Burton T. (et. al.) Wind energy Handbook

Подождите немного. Документ загружается.

was, of course, at the expense of signiﬁcant negative phase sequence currents

ﬂowing in the generators and associated heating and losses, although in this wind

farm nuisance tripping was not reported. The wind farm slightly reduced the

voltage ﬂicker measured at the point of connection using a ﬂickermeter. This was a

complex effect with the generators raising the short-circuit level but also introdu-

cing ﬂuctuations in current. There was a slight increase in total harmonic voltage

distortion with the wind farm in operation, mainly caused by a rise in the ﬁfth and

seventh harmonics when the low-speed, high-impedance winding of each generator

was in service. This increase was probably associated with a parallel resonance of

the high-impedance winding of the generators and the power factor correction

capacitors and the high levels of ambient harmonics on the utility network. The

most signiﬁcant transient event was the energization of the entire wind farm

including 24 turbine transformers, but without the wind turbine s connected. This

led to a transient 8 percent dip on the 33 kV voltage. Energization of the wind farm

in this manner is rare and occurs only after disconnection by the grid interface

protection.

Davidson identiﬁed three main sources of power ﬂuctuation: wind turbulence,

blade passing, and blade pitching. The major effects were found to be at wind

speeds of 9–10 m=s, mainly due to wind turbulence, and above 20 m=s due to

blade-pitching effects, although the pitch control on these turbines was not com-

pletely effective in high wind speeds. The highest wind speeds (. 20 m=s) led to

the greatest ﬂicker. As might be expected siting turbines in turbulent locations

signiﬁcantly increased power ﬂuctuations. Individual wind turbine starts/stops

and speed changes made little impact on network voltage ﬂicker.

These results are typical of the experience of connecting wind farms, with a large

number of relatively small induction generators, on to rural distribution circuits. In

a number of studies some aspects of power quality have been shown to be

improved by the effective increase in short-circuit level. However, the effect of

embedded generation plant on distribution network will vary according to circum-

stances and each project must be evaluated individually. In particular, the connec-

tion of large single generators, as opposed to a group of smaller generators, may be

constrained by consideration of ﬂicker.

Craig (1995) undertook a similar investigation on another WEG MS3-300 wind

turbine connected to an 11 kV network in Northern Ireland over a 2 month period.

A particular feature of this study was the very low network short-circuit level

which varied between 8–15 MVA on the 11 kV point of connection depending on

how the 11 kV utility circuits were arranged. The X=R ratio of the source imped-

ance was approximately 0.5 illustrating the highly resistive nature of the long 11 kV

overhead circuits. The site had a mean wind speed of 9 m=s and a turbulence

intensity of 12.5 percent. Measurements were taken to investigate the voltage

disturbances due to wind turbine starts (with and without an anti-parallel thyristor

soft-start), steady-state voltage variations, dynamic voltage variations and ﬂicker

and also volta ge unbalance.

The conclusions of the study were that, under certain circumstances (particularly

high wind speed cut-out and connection without the soft-start) the dynamic voltage

variations excee ded recommended limits. Steady-state voltage variations could be

predicted reliably from load-ﬂow calculations and were within limits. The wind

POWER QUALITY 585

turbine made some contribution to voltage ﬂicker but this was masked during

working hours by a local quarry which used stone-cutting machinery that created

ﬂicker well in excess of the recommended levels. The background voltage unba-

lance on the network exceeded 1.5 percent at times and this was reduced when the

turbine was conn ected and drew negative phase sequence (unbalanced) current.

10.5.1 Voltage ﬂicker

Voltage ﬂicker describes dynamic variations in the network voltage which may be

caused either by wind turbines or by varying loads (Bossanyi, Saad-Saoud and

Jenkins, 1998). The origin of the term is the effect of the voltage ﬂuctuations on the

brightness of incandescent lights and the subsequent annoyance to customers

(Mirra, 1988). Human sensitivity to variations of light intensity is frequency

dependent and Figure 10.14 indicates the magnitude of sinusoidal voltage changes

which laboratory tests have shown are likely to be perceptible to observers. It may

be seen that the eye is most sensitive to voltage variations around 10 Hz. The

various national and international standards for ﬂicker on networks are based on

curves of this type. The use of voltage ﬂicker as an indicator of accep table dynamic

voltage variations on the network is rather unusual as the assessment is based on

the experimentally measured effect of changes of incandescent lamps on the human

eye and brain. It is, of course, possible to change incandescent bulbs to other types

of lamp with a different response and the use of ﬂicker as a measure of disturbance

does not consider explicitly the effect of voltage variations on sensitive equipment,

e.g., computers. However, the ﬂicker standards are generally used to characterize

transient voltage variations and are of considerable signiﬁcance for embedded wind

generation which: (1) often uses relatively large individual items of plant, (2) may

0.1

100 1000 10000

Voltage changes per min

∆V/V %

Figure 10.14 Inﬂuence of Frequency on the Perceptibility of Sinusoidal Voltage Changes

586

ELECTRICAL SYSTEMS

start and stop frequently, and (3) is subject to continuous variations in input power

from a ﬂuctuating energy source.

Flicker is usually evaluated over a 10 min period to give a ‘short-term severity

value’ P

. The P

value is obtained from a 10 min time series of measured network

voltage using an algorithm based on the nu isance perceived by the human eye in

ﬂuctuating light. This is illustrated in Figure 10.15 wh ich indicates how ﬂicker is

measured. P

is linear with respect to the magnitudes of the voltage change but, of

course, includes the frequency dependency indicated in Figure 10.14. Twelve P

values may then be combined usin g a root of the sum of the cubes calculation to

give a ‘long-term severity value’ P

over a 2 h period (Electricity Association, 1989)

although IEC 614000-21 makes no distinction between P

and P

limits (IEC,

2000b).

If a number of wind turbines are subject to uncorrelated variations in torque then

their power outputs and effect on network ﬂicker will reduce as

˜P

ﬃﬃﬃﬃ

˜ p

where n is the number of generators, P and p are the rated power of the wind farm

and wind turbine respectively and ˜ P and ˜ p are the magnitude of their power

ﬂuctuation.

There is some evidence (Santjer and Gerdes, 1994) that on some sites wind

turbines can fall into synchronized operation and in this case the voltage variations

become cumulative and so increase the ﬂicker in a linear manner. The cause of this

synchronous operation is not completely clear but it is thought to be due to

interactions on the electrical system caused by variations in network voltage.

However, experience on most UK sites, where the winds are rather turbulent, has

been that any synchronized operation lasts only a short while before being

disrupted by wind- speed changes. This may not be the case offshore when the

turbulence of the wind is likely to be lower.

A range of permissible limits for ﬂicker on distribution networks is given in

national and international standards. Engineering Recommendation P28 (Electricity

Association, 1989) speciﬁes an absolute maximum value of P

on a network, from

all sources, to be 1.0 with a 2 h P

value of 0.6. However, extreme caution is advised

if these limits are approached as the risk of comp laints increases at between the

sixth and eighth power of the change in voltage magnitude once the limits are

reached and the approximate assessment method proposed in the same document

is based on P

not exceedin g 0.5. British Standard (1995), which speciﬁcally

excludes embedded generation from its scope, is signiﬁcantly less stringent specify-

ing that over a 1 week period P

must be less than 1 for 95 percent of the time.

Gardner (1996) describes P

limits from a number of utilities in the range 0.25–0.5.

Input voltage

processing

Lamp model Eye model Brain model

Statistical

evaluation

and P

Figure 10.15 Principle of Flicker Measurement

POWER QUALITY 587

10.5.2 Harmonics

Only variable-speed wind turbines inject signiﬁcant harmonic currents into the

network. Fixed-speed wind turbines, particularly those with power-factor correc-

tion capacitors, alter the harmonic impedance of the distribution network and, in

some circumstance, create resonan t circuits. This may be important if ﬁxed- and

variable-speed wind turbines are installed in the sa me wind farm. It is noted in IEC

(2000b) that harmonic currents have been reported from a few installations of ﬁxed-

speed, induction-generator wind turbines but there is no known instance of cus-

tomer annoyance or equipment damage due to harmonic currents from ﬁxed-speed

wind turbines.

Thyristor soft starts (Figure 10.16) are commonly used to connect the induction

generators used on ﬁxed-speed wind turbines. Their mode of operation is initially

for the thyristors to be ﬁred late in the voltage cycle and then the ﬁring angle

advanced (over several seconds) until the entire voltage wave is applied to the

generator. Thus the netw ork voltage is applied gradually to the generator and the

current drawn by the generators controlled to reduce any voltage variations on

the network. When operating, these devices produce varying harmonics as the

ﬁring angle of the thyristors is altered. Generally the soft-start units are only used

for a few seconds during the connection of the induction generator and for this

short period the effect of the harmonics is considered to be harmless and may be

ignored (IEC, 2000b). If the anti-parallel thyristors are not by-passed, but left in

service, then their harmonic currents need to be assessed. This continuous use of

the soft-start unit has been proposed in order to reduce the applied voltage and

hence the iron losses in the induction generator at times of low generator output.

However, as well as the complication of trying to deal with harmonics which

change with the thyristor ﬁring angle there are also potential difﬁculties with the

variation of the applied voltage to the generator altering the dynamic characteristics

of the drive train.

Modern variable-speed wind turbines use network side voltage source conver-

ters, normally based on the Graetz bridge topology (Figure 10.17). These use

insulated gate bipolar transistors (IGBTs) switching at several kHz to synthesize a

sine wave and so eliminate the lower order (e.g., , 19th) harmo nics. The switching

techniques used can vary from rather simple carrier modulated techniques which

compare a reference signal with a trigger signal to those based on space-vector

Soft-start unit

By-pass contactor

Figure 10.16 Soft-start Unit for an Induction Generator (One-phase only shown)

588

ELECTRICAL SYSTEMS

theory with some form of two-axis transformation (Jones and Smith, 1993). The

rapid switching gives a large reduction in low-order harmonic currents compared

to the classical line commutated converter but the output current will have signiﬁ-

cant energy at around the switching frequency of the devices (i.e., in the 2–6 kHz

region). Although these high-frequency currents are relatively easy to ﬁlter they

may inﬂuence the selection of the size of the coupling reactance X

which can be

used to form part of the ﬁlter to block harmonic cu rrents around the switching

frequency.

10.5.3. Measurement and assessment of power quality characteristics

of grid connected wind turbines

Determination of the power quality of wind turbines and prediction of their

performance in service is not straightforward and IEC 614200-21 (IEC, 2000b) has

been written to provide guidance. There are a number of difﬁculties when assessi ng

the power quality of wind turbines as their performance will depend on:

• the design of the entire wind turbine (including the aerodynamic rotor and

control system),

• the conditions of the electri cal network to which it is connected, and

• the wind conditions in which it operates.

For example, simple measurement of voltage variations at the terminals of a test

turbine is not satisfactory as ambient levels of ﬂicker in the electrical network and

the X=R ratio of the source impedance at the test site will obviously have a great

impact on the outcome. Hence, for evaluating ﬂicker, a procedure is proposed

where current measurements are made of the output of a test turbine and used to

synthesize the voltage variations which would be caused on distribution networks

with deﬁned short-circuit levels and X=R ratios of their source impedance. This is

illustrated in Figure 10.18 and is referred to in the Standard as simulation using a

‘ﬁctitious grid’. These voltage variation’ s a re then passed through a ﬂicker algo-

rithm to calculate the ﬂicker which the test turbine would cause on the deﬁned

networks. When the installation of the particular turbine is considered at a point on

the real distribution network these test results are then scaled to reﬂect the actual

Input power from generator

Figure 10.17 Six-pulse Two-level IGBT Voltage Source Converter (Graetz Bridge)

POWER QUALITY 589

short-circuit level and interpolated for the X=R ratio of the point of connection. A

weighting factor, based on an assumed Raleigh distribution of wind spee d, is also

applied to provide ﬂicker coefﬁcients which may be used on sites with various

average annual mean wind speeds.

The Standard also deﬁnes methods to evaluate the impact of wind turbine start-

up at cut-in and rated wind speeds and during speed changing of two speed

generators. Again currents are measured, combined with the ‘ﬁctitious grid’ to

provide a voltage time series and then passed through a ﬂicker algorithm. For

variable-speed wind turbines, harmonic currents are measured over 10 min obser-

vation periods.

10.6 Electrical Protection

All parts of a high-voltage power system are protected by relays which detect

abnormal conditions and circuit breakers which then open to isolate the faulty

circuits. Some lower-voltage circuits are protected by fuses, as these are cheaper but

do not give the degree of control offered by relays and circuit breakers. However,

fuses do have the advantage of operating very rapidly and so limiting the energy

transferred into the fault. On a distribution network the protection syst em is

designed primarily to detect excess currents caused by insulation failure in the

circuits. Failure of insulation, e.g., breakdown of air or of solid insulation, leads to

excess currents either between phases or between phases and gro und. These high

currents are only allowed to persist for up to about 1 s or so in order to limit the

hazards which include: (1) risk to life caused by excess voltages as the high currents

ﬂow into the ground impedance, (2) risk to plant cau sed by the destructive heating

and electromagnetic effects of the high currents, and (3) risk to the stability of the

power system.

The electrical protection of wind turbines and wind farms follows the same

general principles which are applied to any electrical plant (ALSTOM, 1987) but

there are two signiﬁcant differences. Because wind farms are frequently connected

to the periphery of the power syst em it is common to ﬁnd that the fault currents

RjX

Figure 10.18 Use of ‘Fictitious Grid’ to Establish Voltage Variations for Various Potential

Networks (I – measured current (complex quantity), R – resistance of ﬁctitious grid, jX –

inductive reactance of ﬁctitious grid, U – ideal source voltage, V – synthesized voltage jVj

passed through ﬂicker algorithm)

590

ELECTRICAL SYSTEMS

which will ﬂow in the event of insulation failure are rather small. Although this is

desirable from the point of view of reducing hazards, this lack of current can pose

signiﬁcant difﬁculties for the rapid and reliable detection of faults. In particular,

some designs of high-voltage fuses rely on the energy within the arc for their correct

operation. Hence they cannot be relied on to interrupt small fault currents when the

arc energy is low. Secondly, ﬁxed-speed wind turbines use induction machines and

variable-speed wind turbines are interfaced to the network through voltage source

converters. Neither induction generators or voltage source converters are a reliable

source of fault current and so voltage or frequency-sensing relays are needed to

detect abnormal conditions which are fed from wind turbine gen erators.

Figure 10.19 shows the current of an induction generator with a three-phase fault

applied to its terminals. It may be seen that the fault current dies away rapidly as

the stored magnetic energy in the electrical machine decays. There is no sustained

fault current as an induction generator draws its magnetizing current from the

network or local capacitors and this is not possible with the voltage collapsed by a

three-phase fault at the terminals. Some asymmetrical faults (e.g., two-phase faul ts)

can lead to sustained fault currents of two to three times that of full output (Jensen,

1990) but again these are not usually relied on to operate protective relays. How-

ever, it may be necessary to ensure that fault current supplied by wind turbines

does not lead to mal-operation of relays on the distribution syst em by altering the

ﬂow of fault current in the network and, in effect, providing local support of the

network voltage.

The protection of a wind farm has many similarities to the protection of a large

industrial load equipped with machine drives which may back-feed into the

network. This is a useful analogy, for the purposes of protection, as a wind farm

can be thought of as a collection of large electrical machine drives with torque

1.5

0.5

⫺0.5

⫺1

⫺1.5

200 400 600

Figure 10.19 Fault Current of Induction Generator with a Three-phase Fault Applied to

Terminals (Phase Shown with Minimum DC Offset)

ELECTRICAL PROTECTION 591

applied to the machine shafts rather than taken to drive mechanical loads. As in the

case of industrial machine drives, the terminal voltages and frequency of the

electrical machines are determined by conditions on the network. The distribution

network provides a reliable source of fault-current which can be used to detect

insulation failure while there is also the possibility of a fault-current contribution

from the rotating machines. In addition, on a wind farm there is also the hazard of

prolonged generation with abnormal frequency or voltage when the wind farm is

disconnected from the rest of the power system. This is known as islanding and is

an important co nsideration for all embedded generation schemes.

10.6.1 Wind farm and generator protection

Figure 10.20 shows a typical protection arrangement for a wind farm of ﬁxed-speed

wind turbines with generator voltages of 690 V and with a collection circuit voltage

of 11 kV. The 11 kV circuit is fed from a 33=11 kV De lta/Star wound transformer

with the 11 kV neutral grounded either directly or through a resistor. The

11=0:69 kV transformers are also wound Delta/Star and so the 690 V neutral points

of each circuit may be directly grounded. The neutral point of the generators is not

connected to ground.

There are a number of zones of protection. At the base of the wind turbine tower

a 690 V circuit breaker (usually a moulded case type as shown in Figure 10.1) will

be ﬁtted to protect the pendant cables and the generator. This is shown as Zone D.

Zone C in Figure 10.20 is the 690 V cables running from the turbine transformer

to the tower-base cabinet. Fuses or another moulded case circuit breaker may be

Utility Wind farm

Main substation

Zone A

33/11 kV

11 kV RMU

Zone B

Zone C

Zone D

Circuit breaker

11 kV switch fuse

11 kV/690V

Figure 10.20 Protection of a Wind Farm with an 11 kV Connection Circuit (RMU – Ring

Main Unit)

592

ELECTRICAL SYSTEMS

ﬁtted to the 690 V side of the turbine transformer to provide protection of the cables

and also a point of isolation so that all the electrical circuits of the turbine may be

isolated without switching at 11 kV. In some designs of wind turbine, the main

incoming busbar at the bottom of the electrical cabinet at the tower base consists of

exposed conductor. The electrical protection of this area needs careful consideration

as there is the possibility of dropping tools or other equipment on to these busbars.

Zone B is the 11 kV=690 V transformer including the region around its 690 V

terminals. This is a particularly difﬁcult zone to protect as the 11 kV fuses must be

robust enough to withstand the magnetizing inrush current of the transformer

while sensitive enough to detect faults on the 690 V terminals when the fault current

will be limited by the impedance of the transformer. This problem is common to all

11 kV=400 V transformers used on the public distribution network and a typical UK

solution is to use an 11 kV combination fuse-disconnect (sometimes known as a

switch-fuse). Faults on the 11 kV winding of the transformer will lead to high fault

currents which are cleared by the 11 kV fuse. However, for faults on the low-voltage

terminals, the fuse is unable to clear these low fault currents effectively and so, once

the fuse operates, a striker pin operates a mechanism to open the disconnect switch

to clear all the phase s. A more expensive solution adopted on some industrial

installations is to use so-called restricted earth fault protection to detect currents

leaking to ground from the low-voltage winding and terminal area but this requires

an 11 kV circuit breaker. There is obvious commercial pressure to reduce the costs

of transformer protection on a wind farm as there is a transformer for each wind

turbine. However, safety considerations require that all credible faults can be

detected and cleared.

Zone A is the 11 kV cable ci rcuit and this is protected in the conventional fashio n

by overcurrent and earth fault relays operating an 11 kV circuit breaker. The

33=11 kV transformer is protected in a similar manner to a transformer of the same

rating used on the public distribution system.

As wind-turbine ratings increase and larger wind farms are constructed, 11 kV

circuits cease to be cost-effective and so a number of large wind farms have been

constructed with a collection voltage of 33 kV. Figure 10.21 shows a typical

arrangement where a 33 kV wind farm is connected directly to a 33 kV public utility

network. This arrangement poses a number of additional difﬁculties for the

electrical protection. 33 kV switch fuses are not readily available and so it can be

difﬁcult to provide comprehensive protection of the 33 kV=690 V transform er for

the full range of prospective fault currents. Effective protection for a single phase to

ground fault on the low-voltage terminals is particularly difﬁcult. One attempt to

overcome these difﬁculties is to use a source protection relay to detect all faults on

the 33 kV circuit, transformer and low-voltage terminals (Haslam, Cross ley and

Jenkins, 1999). Although a suitable technique has been shown to be technically

successful this type of relay is not in production as the commercial demand for it is

too low.

In the arrangement of Figure 10.20 the 33=11 kV transformer provides two useful

features when considering electrical protection. Its impedance allows easier grading

of overcurrent relays and, as it blocks the passage of zero-sequence current, fast

acting earth fault relays can be applied to the wind-farm circuit. With direct

connection to the utility circuit these desirable features are not available and, in

ELECTRICAL PROTECTION 593

some countries, all embedded generation schemes are required to install an inter-

face transformer even if its voltage ratio is 1:1. This is not an economic solution but

does illustrate the difﬁculty of connecting a wind farm without a main transformer.

Electrical protection of wind farms follows conventional distribution engineering

practice with the network as the main source of fault current. In the protection of a

public supply network it is important only to isolate the faulty section or circuit and

so maintain supply to as many customers as possible. This discrimination of the

protection is less important in a wind farm as only some loss of generation will

occur if correct discrimination is not achieved and so simpler and lower cost

protection may be appropriate. The difﬁculty re mains, however, of ensuring that

effective protection is installed to detect all credible fault conditions even with

limited prospective fault currents.

10.6.2 Islanding and self-excitation of induction generators

Fixed-speed wind turbines use induction generators to provide damping in the

drive train and, as there is no direct access to the ﬁeld of an induction generator, the

magnetizing current drawn from the stator leads to a requirement for reactive

power. In order to reduce the reactive power supp lied from the network it is

conventional to ﬁt ﬁxed-sp eed wind turbines with local power factor correcti on

(PFC) capacitors. As long as the induction machine is connected to a distribution

network its terminal voltage is ﬁxed and so the PFC capacitors merely reduce the

reactive power drawn from the network. However, once the induction generator is

isolated from the network then there is the possibility of a resonant condition,

known as self-excitation, leading to large over-voltages (Allan, 1959, Hindmarsh,

1970). When the generator is isolated from the network it will tend to accelerate as

its load is removed. This increase in rotational speed, and consequently of

frequency, adds to the possibility of self-excitation. There have been a number of

Utility Wind farm

fuse

circuit breaker

Main substation

33/0.69 kV

Figure 10.21 Protection of a Wind Farm with a 33 kV Connection Circuit

594

ELECTRICAL SYSTEMS