The design life of a wind turbine foundation
Reference is made to Figure 8-15 and usually varies between 30 and 50 years. The
Barltrop (1991). For details about methods maximum corrosion is located in the splash
for analysis of laterally loaded piles, zone. The maximum bending stresses in the
reference is made to Section 8.3. pile during extreme loading are found below
this zone. Thus extreme stability will be
Connection between monopile and tower maintained even if a rather large corrosion
The monopile foundation concept includes a depth is found in the splash zone.
conical connection piece between the
monopile and the wind turbine tower. The However, a corroded surface generates areas
connection piece comprises the access with very large stress concentrations and
platform as an integral part. At the top of the cracks will propagate faster in corroded steel
conical connection piece, a flange for con- than in uncorroded steel. Therefore, it is
nection of the wind turbine tower is located. recommended to use corrosion protection in
At the bottom of the conical connection corrosive environments, even if the structure
piece, an in-situ connection to the monopile is capable of accommodating the stresses
foundation is to be established. over the extent of the corroded cross-section
(DS 412, 1998).
The connection between the pile and the
conical connection piece is an essential part With a view to the above numbers, this calls
of the monopile foundation concept. The for implementation of a corrosion protection
basic requirements to this connection are: system.
sufficient strength against extreme and
fatigue loading Three primary options exist for corrosion
easy installation protection of a monopile:
low cost coating (vulnerable to damage during
Two alternative concepts are considered installation)
feasible: cathodic protection by sacrificial anodes
flange connection (fast installation) (a large number of anodes are required
welded connection (strong connection) for a 50-year design life)
Reference is made to Lyngesen and impressed current (facilitated by the
Brendstrup (1997) for more details. presence of electrical power supply)
For a design life in excess of 30 years, the
Corrosion protection preferred option for monopile foundations is
As the surface of the pile is exposed to a protection by use of impressed current.
harsh environment due to waves, abrasion
from suspended sediments and a high The pile has a large surface, but normally
salinity, a large corrosion rate is expected. only a part of this requires protection. The
For unprotected steel, the extreme surface surface exposed to wave action, including
corrosion of an exposed steel pile in ocean the splash zone, shall be protected.
water can be estimated to (DS 464/R, 1988): Depending on the type of sediment, the
0 years after installation 0 mm potential for scour, etc., also a part of the
10 years after installation 8 mm pile below mudline may need protection.
20 years after installation 11 mm Typically, the area to be protected extends
30 years after installation 14 mm down to a level 2 to 10 m below mudline.
Below this limiting level, protection of the
pile surface is usually not required.
212 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
Inside the hollow pile water will be present, shock wave amplitude, and the larger is the
but there will only be little exchange of possibility for pile refusal.
water. Therefore, corrosion on the inside of
the pile is expected to be limited. However, Driving using vibrators
it is recommended that internal zones are In frictional soils, the pile may be vibrated
protected either with coating or with into the soil. Vibration is a very fast
cathodic protection (DNV, 1998). If the pile installation method even compared to
is filled with sand, corrosion will be driving by hammer. Furthermore, it does not
minimised or even prevented (DS 464/R, generate very large shock waves in the soil
1988). that could otherwise have caused damage of
adjacent structures. However, only a few
Installation very large vibration hammers are available
Driving of large diameter piles is well- world-wide. The vibrations generate a lique-
known from the offshore industry. faction of the soil locally around the pile,
Generally, at least three different pile- thus reducing the side adhesion to a mini-
driving methods are available. These mum. Then the pile penetrates into the soil
methods are described briefly in the mainly due to its own weight.
following.
Driving using drilling or excavation
Driving using piling hammer For installation of hollow piles, drilling or
Pile driving by means of a hammer is the excavation may be applied. The soil in the
oldest technique. In its simplest version, it core and below the pile is excavated,
consists of a ram dropped on the top of the generating a hole with a diameter usually
pile from a certain height. Today, hydraulic slightly larger than that of the pile. Then the
fluid is widely used for lifting the ram and pile penetrates into the soil due to the mass
further accelerating it during the downward of the pile. If side adhesion or lateral resis-
stroke. Hydraulic piling hammers are the tance is required, grout may be injected in
most efficient compared to diesel and steam the annulus between the pile surface and the
hammers. The technique capitalises on the soil.
shock wave which is generated in the pile by
the impact from the hammer. The transfer of Notes on pile refusal
the shock wave to the soil generates plastic The pile has to be driven to a particular pen-
deformations of the soil and thereby moves etration depth, which is required to enable
the pile downwards. the pile to carry its design lateral load.
However, pile refusal may be encountered
So-called pile refusal occurs when the shock before the final penetration depth is reached.
wave amplitude is not large enough to create This may happen if the piling hammer does
plastic soil deformations. The penetration of not have sufficient excess capacity
the pile then ceases. Damping of the shock compared to the theoretically calculated
waves caused by soil adhesion along the pile driving resistance.
surface may contribute to pile refusal, in
particular, in cohesive soils. In this context, In case of pile refusal, a decrease of up to
it is important to be aware that the smaller 50% in the driving resistance can be
the ratio between the cross-sectional pile obtained by removing the soil plug inside
area and ram area is, the longer is the shock the pile. In sandy and silty soils, this can be
wave transmission time, the smaller is the achieved by jetting. In clay and tills, this can
be achieved by using an auger. The driving
8 Foundations 213
Guidelines for Design of Wind Turbines DNV/Ris
resistance can also be reduced by installation inspection on the wind turbine tower is
of an external driving shoe on the pile tip, carried out.
however, this may also lead to a reduced
lateral capacity and may therefore not Inspection of Scour Protection
always be recommended. Scour will occur around the monopile. To
provide protection against local scour
Maintenance around the pile, scour protection is installed.
Inspection and maintenance are to be carried Throughout the lifetime of the wind turbine
out to ensure a sufficient structural strength. the scour holes around the pile are to be
measured from time to time. The holes can
Crack Inspection be measured by means of a 3-D echo soun-
Cracks may be initiated at the weld between der. The scour protection is not absolutely
the monopile and the conical transition stable and needs to be maintained. When an
section. Therefore, the weld is to be inspec- unacceptable degree of scour occurs,
ted for fatigue cracks. The inspection will additional scour protection is to be installed.
consist of:
visual inspection and/or non-destructive Spring Inspection
testing (NDT): When the foundation is located in an
magnetic powder inspection (MPI) environment where ice loading may occur, it
eddy current is recommended to perform an annual
ultrasonic examination inspection every spring to verify whether ice
loading has caused damage to any structural
Depending on the fabrication method of the members or coating during the past winter.
monopile, welds may also be located in the In case of damage, repair is to be carried
submerged zone above mudline. These out.
welds also need to be inspected. The extent
of each inspection will be determined as part Dismantling
of the inspection planning. The interval In the following, two different methods are
between each inspection is to be determined outlined for removing the monopile
on the basis of requirements to the lifetime foundation from the seabed after termination
and on the predicted accumulated fatigue of its service life.
damage.
1. Cutting off the piles is a method that can
Inspection of Cathodic Protection be used in general. By this method, the
The monopile foundation is protected monopile is cut below the mudline. The
against corrosion by means of an impressed method is based on the use of a cutting tool
current system. A reference control point lowered down inside the pile. The subsea
(the point with minimum potential) on the cutting is performed by a high pressure jet,
monopile is selected. The electrical potential consisting of water mixed with a non-pollute
of this point is measured and is reported grinding compound (sand).
through the wind turbine monitoring system.
In case the anode disappears or the area of The procedure consists of the following
the anode becomes too small, the potential at steps:
the control point will increase, and the anode the soil core inside the tubular is
needs to be replaced. Further, it is removed to the elevation of the cut line
recommended to perform a visual check of the cutting tool is lowered to the
the cathodic protection system whenever an elevation of the cut line, inside the pile
214 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
8 Foundations 215
Guidelines for Design of Wind Turbines DNV/Ris
216 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
at very low water depths, the distance MP(N) Design plastic moment capacity as
between the upper and lower braces a function of the axial force
becomes too short for the forces to be N Axial force in pile
transferred in any reasonably sized y Nominal steel yield stress
cross-section. WP Plastic section modulus
a solution with the braces protruding the D Pile diameter
water surface is not desirable due to the t Pile wall thickness
possible interlocking or packing of ice A Steel area of pile
and the possibility of collision with
service vessels. As a minimum, the result of the analysis
must contain the pile utilization ratio and a
The tripod causes relatively little blocking verification that the penetration depth is
for wave action and for current flows which sufficient. All piles are allowed to reach the
is of importance for environmentally sensi- design plastic moment. The forces and
tive areas. This even applies when a large moments at the pile heads shall be in
number of foundations are installed as is the equilibrium with the design loading acting
case in an offshore wind farm. upon the structure and shall be smaller or
equal to the allowable design values.
Geotechnical analysis piles
In the ultimate limit state, the foundation In both elastic and plastic analyses, the
analysis shall consist of an elastic and a following features must be taken into
plastic analysis. The two analysis checks can account:
be described as follows: linear-elastic superstructure that
interconnects the piles
In the elastic analysis, a check is carried out non-linear behaviour of the piles
of the stresses in the piles and the structure laterally and axially
with design loads (ULS) and design material pile group effects
properties. One pile only is allowed to reach second-order moments in the piles
its design maximum allowable stress. The soil plug length taken as 0.9 times
allowable design value of the von Mises length of pile embedded in soil
stress is the yield stress fy divided by the
partial safety factor for steel strength. The The procedures for development of t-z and
result of the analysis must as a minimum q-w curves for the axial bearing capacity and
contain the utilisation ratio of the pile steel. p-y for the lateral bearing capacity can be
In the plastic analysis, the entire tripod done by using the API approach (API,
structure is checked using design loads 1993).
(ULS) and design material properties. All
the piles are permitted to yield or fail When characteristic values for the skin
completely, provided that the tripod as a friction and tip resistance are given, the t-z
whole can absorb the design loads. curves for axial pile-soil interaction can be
established by means of any relevant
The plastic design moment shall be recognised computer program. The shape of
calculated by the curves is fully described by
[
M P (N ) = WP cos 2 YN A ,] 2 z G
t= t max arctan
D 3 ( D 2 t ) 3 3 D t max
WP = 6
8 Foundations 217
Guidelines for Design of Wind Turbines DNV/Ris
218 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
load, multiplied by a dynamic amplification lead to a maximum stress above the upper
factor (DAF). Vertical ice load may destabi- joint at two points on a line perpendicular to
lize the structure, because it is applied as an the axis of the applied moment. The stresses
upward load. will vary linearly between these points, but
as the wind direction changes, also the
In the serviceability limit state (SLS), the locations of these points move. Thus, the
maximum deflections are checked with loading (both in terms of magnitude and in
special emphasis on the allowable tilt of the terms of number of stress cycles) shall be
foundation due to differential settlements. In applied for a fixed but critical point in the
this state, no hydrodynamic loads on the cross-section, i.e. the fatigue loads are to be
foundation need to be included, since these applied from one particular direction only,
loads will contribute only marginally to the and this direction is to be taken as the most
overall moment on the tripod. All partial critical direction. The conical transition
load factors are to be set equal to unity, i.e. induces an additional stress, which has to be
characteristic loads are to be used in accounted for. The designer may find it
calculations. Loads from the wind turbine feasible to locate internal ringstiffeners,
shall be taken as the worst damaging load where the cone connects to the tubular, and
case among all dynamic load cases. The thereby lower the overall dimensions of the
maximum allowable tilt under this conical section. Several layouts of the
characteristic loading is 0.5 from vertical. stiffener(s) are possible:
a number of internal ringstiffeners
In the fatigue limit state (FLS), the structure bulkheads
is checked against failure due to fatigue the leg framing may protrude into the
damage. The cumulative fatigue damage for interior of the centre column to be
all load situations during the design life joined at the centre line, thereby
must be taken into account and the distributing the loads directly between
integrated effect should be investigated, e.g. the leg frames
by using the rain-flow counting scheme for The most advantageous option may be
stress cycles. Fatigue loads from the wind chosen, taking into consideration the
turbine shall be considered in conjunction requirements of internal clearance for risers,
with wave-induced fatigue loading, and it is conductors, power cables, etc.
to be verified whether it is acceptable to
disregard contributions to the fatigue Natural frequency analysis
damage from ice crushing. A complete natural frequency analysis shall
be performed for the combined structure
As fatigue is considered critical for the consisting of turbine, tower, tripod and piles.
structure, it is of importance to verify that For this purpose, the non-linear soil must be
the fatigue life does not fall short of the linearised. It is to be verified that the lowest
design life, based on detailed information of frequencies differ from at least 10% of the
waves from different geographical 1P and 3P rotor frequencies at nominal
directions. The areas which are most likely power.
to be dimensioned by fatigue loads are the
centre column transition to the ice cone and Grouted connections
the upper joint at the centre column where Axially loaded pile-to-sleeve connections
the upper leg framing attaches to the struc- are checked using the formula given by
ture. Consider the distribution of stresses DNV (1977), including a safety factor of
due to pure bending of the tower. This will 3.0. The connections can be established by a
8 Foundations 219
Guidelines for Design of Wind Turbines DNV/Ris
220 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
8 Foundations 221
Guidelines for Design of Wind Turbines DNV/Ris
Eide, O., and K.H. Andersen, Foundation Rambll, Havmllefundamenter med sug,
Engineering for Gravity Structures in the UVE-98 J.nr. 51171/97-0047, Fase 1-
Northern North Sea, Norwegian rapport, SEAS, Niras, Rambll, DGI, Ris,
Geotechnical Institute, Publication No. 154, September 1998.
Oslo, Norway, 1984.
Recommendation to Comply with the
Eurocode 2, Design of concrete structures Requirements in the Technical Criteria for
Part 1-1: General basis for buildings and Danish Approval Scheme for Wind Turbines,
civil engineering works, ENV 1992-1- Foundations, Danish Energy Agency,
1:1991 August 1998.
Part 2-2: Reinforced and Prestressed
Concrete Bridges, ENV 1992-2:1996. Reese, L.C. and H. Matlock, Numerical
Analysis of Laterally Loaded Piles,
Hansen, B., Limit Design of Pile Proceedings, Second Structural Division
Foundations, Bygningsstatiske Meddelser, Conference on Electronic Computation,
No. 2, September, 1959. American Society of Civil Engineers,
Pittsburgh, Pennsylvania, pp. 657, 1960.
Hansen, B., Geoteknik og Fundering, Del II
Forelsningsnotater til Kursus 5821 Reese, L.C. and S.-T. Wang, Documentation
Geoteknik 2, Notat nr. 16, Den private of Computer Program Group 4.0, Ensoft,
ingenirfond ved Danmarks tekniske Hj- Inc., Austin, Texas, 1996.
skole, Lyngby, Denmark, 1978.
222 8 Foundations
Guidelines for Design of Wind Turbines DNV/Ris
Soft starter
Wind
Generator Transformer
9.1.1 Generators
Gear Grid
slip rings. In this way, the electrical positive for a motor (nr < ns). If the slip is
characteristics of the rotor can be controlled zero (the generator is in synchronism), the
from the outside by means of electric generator is running idle (it does not
equipment. produce torque). If the slip is 1, the rotor is
blocked. The torque shows a maximum, the
The induction generator has several so-called pull-out torque Tp, at the rotational
advantages such as robustness and mechani- speed np, as shown in Figure 9-2.
cal simplicity. Moreover, as it is produced in
large series, it can be purchased at a Torque
relatively low price.
Motor mode Generator mode
V = 5
10 The maximum voltage Umax and the
2
maximum current Imax are defined as the
peak line-to-line voltage and the peak line
current, respectively.
~ f2 ~ f2
Q G en Current source converter Voltage source convert
Figure 9-5. Reactive power as a function of active
power. Reactive power compensation using a capacitor Figure 9-6. Basic power converter structures for
bank. frequency converter.
There is both a no-load compensation and a Frequency converters will become more and
full compensation. Whereas, the no-load more significant in future control of wind
compensation is only effected when active
turbines. Their most important properties possibility for wind farms to behave as
and implications are highlighted in the active elements in the power system
following: (Srensen P. et al., 2000). A wind farm
with power plant characteristics is able
1. Controllable frequency is the property to:
of the frequency converter which has control the active/reactive power.
the unique possibility to connect a influence positively the stability of
variable speed wind turbine to the grid, the network.
thereby allowing the generator improve the power quality.
frequency to differ from grid frequency.
Some of the implications of the
controllable frequency property are: 9.2 Wind turbine configurations
the rotor behaves as an energy
storage (as the variations in wind In this section the most commonly applied
speeds are absorbed by the rotor generator and power electronic configu-
speed changes). rations in wind turbines are presented.
the loads on the gear and drive train Further details can be found in (Hansen L.H.
can be reduced. et al., 2001a).
the power capture at low wind
Wind turbines can be divided into two main
speed can be improved.
groups, i.e. fixed speed and variable speed
the acoustical noise emission can be
wind turbines.
reduced (as the wind turbines can
have a low rotational speed at low
The fixed speed wind turbine, equipped with
wind speeds).
a generator connected directly to the grid, a
the frequency converter can replace softstarter and a capacitor bank, is the most
the softstarter and the capacitor common type of wind turbine. In order to
bank. increase power production, some of the
the frequency converter is a fixed speed wind turbines are equipped with
practical necessity for the gearless two speed generators instead of with one
wind turbine. generator only.
2. Controllable reactive power is The advantages of the fixed speed wind
another property of the frequency turbine are:
converter, which makes it possible to simplicity.
improve the power quality. Some of the
low price of the electrical system.
implications of the controllable reactive
power property are:
The disadvantages are:
the voltage stability is improved.
limited power quality control.
the flicker level is reduced.
mechanical stress on gear and drive
the frequency converter can replace train.
the capacitor bank.
the frequency converter can be used The variable speed wind turbine is today not
as a local reactive power source. as common as the fixed speed wind turbine.
However an increasing number of turbines
3. Power plant characteristics is an have variable speed rotors. Two general
important property of the frequency
converter because it provides the
configurations of variable speed wind This type of variable speed wind turbine is
turbines exist: used by several large manufacturers. The
1) one where the generator stator is rotor is connected to the grid through a
connected to the grid, while the frequency converter, which is not a full-
rotor frequency is controlled. The scale power converter, since typically only a
speed can typically be varied fraction (up to 70%) of the speed range will
within a limited range only. be utilised.
2) one where the generator stator is
connected to the grid through a
frequency converter. The speed can ~
be varied to an unlimited extent. ~
The advantages of variable speed wind Gear
turbines are:
improved power quality.
reduced mechanical stress.
Figure 9-8. Schematics of a wind turbine equipped with
a doubly-fed induction generator with a converter
The disadvantages are:
connected to the rotor circuit.
losses in power electronics.
price of frequency converter. Full variable speed systems are equipped
with a frequency converter. Two different
Two possible configurations exist for modes of implementation with full variable
variable speed wind turbines with controlled speed can be found.
rotor frequency (limited variable speed):
One is the classical configuration, where a
cage rotor induction generator is connected
to the grid through a full power frequency
converter see Figure 9-9. The range of the
variable speed operation is, in principle,
Gear from zero to the maximum which can be
handled by the wind turbine.
Optislip
Power quality
Synchronous
generator
~
~
Wind Turbines Grid
Figure 9-10. Schematics of a variable speed system
with a synchronous generator.
Voltage
Frequency
Failures
9.3 Power quality and grid
connection
Figure 9-11. Power quality expressed in grid
The term power quality of a wind turbine interference of the wind turbines.
describes the electrical performance of the
turbines electricity generating system in its The main interference of wind turbines with
interaction with the grid. Methods which can the grid is caused by voltage distortions. The
be used to measure and quantify the power frequency of large power systems is usually
quality of wind turbines have earlier on been very stable, and therefore wind turbines do
developed at a national level. However, the not typically influence the frequency.
need for a common frame of reference However, this is not the case for small
across national borders has called for autonomous grids, where wind turbines may
international standardisation work in the cause frequency variations. Wind turbines
field. do not normally cause any interruptions/
failures on a high voltage grid. Wind
A perfect power quality means that the turbines, as they are designed today, are
voltage and current are continuous and disconnected from the grid by the protection
sinusoidal, implying that they have a system, if for example grid failures occur.
constant amplitude and frequency. This aspect can generate serious problems in
the future. For example in Denmark where it
Grid connected wind turbines affect the is envisaged to cover 50% of the overall
power quality of the grid. The wind turbines electricity consumption by wind power in
and the grid are in continuous interaction, as 2030. Thus, grid failures can have severe
illustrated in Figure 9-11. consequences for the stability of the whole
power system, if large wind farms are
Power quality depends on the interaction disconnected from the grid simultaneously.
between the grid and wind turbines. IEC As a consequence hereof, research and
61400-21 defines power quality development are today focused on making
wind farms more robust towards grid faults, distortions see Table 9-1. In switching
for example by using power electronic operation focus is on flicker and voltage
devices. drop generated at the connection of the
generator, and on voltage changes generated
Voltage distortions can be described in at the connection of the capacitor bank.
different time intervals:
steady-state voltage variations are Depending on the grid configuration and on
changes in the r.m.s. value of the the specific type of wind turbine used,
voltage, occurring with a frequency less different power quality problems may arise.
than 0.01 Hz. For example, in the case of a fixed speed
flicker disturbances are voltage wind turbine, natural variations of the wind
fluctuations between 0.01-35 Hz that and the tower shadow will result in
can cause visible variations in domestic fluctuating power, which may cause flicker
lighting. disturbances. In the case of a variable speed
harmonics are voltage fluctuations wind turbine, the presence of the frequency
above 50 Hz, caused by the presence of converter will inject harmonic currents into
power electronics (frequency the grid. The frequency of a large power
converter). system is normally very stable. However, in
transients are random voltage the case of a weak or autonomous grid,
fluctuations caused by e.g. connection where diesel engines are used, wind turbines
of capacitor bank. may cause frequency fluctuations.
every location where electrical systems or In other words, wind farms must develop
equipment are located. power plant characteristics. The two system
responsible utilities in Denmark, Eltra
It is recommended that disconnection of the (Eltra, 2000) and Elkraft System have laid
electrical system from all electric sources, as down requirements for the influence of wind
required in the case of maintenance or farms on grid stability, power quality and on
testing, shall be possible. Semiconductor the control capabilities of wind farms.
devices should not be used as disconnection
devices alone. Where lighting or other Another consequence of the increased size
electrical systems are necessary for reasons of future wind farms is that large wind farms
of safety during maintenance, auxiliary will be connected directly to the
circuits should be provided with separate transmission level. Until now, wind turbines
disconnection devices, such that these and wind farms have been connected to the
circuits may remain energised while all distribution system, typically on 10/20 kV
other circuits are deenergized. It is grid or on 50/60 kV grid. Consequently,
recommended that any electrical system main focus has been on the influence of
operating above 1000 V AC or 1500 V DC wind farms on the power quality in the
shall be able to be earthed for maintenance. distribution system. In Denmark, this has
been regulated by the Danish Utilities
The IEC 61400-1 requires that all electrical Research Institutes (DEFU) requirement for
components and systems shall meet the grid connection of wind turbines (DEFU
requirements set forth in IEC 60204-1, and KR111, 1998) to the distribution system.
compliance with the requirements of IEC However, stricter requirements for large
60364 for design of the electrical system of wind farms, connected directly to the
a wind turbine is also required. transmission system (Eltra, 2000), are now
issued by a.o. transmission system operators
It is recommended to make provisions to in Denmark. As mentioned before, national
enable the wind turbine to be isolated from standards for power quality of wind turbines
the public grid in a safe manner. This have been supplemented by the new
applies to normal situations, abnormal standard IEC 61400-21 (IEC 61400-21,
and/or faulty conditions as well as during 2001) for measurement and assessment of
maintenance and repair. the power quality of grid connected wind
turbines.
2001a). Simulation of the wind farm Hansen L.H., Helle L., Blaabjerg F., Ritchie
interaction with the grid may thus provide E., Munk-Nielsen S., Bindner H., &
quite valuable information and may even Srensen P, Conceptual survey of
lower the overall grid connection costs. In generators and power electronics for wind
addition, these models can be used to study turbines, Ris-R-1205 (EN), 2001.
control strategies for wind farms.
Hansen L.H., Madsen P.H., Blaabjerg F.,
Large research projects have been initiated Christensen H.C., Lindhard U., & Eskildsen
to analyse different control strategies of K, Generators and Power Electronics
large wind farms. There are many control Technology for Wind Turbines, In , IECON '
topologies, and they all have their particular 01, Denver, 2001.
advantages and disadvantages (Hansen L.H.
et al., 2001b). One option is a decentralised Heier S, Grid Integration of Wind Energy
control structure with an internal AC grid Conversion Systems, 1998.
connected to the main grid, where each wind
turbine has its own control system with its IEC 61400-21, Final Draft International
own frequency converter. Such a system has Standard - Wind turbine generator systems -
the advantage of ensuring that each wind Part 21: Measurement and assessment of
turbine can work optimally with respect to power quality characteristics of grid
its local wind conditions. Another option is a connected wind turbines, 2001.
centralised control structure where, for
example, the wind farm is connected to the Mohan N., Undeland T.M., & Robbins
grid via an HVDC connection. Here, the W.P., Power Electronics: converters,
internal behaviour of wind turbines is applications and design, 1989.
separated from the grid behaviour, thus
enabling the wind farm to become Ris and DEFU. Power quality and grid
sufficiently robust to withstand possible connection of wind turbines - summary
failures on the grid. Another way to control report. Ris-R-853 (Summ.)(EN), 1996.
wind farms is the combination of wind
farms and energy storage systems, which Srensen P., Hansen A.D., Janosi L., Bech
makes it possible to buffer some of the J., & Bak-Jensen B., Simulation of
energy. interaction between wind farm and power
system, Ris-R-1281, Ris National
Laboratory, 2001.
REFERENCES
Srensen P., Unnikrishnan A.K., & Sajan
DEFU KR111, Connection of wind turbines A.M. Wind farms connected to weak grids in
to low and medium voltage networks, 1998. India, Wind Energy, 4. 2001.
Eltra, Tilslutningsbetingelser for vindmlle- Srensen P., Bak-Jensen B., Kristiansen? J.,
parker tilsluttet transmissions-nettet, TP98- Hansen A.D., Janosi L., & Bech J., Power
328b, 2000. plant characteristics of wind farms, In
Kassel, 2000.
Hansen A.D., Srensen P., Janosi L., &
Bech J, Wind farm modelling for power
quality, In Denver , IECON ' 01, (2001).
10 Manuals 233
Guidelines for Design of Wind Turbines DNV/Ris
In most cases, the wind turbine parameters The purpose of power performance
must be correlated with the wind conditions measurements is mostly economic as the
at the site and for that purpose, a performance determines the overall
meteorology mast is erected close to the economics of the wind turbine installation.
wind turbine. In some cases, measurements Therefore, power performance measure-
can be made without correlation with the ments are part of contractual matters
wind speed on the site. A meteorology mast, between wind turbine manufacturers and
which is to measure the given wind speed on developers.
the site, should not be erected too close to
the wind turbine. This is due to the fact that When power performance is measured, the
the wind speed sensors are influenced by the smallest uncertainty of the measurement is
wind turbine rotor. Neither should the mast found in flat and non-complex terrain. In
be erected too far from the wind turbine, in most power performance measurement
which case the correlation is decreased procedures, IEC, Danish Energy Agency
substantially. In a flat and non-complex and MEASNET, requirements to the terrain
terrain where the topography causes have been set up, so-called ideal site, that
insignificant flow distortion, a good do not require a site calibration. If these
correlation with the wind velocity at the requirements are not met, a site calibration
wind turbine position is found. In a complex shall be made. A site calibration can be
terrain, the correlation between the wind at made even in flat terrain to reduce
the mast and the wind at the wind turbine uncertainty of the wind speed measurement,
position is poorer. In such cases, a site which shall otherwise be stated as 2-3% due
calibration must be made. Meanwhile, a
234 11 Measurements
Guidelines for Design of Wind Turbines DNV/Ris
Meteorology mast at 4 D
Distance of meteorology
mast to WTGS between
2 D and 4 D; 2,5 D is 2,5 D
recommended
2D
Wind
D
WTGS
Figure 11-2. Requirements in IEC 61400-12 as to distance of the meteorological mast and maximum allowed
measurement sectors.
120
Undisturbed
100
80
60
40
Disturbed
20
2 4 6 8 10 12 14 16 18 20
Relative distance Le/De or Ln/Dn
IEC 149/98
Figure 11-3. Requirements in IEC 61400-12 on the exclusion of particular sectors due to wakes of neighbouring and
operating wind turbines and significant obstacles.
11 Measurements 235
Guidelines for Design of Wind Turbines DNV/Ris
236 11 Measurements
Guidelines for Design of Wind Turbines DNV/Ris
Load measurement procedures are found in more detailed measurement procedure, IEC
the recommendations for compliance with 61400-11.
the Technical Criteria, July 1992, and in IEC
61400-13.
REFERENCES
11 Measurements 237
Guidelines for Design of Wind Turbines DNV/Ris
238 11 Measurements
Guidelines for Design of Wind Turbines DNV/Ris
with melted zinc at 540 in few minutes for special bolts threads may be cut on a
forming a zinc layer of approximate 50- lathe or by hand cutting tools (dies).
60m. The actual lifetime for hot dip
galvanized bolts are dependent on the A widely used approach for lifetime
environmental conditions (pollution in the prediction and strength assessment is the
atmosphere). German guideline VDI-2230. VDI
distinguishes between thread form-rolled
Environment Average before heat-treated and thread form-rolled
lifetime after heat-treated.
[years]
Open agricultural areas 45 In the latter case the thread rolling process
Small towns 30 introduces compression stresses in the
Large towns 12 surface, causing better fatigue properties but
Heavy polluted industrial areas 5 also dependence of the mean-stress in the
Coast-near areas 30 bolt. For finally heat-treated bolts the
Table A-1. Average lifetime for hot dip galvanized following equation will apply
bolts. (Arvid Nilsson).
ASV = 0.85 (150/d + 45)
For materials with hardness > HV 300, hot
dip galvanizing implies a risk of hydrogen 50
embrittlement. Consequently such materials 50
48
shall be heat treated at approx. 200 for 2-4
hours. In general it is not recommended to ASV
(d)
46
hot dip galvanize bolt at higher grade than 44
8.8. Grade 10.9 may be hot dip galvanized 42
with great caution. It is not possible to hot 40 40
dip galvanize materials with higher strength 10 20 30 40 50 60
because of temper effects and zinc 10 d 60
depositions in the grain boundaries causing Figure A-1. Fatigue strength as function of bolt
diameter for finally heat-treated bolts (VDI 2230)
micro cracks in the structure. The (stress amplitudes).
consequences in such cases may be
catastrophic. For finally thread form-rolled bolts the
following equation will apply
A.5 S-N curves ASG = (2 FSm/F0.2) ASV
For design of bolt connections, the single
most important issue is to consider the FSm = (FSao + FSau)/2 + FMzul
proper choice of an S-N curve for fatigue
predictions. This S-N curve is much Valid for: 0.3 < FSm/F0.2 < 1
dependent on how the bolt is manufactured.
Several possibilities exist for this FSm/F0.2 = Rb
manufacturing, including:
thread form-rolled and then heat-treated FSao higher force from external load
material of near nominal diameter fully FSau lower force from external load
heat-treated before form-rolling of FMzul allowable pre-tensional force
thread
80
80 These S-N curves are intended for use at
loads of constant amplitude and are not
ASG16
(Rb) 70 unconditionally valid for loads with varying
ASG24
(Rb) amplitudes (defined as a load spectrum)
ASG36
(Rb)
60 (VDI 2230).
ASG48
(Rb)
50
A.5.1 S-N curves in structural steel
40 40
codes
0.3 0.4 0.5 0.6 0.7 0.8 0.9
0.3 Rb 1 S-N curves in the structural steel codes
DS412 and Eurocode 3 are given for 97.7%
survival probability in the following general
Figure A-2. Fatigue strength of bolts M16 (red), M24 form
(blue), M36 (green) and M48 (brown) as a function of
mean stress ratio FSm/F02 with bolt diameter as
parameter for finally form-rolled threads (stress log10N = log10a mlog10r
amplitudes).
where N is the number of cycles to failure at
These values are valid for grade 8.8-12.9 stress amplitude r, and log10a and m are
bolts subjected to more than 2106 cycles the parameters of the S-N curve. Depending
and correspond to 99% survival probability. on the material quality, heat-treatment and
pre-stressing different S-N curves can be
For a number of cycles lower than 2106 the established.
corresponding stress to failure can be found
from Standard bolts without controlled pre-
stressing
AZSV = ASV(ND/NZ)1/3 Standard bolt means mass produced bolts
according to ISO 898/1-2. For finally heat
AZSG = ASV(ND/NZ)1/6 treated or cut bolts category *36 according
to DS 412 / Eurocode 3 will apply (stress
For finally heat-treated and finally thread amplitudes):
form-rolled ND = 2106 and NZ is the actual
number of load cycles. N < 107: log10a = 11.101, m = 3
N > 107: log10a = 13.385, m = 5
N > 108:
3
1 .10
500 fat = 7.5 MPa
AZSG ( NZ )
N < 107: log10a = 11.551, m = 3
10 N > 107: log10a = 14.585, m = 5
N > 108: fat = 11.5 MPa
1 1
Standard bolts with controlled pre-
1 .10
3
110
3
4
1 .10
5
1 .10 1 .10
NZ
6
1 .10
7
1 .10
8 9
1 .10
110
9 stressing
For class 8.8 and 10.9 standard bolts with
Figure A-3. S-N curve for finally heat-treated (red) and controlled pre-stressing category 71
finally thread form-rolled (blue) (VDI 2230) (stress according to DS 412/Eurocode 3 may be
amplitudes).
applied based on the S-N curves reported in A.5.2 Allowable surface pressure
VDI 2230 (stress amplitudes).
The maximum surface pressure between
mating surfaces shall be evaluated. Be aware
N < 5106: log10a = 11.851, m = 3
that the allowable surface pressure PG is not
N > 5106: log10a = 15.286, m = 5
proportional to the ultimate strength of the
N > 108: fat = 14.5 MPa
material. The maximum bolt force shall, of
3
course, always be adjusted to the material
1.10
with the lower PG in the actual bolt
connection. Figures for some commonly
71( N) 100
used materials are shown in Table A-2.
50( N)
Material Ultimate Allowable
36( N)
10 strength surface
Rm,min stress PG
[MPa] [MPa]
1
3
1.10 1.10
4 5
1.10
6
1.10
7
1.10 1.10
8
1.10
9 St 37-2 340 490
N St 50-2 470 710
Figure A-4. S-N curve (stress amplitudes) for bolts
St 52-3 510 760
without pre-stressing *36 (black) and *50 (blue) and 34CrMo4 1000 870
with pre-stressing 71 (red) According DS412/ 34CrNiMo6 1200 1080
Eurocode 3. X5 CrNi 18 12 500 630
3
GG-25 250 900
1 .10
500
GGG-40 400 700
GGG-50 500 900
GGG-60 600 1000
AZSV( NZ ) 100
AlMgSi 1 F28 260 230
AZSG ( NZ )
Table A-2. Allowable surface pressure PG (VDI 2230)
36( NZ )
71( NZ )
10
A.6 Pretension
1
Depending on the bolt diameter and the
required accuracy, several types of pre-
1
3 4 5 6 7 8 9
1 .10 1 .10 1 .10 1 .10 1 .10 1 .10 1 .10
3 NZ 9
110 110
stressing methods are in use. The most
Figure A-5. Comparison of S-N curves (stress commonly used methods are:
amplitudes) from VDI 2230 (red, blue) and category 71 torque wrench
and *36 from DS412/Eurocode 3 (violet, black). VDI
curves are valid for M36 with a pre-tension of 0.7 times
turning angel method
02 hydraulic tension device
measurement of bolt extension
These S-N curves will apply for M12 M36
untreated (black) bolts. For hot dip In this context only the two first methods
galvanised bolts the figures will be 12-15% will be mentioned since a large majority of
lower. bolts are pre-stressed with these methods.
For selected bolt diameters, surface
treatments and lubrication conditions,
suitable pre-tension and pretension torque,
can be taken from Table A-3 to A-6.
Bolt Area Bolt force at 02 kN] combinations of surface and lubrication can
dia. AS be calculated from:
[mm] [mm2] 8.8 10.9 12.9
M12 84.3 54 76 91 FSm = Fs0 GF
M16 157 100 141 170
M20 245 157 220 265 Mv = MV0 C
M24 353 226 318 381
Scatter on the average pre-tension:
M27 459 294 413 496
M30 561 359 505 606
FS = FSm SF/FSM
M33 694 444 625 750
M36 817 523 735 882 SF is the scatter on average pre-tension force
M42 976 624 878 1054 FSM is the average pre-tension force.
Table A-3. Bolt forces at 02. (Arvid Nilsson)
For an M20 8.8 black oiled bolt the with C =
Bolt Area Pre-tension torque MV0 1.0 and GF = 0.71 and MV0 = 385Nm it
dia. AS [Nm] results in a average pre-stress of 0.71 times
[mm] [mm2] 8.8 10.9 12.9 02. For the same bolt lubricated with wax
M12 84.3 81 114 136 the average pre-stress will arrive at 0.83
M16 157 197 277 333 times 02 with a corresponding pre-tension
M20 245 385 541 649 torque of 0.63 times MV0.
M24 353 665 935 1120
Surface Lubri- SF/FSM GF C
M27 459 961 1350 1620
cation
M30 561 1310 1840 2210
Black Dry 0.29 0.62 0.96
M33 694 1770 2480 2980 Oil 0.16 0.71 1.00
M36 817 2280 3210 3850 MoS2 0.16 0.75 0.86
M42 976 3640 5110 6140 Hot dip Dry 0.29 0.55 1.17
Table A-4. Pre-tension torque for oil lubricated black
galva- Oil 0.16 0.69 1.07
bolts. (Arvid Nilsson)
nised
Bolt Area Pre-tension torque All Wax 0.11 0.83 0.63
dia. AS MV0 [Nm] Stainless Oil 0.29 0.55 0.84
[mm] [mm2] 50 70 80 steel Wax 0.23 0.65 1.00
Table A-6. Modifying coefficients for pre-tension force
M12 84.3 27 57 76 GF and pre-tension torque C as well as scatter at
different surface treatments and lubrication conditions.
M16 157 65 140 187 (Arvid Nilsson)
M20 245 127 273 364
M24 353 220 472 629 For the turning angel method following
M27 459 318 682 909 guidelines (for grade 8.8) can be given in
M30 561 434 930 1240 which d is the bolt diameter:
M33 694 585 1250 1670
M36 817 755 1620 2160
Table A-5. Pre-tension torque for stainless steel bolts
lubricated with wax. (Arvid Nilsson)
Main influence factors to consider for loss A.7 Minimum depth of threaded
of pretension: holes
plastic deformations between bolt
head/nut and contact face For standard bolts with nuts in same strength
number of joint faces class no further calculation of strength of the
For dW DA dW + lK:
Aers = / 4(dW2 dh2) + / 8
dW(DA dW)[(x + 1)2 1]
where
x = ((lKdW)/DA2)1/3
For DA dW + lK:
Aers = Aers for DA= dW + lK
FV,R = c3Afub/m
Figure A-7. Illustration of the force inlet ratio n.
FV,R design carrying capacity in shear
Calculation of more complicated bolt
connections e.g. eccentric loaded connec- c3 = 0.6 for quality class 4.6, 5.6, and 8.8
tions is beyond the scope of this book but with section through thread.
more informations can be taken from e.g. c3 = 0.5 for quality class 5.8, 6.8 and
VDI 2230. 10.9 with section through
thread.
c3 = 0.6 for all quality classes with
A.9 Connections subjected to section through shaft
shear A either the stress area of the
thread or area of bolt shaft
Three categories of shear connections in
fbu characteristic ultimate strength
structural steel constructions are defined in
for the bolt material
DS412/Eurocode 3:
m partial safety factor for material
A. Bearing type
For bolts with finally cut threads c3 shall be
B. Combination of bearing/friction type
multiplied by 0.85.
C. Friction type
Bearing resistance:
Category A:
it shall be shown that the actual shear
Fb,R = 2.5c1c2dtfu / m
force is lower than the shear resistance
of the bolt material incl. the relevant
c1 and c2 can be taken from Table A-9
partial safety factors
d diameter of the bolt
it shall be shown that the actual shear t thickness of the plate
force is lower than the bearing fu characteristic ultimate strength
resistance of the hole inclusive the for the plate material.
relevant partial safety factors. m partial safety factor for material
If movements due to clearance in the holes
1.2d0e1<3.0d0 c1 = e1 /3d0 Low-
are not allowed in the connection (e.g.
2.2d0p1<3.75d0 c1 = p1 /3d0-1/4 est
caused by change in load direction) tight-
1.2d0e2<1.5d0 c2 = e2/0.9d0-2/3 value
fitted bolts should be used. In that case the
2.4d0p2<3.0d0 c1 = p2/1.8d0-2/3
tolerances for bolt/hole shall be h13/H11. Table A-9. Magnitude of reduction factors c1 and c2.
REFERENCES
A.12 Execution of bolt connections
VDI 2230, Systematische Berechnung
The following apply, according to ENV
hochbearspruchter Scraubenverbindungen
1090-1 and Eurocode 3.
October 2001
Straightness of mating surfaces in
Zeit- und Daurfestigkeit von schwarzen und
compression
feurverzinkten hochfesten scrauben,
Must be < 0.5mm between an arbitrary
Bauingenieur 61 (1986)
located straight edge and the surface.
ENV 1090-1, General rules and rules for
Washers
buildings.
For connections in category A washers are
not required.
ENV, Eurocode 3. Design of steel
structures, 1993-1-1.
For connections in category B and C with
class 8.8 bolts tempered washers are
DS 412 Steel Constructions, 1998-07-02
required beneath the turning component.
RCSC, Specification for Structural Joints
For connections in category B and C with
using ASTM A325 or A490 Bolts, 1985.
class 10.9 bolts tempered washers are
required beneath both head and nut.
Bossard Handbuch der Verscraubungs-
technik, Expert Verlag
B.3 Nacelle
B.3.1 Main shaft
A rough estimate of the main shaft diameter
signifies that it amounts to 1 % of the rotor
diameter.
B.4 Noise
Other things being equal, sound pressure
will increase with the fifth power of the
speed of the blade relative to the surround-
ing air.
REFERENCES
N = Ni + Np
Ni initiation time
Np propagation time
Figure C-1. Simulated time series of bending stress in The initiation and the propagation time for
the main shaft during production at 12 m/s. steel components is often expressed by
formulas on the following form, where the defects, which may often be located in areas
latter is known as Paris' power law with stress concentrations, see Section C.5.
Fracture mechanics will mostly apply to
n calculation of the remaining life of detected
K
Ni = B cracks.
C.3 S-N curves
da
= C K m
dN p Fatigue strength data are usually reported as
S-N curves, also known as Whler curves.
The S-N curve is established through
n empirical material constant numerous sample tests with different load
B empirical material constant ranges, resulting in pairs of stress range S
dependant on geometry of stress and a number of stress cycles N to failure at
concentration this stress range. The S-N curve can usually
a crack length be expressed as
C empirical material constant
m empirical material constant
N = KS -m
K stress intensity range
On logarithmic form this results in a linear
The stress intensity range K for cracks relationship between logS and logN
through a thickness is in its general form
written logN = logK m logS
m1
As indicated, a fracture mechanics approach 1
to fatigue requires explicit data of material
behaviour and stress conditions that are SD m2
often difficult to obtain. SE
1
is bilinear which is commonly used for steel. When logN is not known, it can be set equal
For such a curve two different sets of logK to 0.46 for plain steels free from welds, see
and m are given depending on whether N < DNV (1984).
ND or N > ND. The endurance limit SE under
which failure will not occur, is a property The S-N curve valid for analysis of a
specific for ferrous and titanium alloys. structural detail is to be applicable for the
material, the structural detail, the state of
The material will exhibit a natural variability stress considered and the surrounding
of logN about the mean as given by this environment. The S-N curve should take into
curve - often expressed in terms of the account possible material thickness effects.
standard deviation . Usually, the mean
curve is specified in such a way that in 50% Recommended characteristic S-N curves for
of the cases the true value of logN will be steel are described in Section C.6 and C.7,
larger than given by the curve, and in 50% and recommended characteristic S-N curves
of the cases it will be smaller than given by for steel bolts are given in Appendix A. For
the curve. For design of wind turbine determination of S-N curves for cast iron,
structures, it is required to use a reference is made to Hck et al.
characteristic S-N curve, which is to be
chosen in a conservative manner such that For design purposes, fatigue analysis based
logN in a large fraction, 1p, of the cases on S-N curves from fatigue tests is normally
exceeds the value given by the characteristic straightforward and the most suitable
curve. Usually, 1p is taken as 97-98%. method. It is then used in conjunction with
When the mean S-N curve is given, the Palmgren-Miners rule for prediction of
characteristic S-N curve can be found by cumulative damage.
keeping the values of logN fixed and
changing the corresponding S values
according to the following formula C.4 The Palmgren-Miner rule
The fatigue life or in other terms the
log N cumulative damage under varying loads
S1 p = S 50% exp(k )
m can be predicted based on the S-N curve
approach under the assumption of linear
in which S50% is the stress range cumulative damage by Palmgren-Miners
corresponding to the mean S-N curve for rule. The total damage that a structure will
given N, S1p is the corresponding stress experience during its design life may be
range corresponding to the characteristic S- expressed as the cumulative damage from
N curve with 1p probability of exceedance, each load cycle at different stress levels,
logN is the standard deviation of logN independent of the sequence in which the
representing the natural variability in logN stress cycles occur, i.e. no sequence-
for given S, and k depends on p and can be dependency or so-called load cycle effect
taken from Table C-1. is present.
Table C-1. Quantile coefficient k vs. fraction p According to Palmgren-Miners rule, the
Fraction 20 10 5 2.3 2 1 0.1 accumulated damage D can be predicted as
p% follows
K 0.84 1.28 1.65 1.96 2.05 2.33 3.09
k
n( S i ) arrive is important. In such cases, the
D= expression for D by the Palmgren-Miners
i =1 N (S i )
sum may underpredict the true accumulated
damage.
in which n denotes the number of stress
cycles of stress range S in the lifetime of the Note that the Miners sum formulation for
structure, and N is the number of cycles to the cumulative damage used with the quoted
failure at this stress range. n is determined S-N curve model disregards a possible
from the long-term distribution of the stress dependency on the stress ratio R = min/max.
ranges as the one shown in Figure C-2, and This is sufficient for most components in
N is determined from the expression for the wind turbines since they are subjected to
S-N curve; logN = logK m logS. The sum fatigue loading with an approximately zero
is over all stress ranges Si in a sufficiently mean stress m, as is the case for the main
fine discretisation of the stress range space shaft in Figure C-1. However, in some cases
into k blocks of constant range stress cycles. the influence of the mean stress needs to be
taken into account. Different S-N curves for
It is noted that the expression for the different values of m and R can be
accumulated damage D can be interpreted as evaluated or a fatigue life can be assessed
a sum of partial damage owing to load through methods that take the mean stress
cycles at various stress ranges, regardless of directly into account. See Section C.9.
the sequence in which the load cycles occur.
When the long-term stress range distribution C.5 Fatigue in welded structures
is a Weibull distribution with scale
parameter s0 and shape parameter h, Welds in structures usually possess defects
which may form the basis for crack growth
s h and eventually lead to fatigue failure. For
FS ( s ) = 1 exp(( ) ), welded joints involving potential fatigue
s0
cracking from the weld toe, an improvement
in strength by a factor of 2 on the fatigue life
and the total number of stress cycles in the can be obtained by controlled local
design life is ntot, then the cumulative
damage D can be calculated as
n tot m m
D= (1 + ) s 0
K h
machining or grinding of the weld toe. This C.7 Characteristic S-N curves for
can be carried out by means of a rotary burr. forged or rolled steel
The treatment should produce a small
S-N curves for alloyed steel are not
concave profile at the weld toe with the
standardised in the same manner as in the
depth of the depression penetrating into the
case of structural steel. Based on empiric
plate surface at least 0.5 mm below the
relationships between the static material
bottom of any visible undercut. The
strength and the fatigue strength, so-called
undercut should be small enough to allow
synthetic S-N curves are established.
the maximum required grinding depth to
remain within 2 mm or 5% of the plate
For forged and rolled steel subject to fully
thickness, whichever is the smaller.
reversed stress cycles about a zero-mean
Reference is made to Figure C-4.
stress, i.e. the stress ratio R = min/max. =
By grinding the weld profile such that a 1, the following applies:
circular transition between base material and
weld surface is achieved, the stress The fatigue strength limit for rotating
concentration becomes reduced as the bending with 50% survival probability at 106
grinding radius is increased. This may lead cycles can be calculated as
to additional improvement of the welded
joint. D = 1.25(0.436y + 77) f(Ry,B) f(d,r)
The benefit of grinding may be claimed only y yield strength or 0.2% proof stress for
for welded joints, which are adequately the shaft material, related to the actual
protected from seawater corrosion. The dimension. (MPa)
benefit of grinding should usually not be B ultimate strength for the shaft material,
taken into account during design, as related to the actual dimension. (MPa)
grinding is considered as the only
improvement, which can be carried out f(d,r) is a size factor, which depends on the
during fabrication, to increase the fatigue shaft diameter d and the fillet radius r. For
strength. unnotched parts, the size factor is
2
f (d , r ) = 0.8 +
C.6 Characteristic S-N curves for d
structural steel
and for notched shafts it is
S-N curves for welds and base material of
structural steel are classified in DS412 and
Eurocode 3. In these standards, guidance is 0.9 + 1 / r for r 10 mm
f (d , r ) =
given to selection of S-N curves, on the form 1.0 for r < 10 mm
logN = logK m logS for numerous
construction and weld details, to be used in These expressions for the size factor require
conjunction with the Palmgren-Miner rule. d and r to be given in units of mm.
B Ry
f (R y , B ) = 1 log 10 ( )
4000 10
12
m= +3 Figure C-5. Characteristic S-N curve for non-welded,
2 forged or rolled machine steel.
a Su
R=0 Se a
Sem 45
Goodman line m
Gerber line
Se
m 45
Syt Sut Sy
Figure C-6. Different fatigue lines for non-zero mean Figure C-7. Modified Goodman diagram.
stress.
Sy yield strength
a stress amplitude Su ultimate strength
m mean stress Se endurance limit
Syt yield strength in tension
Sut ultimate strength in tension
Sem modified endurance limit, i.e. reduced REFERENCES
for effects of surface, size, reliability
etc. DNV, Fatigue Strength Analysis for Mobile
R = min/ max Offshore Units, Classification Notes No.
30.2, Det Norske Veritas, Hvik, Norway,
The Gerber diagram is a parabola given by 1984.
m
2 DNV, Fatigue Assessment of Ship
a = S em 1
Structures, Classification Notes No. 30.7,
S ut Det Norske Veritas, Hvik, Norway, 1998.
The Goodman line is given by DS 412, Code of Practice for the structural
use of steel, 3rd edition, 1998-07-02.
analyse analogous types of problems define the model and the boundary con-
involving other time-dependent quantities ditions. Hence the coordinate system valid
than temperature. This applies to problems for the elements and boundary conditions
governed by the same differential equation should be checked, e.g. by plots. This is
as the one which governs heat transfer. An particularly important for beam elements
example of such an application can be found given that it is not always logical which axes
in foundation engineering for analysis of the are used to define the sectional properties.
temporal evolution of settlements in
foundation soils. Similarly, as a wrong coordinate system for
symmetry conditions may seriously corrupt
Types of analyses the results, the boundary conditions should
The analyses mentioned above only be checked.
encompass some of the types of analyses
that can be performed by FEM analysis. Insofar as regards laminate elements, the
Other types of analyses are: plastic analyses default coordinate system often constitutes
and analyses including geometric non- an element coordinate system, which may
linearities. have as a consequence that the fibre
directions are distributed randomly across a
Furthermore, combinations of several model.
analyses can be performed. As examples
hereof, the results of an initial frequency Material properties
analysis can be used as a basis for Several different material properties may be
subsequent dynamic analysis. Finally, the used across a model, and plots should be
result of a thermal analysis may be used to checked to verify that the material is
form a load case in a subsequent static distributed correctly.
analysis.
Drawings are often made by means of using
units of mm to obtain appropriate values.
D.2 Modelling When the model is transferred to the FEM
program, the dimensions are maintained. In
The results of a FEM analysis is normally
this case care should be taken in setting the
documented by plots and printouts of
material properties (and loads) correctly, as
selected extreme response values. However,
kg-mm-N-s is not a consistent set of units. It
as the structural FEM model used can be
is advisable to use SI-units (kg-m-N-s).
very complex, it is important also to
document the model itself. Even minor
Material models
deviations from the intention may give
The material model used is usually a model
results that do not reflect reality properly.
for isotropic material, i.e. the same
properties in all directions. Note, however,
D.2.1 Model
that for composite materials an orthotropic
The input for a FEM model must be docu- material model has to be used to reflect the
mented thoroughly by relevant printouts and different material properties in the different
plots. The printed data should preferably be directions. For this model, material
stored or supplied as files on a CD-ROM properties are defined for three orthogonal
directions. By definition of this material, the
Coordinate systems choice of coordinate system for the elements
Different coordinate systems may be used to has to be made carefully.
programs can set up such constraints Furthermore, the shape and order of the
automatically. elements influence the required number of
elements. Triangular elements are more stiff
Element size/distribution than quadrilateral elements, and first-order
The size, number and distribution of elements are more stiff than second-order
elements required in an actual FEM model elements.
depend on the type of analysis to be
performed and on the type of elements used. This can be seen from the example below, in
which a cantilever is modelled by beam,
Generally, as beam and shell elements have membrane, shell and solid elements.
five or six degrees of freedom in each node,
good results can be obtained with a small E = 2.1105 N/mm2 100 N
number of elements. As solid elements only
have three degrees of freedom in each node, 10 mm
The aspect ratio is the ratio between the side include element models of structural parts
lengths of the element. This should ideally other than the particular one to be
be equal to 1, but aspect ratios of up to 3-5 investigated. One situation where this comes
do usually not influence the results and are about is when the true supports of a
thus acceptable. considered structure have stiffness pro-
perties which cannot be well-defined unless
Element warping is the term used for non- they are modelled by means of elements that
flatness or twist of the elements. Even a are included in the FEM model.
slight warping of the elements may
influence the results significantly. When such an extended FEM model is
adopted, deviations from the true stiffness at
Most available FEM programs can perform the boundary of the structural part in
checks of the element quality, and they may question may then become minor only. As a
even try to improve the element quality by consequence hereof, the non-realistic effects
redistribution of the nodes. due to inadequately modelled boundary
conditions are transferred further away to
The quality of the elements should always the neighbouring structural parts or sub-
be checked for an automatically generated parts, which are now represented by
mesh, in particular, for the internal nodes elements in the extended FEM model.
and elements. It is usually possible to
generate good quality elements for a Types of restraints
manually generated mesh. The types of restraints normally used are
constrained or free displacements/rotations
With regard to automatically generated or supporting springs. Other types of
high-order elements, care should be taken to restraints may be a fixed non-zero displace-
check that the nodes on the element sides are ment or rotation or a so-called contact, i.e.
placed on the surface of the model and not the displacement is restrained in one
just on the linear connection between the direction but not in the opposite direction.
corner nodes. This problem often arises
when linear elements are used in the initial The way that a FEM program handles the
calculations, and the elements are then fixed boundary condition may vary from one
changed into higher-order elements for a program to another. One approach is to
final calculation. remove the actual degree of freedom from
the model, another is to apply a spring with
Benchmark tests to check the element a large stiffness at the actual degree of
quality for different element distributions freedom. The latter approach may lead to
and load cases are given by NAFEMS. singularities if the stiffness of the spring is
These tests include beam, shell and solid much larger than the stiffness of the element
elements, as well as static and dynamic model. Evidently, the stiffness can also be
loads. too small, which may again result in
singularities.
D.2.3 Boundary conditions
An appropriate value for the stiffness of
Definition of boundary conditions
such a stiff spring may be approximately 106
The boundary conditions applied to the
times the largest stiffness of the model.
model should as a matter of course be as
realistic as possible. This may require that
the FEM model becomes extended to
As the program must first identify whether Structural loads consist of nodal forces and
the displacement has to be constrained or moments and of surface pressure. Nodal
free, the contact boundary condition requires forces and moments are easily applied, but
a non-linear calculation. may result in unrealistic results locally. This
is due to the fact that no true loads act in a
Symmetry/antimetry single point. Thus, application of loads as
Other types of boundary conditions are pressure loads will in most cases form the
symmetric and antimetric conditions, which most realistic way of load application.
may be applied if the model and the loads
possess some kind of symmetry. Taking Load application
such symmetry into account may reduce the The loading normally consists of several
size of the FEM model significantly. load components, and all of these com-
ponents may be applied at the same time. As
The two types of symmetry that are most a slightly different load combination in a
frequently used are planar and rotational new analysis will require an entirely new
symmetries. The boundary conditions for calculation, this is, however, not very
these types of symmetry can normally be rational.
defined in an easy manner in most FEM Instead, each of the load components should
programs by using appropriate coordinate be applied separately as a single load case,
systems. and the results found from each of the
corresponding analyses should then be
The loads for a symmetric model may be a combined. In this way, a large range of load
combination of a symmetric and an combinations can be considered. To
antimetric load. This can be considered by facilitate this procedure, unit loads should be
calculating the response from the symmetric used in the single load cases, and the actual
loads for a model with symmetric boundary loads should then be used in the linear
conditions, and adding the response from the combinations.
antimetric loads for a model with antimetric
boundary conditions. As only one or more parts of the total
structure is modelled, care should be taken
If both model and loads have rotational to apply the loads as they are experienced by
symmetry, a sectional model is sufficient for the actual part. To facilitate such load
calculating the response. application, dummy elements may be
added, i.e. elements with a stiffness
Some FEM programs offer the possibility to representative of the parts which are not
calculate the response of a model with modelled these are often beam elements.
rotational symmetry by a sectional model, The loads can then be applied at the
even if the load is not rotational-symmetric, geometrically correct points and be
as the program can model the load in terms transferred via the beam elements to the
of Fourier series. structural part being considered.
D.2.4 Loads
D.3 Documentation
The loads applied for the FEM calculation
are usually structural loads, however, D.3.1 Model
centrifugal and temperature loads are also
The result of a FEM analysis can be
relevant.
documented by a large number of plots and
procedure may in some cases also result in For models with rotational symmetry a plot
too dense meshes. Instead, an indication of of the deflection relative to a polar
whether the model or mesh is sufficient coordinate system may be more relevant for
would be preferable. evaluation of the results.
Load combinations
If the results are found as linear
combinations of the result from single load
cases, the load combination factors should
of course be clearly stated.
Displacement
The global deflection of the structure should
be plotted with appropriately scaled de-
flections. For further evaluation, deflection
components could be plotted as contour
plots to see the absolute deflections.
Actuator: Device that can be controlled to the total kinetic energy contained by the
apply a constant or varying force and wind at a given wind speed within a given
displacement. capture area. (Established by the German
Anemometer distance constant: Quantity scientist A. Betz).
related to the response time of an Bin: Interval for test data grouping.
anemometer. Blade Angle: Angle between the plane of
Angle of attack: Angle between the the rotor and the local chord line of the
direction of the resulting wind velocity on blade profile
the blade and the chord line of the blade Blade cone angle: Angle between the rotor
airfoil section. plane and the blade quarter chord line.
Annual average: Mean value of a set of Blade load distribution: Description of the
measured data of sufficient size and duration manner in which blade loads are distributed
to serve as an estimate of the expected value along the quarter chord line.
of the quantity. The averaging time interval Blade root: Part of the rotor blade that is
shall comprise a whole number of years to closest to the hub.
average out non-stationary effects such as Buckling: Failure mode characterised by a
seasonality. non-linear increase in deflection with a
Augmenter: Device or structure which change in compressive load.
increases an air flow's speed by reducing the Calibration load: Forces and moments
area through which it passes. applied during calibration.
Autonomous system: Wind turbine in Capacity factor: Ratio of the average power
combination with at least one other source produced by a wind turbine to its rated
of energy (e.g. diesel generator, solar power. (Usually taken over a period of time
collector, biogas system, etc.) commonly - daily, weekly, monthly, annually - used as
used to improve the continuity of supply. a prefix).
Auto-reclosing cycle: Event with a time Capacity: Total rated power of a wind
period, varying from approximately 0.01 turbine or group of wind turbines. In the
second to a few seconds, during which a case of electricity producing turbines, it
breaker, released after a grid fault, is denotes the rated generator power.
automatically reclosed and the line is Capture matrix: Organisation of the
reconnected to the network. measured time series according to mean
Availability: Percentage of time that a wind wind speeds and turbulence intensities.
turbine generator is available for production. Catastrophic failure: Disintegration or
Axial: Direction or motion perpendicular to collapse of a component or structure.
the rotor plane. Characteristic value (of a material
Betz limit: Maximum energy conversion property): Value having a prescribed
efficiency theoretically obtainable from a probability of not being attained in a
wind turbine rotor, equal to 16/27 0.593 of hypothetical unlimited test series.
Chord line: Straight line connecting an
airfoil's leading and trailing edges. amplitude and mean value during a fatigue
Complex terrain: Surrounding terrain that test.
features significant variations in topography Co-ordinate system: Co-ordinate systems as
and terrain obstacles that may cause flow defined in DS 472 and IEC 61400 are shown
distortion. in Figure F-1 and Figure F-2, respectively.
Cone angle: See Blade Cone Angle.
Constant amplitude loading: Denotes the
application of load cycles with a constant
zb zb
yb xb
xb
yb
zh zh
yh xh xh
yh
zt zt
yt xt
Electric power network: Particular Guy (wire): Cable or wire used as a tension
installations, substations, lines or cables for support between the ground and a tower.
the transmission and distribution of electri- Horizontal axis wind turbine (HAWT):
city. Wind turbine whose rotor axis is
Electromagnetic interference: Impairment substantially parallel to the wind flow.
of an electromagnetic signal, either by Hub height: Height of the centre of the
electrical field interference or by deflection rotor above the terrain surface.
of a radio transmitted signal. Idling: Condition of a wind turbine
Environmental conditions: Characteristics generator that is rotating slowly and not
of the environment (altitude, temperature, producing power.
humidity, etc.) which may affect the WTGS Inboard: Towards the blade root.
behaviour. Inertial sub-range: Frequency interval of
External conditions: Factors affecting the wind turbulence spectrum, where eddies
operation of a wind turbine, including wind - after attaining isotropy - undergo
regime, electrical grid, temperature, snow, successive break-up with negligible energy
ice, etc. dissipation.
Fatigue strength: Measure of the load- NOTE - at a typical 10 m/s wind speed, the
bearing capacity of a material or structural inertial sub-range is roughly from 0.02 Hz
element subjected to repetitive loading. to 2 kHz.
Fatigue test: Test in which a cyclic load of Infinite life design: Design where the
constant or varying amplitude is applied to operating stresses does not result in damage.
the test specimen. In-plane: Direction or motion parallel with
Feathering: To change the blade pitch the rotor plane.
angle of all or part of a blade to reduce the Kite anemometer: Kite which has been
aerodynamic lift. calibrated to give quantitative wind speed
Flap: Direction which is perpendicular to data.
the swept surface of the non-deformed rotor Leading edge: Part of a blade at the
blade axis. incidence of the air flow.
Flapping: Blade motion in the out-of-plane Lead-lag: Blade motion in the plane of
of rotation direction. rotation (HAWT).
Flapwise: Direction that is perpendicular to Loads: Typical term for the different load
the surface swept by the non-deformed rotor components with reference to Figure F-3:
blade axis. Blade loads
Flatwise: Direction that is perpendicular to Term refer to directions in and
the local chord, and spanwise blade axis. perpendicular to the rotor plane (while the
Freewheeling: Free rotation of the wind terms in parenthesis refer to directions along
turbine with the generator disconnected. and perpendicular to the local blade chord).
Furling: To adjust the wind turbine Mbe: Leadwise (edgewise) bending
structure in order to reduce the driving force moment
of the wind, for instance feathering - yawing Mbf: Flapwise (flatwise) bending moment
out of the wind. Mbt: Torsional moment
Gamma function: The Euler gamma Fbe: Leadwise (edgewise) shear force
Fbf: Flapwise (flatwise) shear force
function is given by ( x) =
0
e u u x 1 du Fbs: Spanwise (axial) force
Hub loads
Grid: Network for transmission and Mhr: Rotor torque/roll
distribution of electricity. Mht: Tilt moment
Strain: Ratio of the elongation (or shear System which converts kinetic energy in the
displacement) of a material subjected to wind into electric power.
stress, to the original length of the material. Wind turbine heating system (WTHS):
Support structure: Part of a wind turbine Wind turbine where the output energy is in
comprising the tower and foundation. the form of heat.
Survival wind speed: Maximum wind speed Wind turbine pump (WTP): Wind turbine
(normally a 3-second gust) that a WTG has directly coupled to a pump.
been designed to sustain without damage to Wind vane: Device for indicating or
structural components or loss of ability to recording wind direction.
function normally. Wind velocity: Vector describing the speed
Swept area: Projected area perpendicular to and the direction of the wind.
the wind direction that a rotor will describe Wind-diesel system: See Autonomous
during one complete rotation. System.
Tare loads: Forces and moments created by Yaw rate: Rate of change nacelle yaw
gravity. position (usually measured in /sec.).
Tip-speed: Linear speed of a blade tip.
Vtip = r (rotor radiusangular speed)
Tower shadow: Disturbance of the flow REFERENCES
field around, and created by, the tower.
Trailing edge: Aft portion of a blade, DS472, Last og sikkerhed for
normally pointed. vindmllekonstruktioner, Dansk Ingenir-
Transient event: Event during which the forening, 1st edition, Copenhagen, Den-
state of operation of the wind turbine mark, 1992.
changes, such as during shut-down or gust.
Twist: Spanwise variation in angle of the Elliot, George, Recommended practices for
chord lines of blade cross-sections. wind turbine testing, 8. Glossary of terms,
Ultimate strength: Measure of the National Wind turbine Centre, Dept. of
maximum (static) load-bearing capacity of a Trade & Industry, National Engineering
material or structural element. Laboratory, Glasgow, UK, 2nd edition 1993.
Unscheduled maintenance: Maintenance
carried out, not in accordance with an IEC, Wind turbine generator systems Part
established time schedule, but after 1: Safety requirements, International
reception of an indication regarding the state Standard, IEC61400-1, 2nd edition, 1999.
of an item.
Upwind: Denotes the direction opposite to IEC, Measurement of mechanical loads, TS
the main wind direction. 61400-13, 1st edition, 2001-06.
Utility grid: Electrical distribution system
for public supply. IEC, Full-scale structural testing of rotor
Utility interconnection: Electrical blades, TS 61400-23, 1st edition 2001-04.
connection between a wind turbine
generator system and a utility grid in which
energy can be transferred from the WTGS
to the utility grid and vice versa.
Variable amplitude loading: Application of
load cycles of non-constant mean, and/or
cyclic range.
Wind turbine generator system (WTGS):
2
u0 = E [U 10 ]
u = u0 x ,
Index
Index 279
Guidelines for Design of Wind Turbines DNV/Ris
280 Index
Guidelines for Design of Wind Turbines DNV/Ris
Index 281
Guidelines for Design of Wind Turbines DNV/Ris
Generator Inertia
asynchronous .... 3, 152, 154, 155, 224 coefficient ................................. 95, 96
induction ....................................... 223 loads ................................... 58, 96, 98
multiple-poled ............................... 157 Inflow angle .........64, 65, 66, 71, 76, 166
multipole ....................................... 228 In-rush current .................................. 225
overload .................. 13, 115, 155, 157 Inspection ............................. 11, 13, 176
reactive power ....... 224, 225, 226, 227 intervals .................................. 26, 176
size ....................................................5 Installation 28, 29, 55, 142, 175, 194,
slip ........................................ 152, 224 199, 209, 212, 213, 220, 233
synchronous 152, 155, 157, 223, 224, Insulation .................... 28, 153, 156, 157
229 Integral temperature criterion ........... 140
variable speed 6, 69, 154, 181, 227, Interference factor .. See Induction factor
228, 229, 230
Geological study ............................... 187 JONSWAP spectrum .......................... 93
Geometrical size effect ............. 121, 122
Geophysical survey ........................... 187 Kaimal spectrum .............. 38, 39, 62, 88
Geostrophic wind speed ................ 40, 44
Geotechnical analysis ....... 187, 188, 218 Labour safety ................ 28, 29, 180, 233
Gerber diagram ................................. 258 Labyrinth seals ......................... 130, 131
Germanischer Lloyd ...........................30 Lift
Glauerts correction ............................67 coefficients ......64, 65, 66, 71, 76, 158
Goodman diagram ............................ 258 curve ................................... 65, 70, 75
Gravity loads ........... 58, 87, 90, 112, 117 forces .......................... 60, 63, 64, 104
Grid connection .... 13, 80, 223, 229, 231 Light conditions .................................. 29
Grid loss ............................................ 148 Lightning .......28, 53, 106, 113, 114, 230
Grinding ............................................ 255 Limit states 11, 18, 19, 23, 24, 27, 56,
Gumbel distribution 23, 44, 83, 85, 86, 117, 145, 210, 217, 218, 248
89 Load
Gust ................. 43, 55, 57, 87, 152, 174 distribution ......................... 24, 25, 89
Gyroscopic effects ............ 3, 58, 59, 161 verification ............... 93, 96, 170, 171
Load combination
Hail ...................................................51 extreme ........11, 20, 99, 174, 260, 267
Harris spectrum ............................. 37, 38 fatigue ..................81, 90, 91, 120, 174
HawC ............................................ 61, 70 Logarithmic decrement ................. 72, 73
High-cycle fatigue ..............................77 Longuet-Higgins distribution ............. 95
Homogeneous terrain .. 33, 35, 37, 39, 62 Low-cycle fatigue ....................... 77, 126
Humidity ....................... 41, 50, 110, 153 Lubrication 27, 130, 136, 137, 141, 165,
Hurricanes ...........................................45 233
grease ............................ 130, 131, 166
Ice ..................................... 33, 50, 214 oil ................................ 132, 137, 141
cone ............................... 210, 216, 219 relubrication interval .... 131, 132, 165
loads .... 50, 60, 99, 210, 214, 216, 218 viscosity ratio ....................... 136, 137
Ideal site ............................................ 234
Idling . 13, 15, 55, 56, 57, 79, 80, 81, 155 Main shaft ....87, 120, 130, 133, 251, 257
Induction factor ...................................66 Mann model .................................. 62, 63
Induction generator ........................... 223 Manual operation ........................ 56, 159
Manuals ...................................... 29, 233
282 Index
Guidelines for Design of Wind Turbines DNV/Ris
Material certificates .................... 11, 143 Palmgren-Miner rule 24, 77, 125, 176,
Material properties ............................ 268 177, 207, 211, 254, 255, 256
Mean stress 78, 79, 81, 110, 112, 117, Parametrised load spectra ................... 89
123, 124, 241, 255, 256, 258 Parks See Wind farms
Measurements ............................. 49, 234 Partial safety factors 11, 20, 21, 22, 23,
damping .................................... 72, 73 26, 101
full-scale test ...................................26 calibration ........................... 20, 21, 22
load ....................................... 170, 236 Passive brakes .................................. 148
model verification ............. 62, 70, 171 Peak counting method .................. 77, 78
noise .............................................. 237 Peak-over-threshold method ............... 89
power performance ................... 7, 234 Periodical maximum method .............. 89
power quality ................................ 237 Personal safety ............See Labour safety
uncertainty ........................ 27, 46, 234 Pile foundation 187, 189, 193, 203, 205,
wind speed ........................ 37, 46, 234 209
Meteorology mast ............................. 234 Pile refusal ........................................ 213
Micro pitting ............................ See Gear Pile-driving 194, 210, 211, 213, 215, 220
Middelgrunden ....................................98 Pile-soil-pile interaction ........... 194, 198
Miner .....................See Palmgren-Miner Pitch angle .............................. 4, 69, 104
Minimal cut set analysis .....................17 Pitch control .........See Power regulation
Modal analysis ...................... 61, 70, 181 Planetary gears ................................. 134
Monin-Obukhov length ......................41 Polyamide (PA) ................................ 165
Monitoring 4, 12, 13, 15, 132, 141, 150, Polyethylene Terephtalate (PET) ..... 165
220 Polyurethane (PUR) ......................... 165
Morisons equation .............................95 Potential annual energy ........................ 7
Power
NACA profiles ....................................71 coefficient ................................. 6, 250
Natural frequency 73, 88, 91, 113, 115, electrical ........................................... 6
171, 173, 204, 218, 219, 260 iso-power curve ................................ 4
Near-coastal locations .........................33 mechanical ...................................... 69
Network connection .. 12, 13, 55, 57, 226 nominal ................................... 12, 155
Nitrided gears ................................... 145 performance .......................... 5, 7, 234
Noise ......... 1, 3, 130, 158, 227, 237, 251 curve ................................................. 6
Nominal power ...................... See Power quality ........................... 229, 232, 237
Non destructive testing ..... 118, 127, 214 rated .......................................... 4, 152
Notch sensitivity factor ............. 122, 123 Power regulation ........................ 3, 6, 69
active stall ......................................... 4
Offshore turbines 8, 33, 50, 51, 179, 187, pitch ...................................... 4, 6, 155
208 stall ................................. 4, 6, 63, 152
Operation range ..................................13 Power spectral density 38, 39, 40, 75, 93
Operation temperature ................ 48, 120 Prandtls tip loss factor ....................... 67
Operational conditions ............ 55, 60, 79 Probabilistic method 22, 24, 26, 28, 100,
Operational load ..... See Functional load 181
Operational procedures ............... 29, 233 Protection system 10, 11, 15, 149, 150,
Outgassing ........................................ 108 153
Overconsolidation ratio ............ 197, 202 P-y curves ................. 197, 198, 199, 200
Overloading ...................................... 155
Overspeed ............. 12, 13, 155, 157, 237
Index 283
Guidelines for Design of Wind Turbines DNV/Ris
284 Index
Guidelines for Design of Wind Turbines DNV/Ris
Index 285
Guidelines for Design of Wind Turbines DNV/Ris
Yaw
active ........................... 3, 86, 159, 161
brake ............................... 29, 146, 162
error 12, 56, 71, 81, 112, 158, 159,
160, 161, 162, 166
loads ........................................ 60, 160
passive .............................. 3, 158, 161
system ......... 2, 59, 146, 157, 158, 233
Yield strength 23, 122, 124, 125, 146,
239, 256, 258, 268, 269
Youngs modulus ...... 138, 165, 203, 204
286 Index