Вы находитесь на странице: 1из 8

Benchmark of Annual Energy Production for

Different Wind Farm Topologies


Stephan Meier Philip C. Kjær
Royal Institute of Technology Vestas Wind Systems A/S R&D, Converter Design
Electrical Machines and Power Electronics Frankrigsvej 15
100 44 Stockholm, Sweden 8450 Hammel, Danmark
stephan@ekc.kth.se pck@vestas.com

Abstract— Wind power generation has become an established


alternative power source. Especially large wind farms in remote
or offshore locations are emerging strongly. Their grid connection
(1)
demands new transmission solutions as distances increase. A
newly proposed voltage source converter (VSC) based HVDC
transmission system looks promising compared to conventional
AC and DC transmission systems.
This paper presents a benchmark of the estimated annual energy
production (AEP) of a 200 MW wind farm depending on the (2)
transmission distance and the average wind speed. The proposed
system is compared to two state-of-the-art wind farm topolo-
gies: Variable-speed wind turbines with doubly-fed induction
generators (DFIG) and either AC or DC transmission systems.
The benchmark comprises detailed drive-train, converter, trans- (3)
former, distribution and transmission loss models. The total
system losses as well as the loss distribution between the different
components as a function of the transmission distance and the
average wind speed allows important conclusions for future wind
farm projects.
Fig. 1. Considered variable-speed wind farm topologies:
I. I NTRODUCTION (1) DFIG wind turbines with conventional HVAC transmission,
(2) DFIG wind turbines with conventional VSC based HVDC transmission,
Today, quite a few wind farms in the power range of several (3) Proposed VSC based HVDC transmission system.
hundred megawatts are under planning [1]. Promising sites
however are often situated in remote places or offshore, due
to better wind conditions. This leads to increasing distances regulate the network frequency. The independent control of
between wind farms and suitable grid connection points. Un- the reactive power at both cable ends allows to both excite
fortunately, AC cables inherently generate reactive power that the induction generators of the wind turbines and to assist the
limits the maximum permissible AC cable length. DC cables voltage control in the AC network [2].
however are not affected by cable charging currents and may
be as long as needed. Thus, for increasing distances, HVDC II. C ONSIDERED TOPOLOGIES
transmission based on VSCs, also called VSC transmission, This paper presents a benchmark of the AEP of three differ-
is a feasible and reliable solution compared to traditional AC ent wind farm topologies. Today, variable-speed wind turbine
transmission. generator systems (WTGS) with DFIGs are the predominant
An augmented application of VSC transmission systems is solution. In DFIG wind turbines, the stator of the induction
mainly limited by the cost of the expensive converter stations generator is directly connected to the wind farm grid via a
at both transmission ends and the semiconductor losses. Es- transformer, whereas the rotor windings are connected to a
pecially the high-frequency PWM switching generates losses frequency converter (back-to-back VSC) over slip rings. Even
and decreases the system efficiency. On the other hand, VSC though its slip rings are maintenance intensive, the DFIG is
transmission offers some valuable advantages. Currently, many beneficial as it allows the wind turbine to operate over a wide
new installed transmission lines are cables and no longer speed range with a frequency converter that is only rated at a
overhead lines. The right-of-way permission for underground fraction of the nominal power.
or submarine DC cables are often easier to obtain due to A topology with DFIG wind turbines and conventional HVAC
reduced environmental impacts. Furthermore, the VSC stations transmission as shown in Fig. 1.1 offers a simple and cost-
contribute to stabilize the AC network at their connection efficient solution for the grid connection of wind farms. In
points. The controllable active power flow can contribute to Fig. 1.2, the grid connection is instead realized by a conven-
+/− 150 kV 150 kV AUX
DC 500 Hz 50 Hz

33 kV, 500 Hz
3 kV, 500 Hz
1 kV, fvar

Gearbox
Squirrel−cage induction generator
AUX
Single−phase VSC

Cycloconverter

Auxiliary power 500/50Hz−converter


Single−phase MF transformer

Circuit breaker
AUX
Disconnector

Second−order shunt filter

Cycloconverter output filter

Fig. 2. Topology of the proposed VSC transmission system.

tional VSC based HVDC link. One of the three considered single-phase VSC, whereas one of the transformer terminals
topologies is new, and differs strongly from previously pub- is connected to the midpoint in the DC link created by bus-
lished wind power transmission systems, as the wind turbine splitting capacitors. These capacitors provide the DC voltage
converters and the power collection between the wind turbines source necessary for the dynamics of the system and limit the
are inseparable and specific. Fig. 1.3 shows this system, which voltage ripple on the DC line. Series-connected IGBTs with
is described in the following section. antiparallel diodes and snubber capacitors form the valves of
the VSC. The snubber capacitors allow the IGBTs to turn-off
III. P ROPOSED SYSTEM at zero-voltage conditions.
The topology of the proposed soft-switched AC/DC con-
A. Principle of operation
version system is shown in Fig. 2. It basically incorporates a
single-phase VSC and cycloconverters connected via a single- By alternately commutating the cycloconverters and the
phase medium-frequency (MF) collection bus. VSC it is possible to achieve soft commutations for all the
Thereby, all wind turbine nacelles are equipped with the semiconductor valves [3]. The cycloconverters can be solely
following components: The drive train comprises a gearbox, operated by natural commutation, whereas snubbered or zero-
a squirrel-cage induction generator and a cycloconverter with voltage commutation is always enabled for the VSC.
a passive output filter. The valves of the cycloconverter do Thereby, the VSC is commutated at fixed time instants with
not need any turn-off capability and can be realised by constant intervals (500 Hz switching frequency), thus generat-
fast thyristors connected in anti-parallel. The auxiliary power ing a square-wave voltage. The cycloconverter phase legs are
demand is directly supplied from the wind turbine mains over then commutated in order to obtain the desired PWM generator
a single-phase transformer and a frequency converter. The voltages from the MF square-wave voltage. A more detailed
connection of the wind turbine to the local collection bus is description of the operation principle and the basic waveforms
integrated in the bottom of each tower. An MF transformer was previously published in [4].
increases the voltage to 33 kV, where a circuit breaker enables
the wind turbine to disconnect from the collection bus (e.g. B. Specific features
during faults or at low wind speeds and during maintenance). The proposed converter topology differs considerably from
A second-order shunt filter dampens the ringing of the square- conventional VSC transmission systems, as e.g. HVDC Light
wave voltage caused by the cable resonance of the collection from ABB [5].
grid. A main ambition of the proposed system is lower initial costs.
The power collection from the individual wind turbines is Single-phase MF transformers are not only cheaper than three-
realized by a single-phase MF bus. It connects the wind phase transformers but also more compact which simplifies
turbines to an offshore platform comprising the main circuit their integration in the wind turbines. However, the design
breakers, the main transformer and a single-phase VSC. The of the MF transformers has to be adapted to the specific
main MF transformer raises the bus voltage to 150 kV for characteristics of the proposed topology. Especially the design
onward transmission, which is half the DC link voltage. of the transformer insulation needs to withstand high voltage
The high-voltage side of the transformer is connected to a derivatives (however limited by the snubber capacitors of the
TABLE I
VSC). The initial costs are further decreased by a significant
S YSTEM PARAMETERS
reduction of series-connected IGBT valves in the VSC. IGBTs
are expensive and require complex gate drives and voltage- Electrical system parameters
Rated power 201 MW
sharing circuitries. A prior publication revealed that the IGBT Collection bus voltage 33 kV
power rating of the main VSC is reduced by approximately HVDC link voltage ±150 kV
30 % compared to a conventional VSC for the same effective AC transmission voltage 132, 150 or 220 kV1
Wind turbine parameters
switching frequency [6]. Compared to the IGBT-based fre-
Rated active power 3 MW
quency converters in the DFIGs, the cycloconverter valves con- Number of wind turbines 67
sist of comparatively cheap and well-established fast thyristors. DFIG frequency converter rating 300 kW
1
However, unlike the frequency converters, the cycloconverters The choice of the cable voltage is discussed in section V-E.2.

are rated at nominal power.


The high-frequency PWM switching losses of the VSCs con- Offshore platform
tribute considerably to the total system losses. The proposed
topology however offers soft-switching of all semiconductor
valves and reduces the switching losses substantially. In ad-
dition, the thyristors in the cycloconverters have low losses

4 km
compared to IGBTs. Thus, the converter efficiency in the wind
farm (both cycloconverters and single-phase VSC) PSfrag replacements
is expected
to reach up to approximately 99 % [6]. The application of
single-phase MF transformers reduces the total system losses
10.4 km
further. However, the square-wave voltage on the MF collec-
tion bus causes higher cable and filter losses within the wind
Fig. 3. Wind farm layout with 33 kV collection bus.
farm.

IV. S YSTEM PARAMETERS


measuring the power performance characteristics of WTGSs
The considered wind farm is rated at 201 MW, collecting the connected to the electric power network. The power perfor-
power from 67 identical 3 MW wind turbines. The two refer-
mance characteristics of WTGSs are defined by the measured
ence topologies have DFIGs, where the generator excitation is
power curve and the estimated AEP. The AEP is an estimate
provided from the rotor converter, allowing the wind turbine of the total energy production of a WTGS during a one-year
to generate power at unity, or any other, power factor. For
period by applying the measured power curve to different
the proposed topology, the cycloconverters have to supply the
reference wind speed frequency distributions at hub height,
reactive power to compensate the magnetising currents of the
assuming 100 % availability. Thereby, a Rayleigh distribution
induction generators. A typical DFIG wind turbine operates
is used as the reference wind speed frequency distribution.
in the slip-range ±10-15 %, which approximately reflects the AEP calculations should be made for annual average wind
frequency converter rating. Modern wind turbines must also
speeds Vavg of between 4 and 11 m/s [8] as
comply with new grid connection requirements [7] concerning
fault ride-through and particular power-factor ranges. To meet N   P 
i−1 + Pi
X
these requirements, the converter rating for the DFIGs may AEP = Nh F (Vi ) − F (Vi−1 ) , (1)
i=1
2
increase somewhat in the near future. The wind turbines
generate into a collection bus of 33 kV, before the voltage where Nh is the number of hours in one year ≈ 8760 h;
is transformed up for onward transmission. In case of HVDC N is the number of measuring points;
transmission, the DC cable voltage is ±150 kV. Fig. 3 shows Vi and Pi are the normalized and averaged wind speed
the chosen wind farm and collection bus layout, with a distance respectively power output;
of nine times the blade diameter between two wind turbines and F (Vi ) is the Rayleigh cumulative probability distribution
perpendicular to the prevailing wind direction. Table I presents function for a certain wind speed as shown in Fig. 4.
an overview of the system parameters. In addition to standard IEC61400-12, this benchmark includes
also the transmission losses between the wind farm and the
V. M ETHODOLOGY AC grid. Thus, the significant estimated AEP that becomes
This section presents the applied methodology in order to available for onward distribution can be expressed as a func-
determine the AEP and the losses in the different system tion of the transmission distance and the average wind speed.
components.
A. Annual energy production B. Wind turbine power curve
The estimated AEP for the considered wind farm topologies According to standard IEC61400-12, the single-turbine
is calculated according to the IEC wind energy standard power curve was determined by collecting simultaneous mea-
IEC61400-12 [8]. This standard specifies the procedure for surements of wind speed and power output on the high-voltage
1 1.5

0.9
Cumulative probability F (V )

4 m/s

AEP gain (%)


0.8
1
0.7

0.6
Vavg
0.5
0.5
11 m/s
replacements 0.4
PSfrag replacements
0.3 0
4 5 6 7 8 9 10 11
0.2
Average wind speed Vavg (m/s)
0.1

0
0 5 10 15 20 25 30
Fig. 6. Increased AEP of the proposed system compared to a DFIG-system
Wind speed V (m/s) due to lower drive train losses

Fig. 4. Rayleigh cumulative probability distribution function for average


wind speeds Vavg between 4 and 11 m/s.
transformer, the drive train losses and the turbine’s auxiliary
1
power consumption are already included in the power curve
Single−turbine
of the wind turbine.
Normalized power P (p.u.)

Multi−turbine

0.8 Howsoever, the proposed system differs considerably from


the DFIG-system of the Vestas V90-3MW. The frequency
0.6
converter is replaced by a cycloconverter and the three-phase
transformer by a single-phase MF transformer. The reduced
replacements 0.4
drive train losses of the proposed system were considered in a
0.2 power gain compared to the DFIG-system. Fig. 6 shows this
power gain in the AEP depending on the average wind speed.
ingle-turbine 0 It can be seen that the smaller no-load losses of the proposed
0 0.5 1 1.5 2 2.5 3
Multi-turbine Normalized wind speed V (p.u.) system increase the power gain at low average wind speeds.
Fig. 5. Normalized single- and multi-turbine power curves. The average D. Converter losses
wind speed for the calculation of the multi-turbine power curve is 8 m/s and
the normalized standard deviation σ is 0.1 relative to the average wind speed. The VSC losses are analytically calculated as a function of
the wind speed, respectively power output. The losses in the
three-phase VSCs have been confirmed with measurements
side of the wind turbine transformer. The power output of from ABB on the Cross Sound Cable Interconnector [11].
variable-speed wind turbines is rather smooth as they are The frequency modulation ratio for both wind farm and
storing wind gusts in rotational energy. Fig. 5 shows the grid side VSCs is chosen as p = 21, thus assuring the
normalized power curve of the commercially available 3 MW same effective switching frequency as the proposed topology.
wind turbine Vestas V90-3MW [9], where the output power is Detailed calculations of the converter losses were previously
constant above a wind speed of 14 m/s. The rated wind speed published in [6].
however is 9 m/s, above which the output power is gradually
reduced by pitch control. E. Transmission and distribution losses
Fig. 5 shows an example of a multi-turbine power curve that 1) DC cable losses: The choice of an appropriate DC cable
considers the smoothing effects of the aggregated power output for the specific nominal power of the wind farm is based on
from the whole wind farm [10]. It is used for loss calculations a comparison with similar HVDC transmission projects. With
on the wind farm level. The multi-turbine power curve takes the choice of the cable NOVA-L 1x630 mm2 Cu from Nexans
the spatial wind speed distribution into account, while the (see Table III), a reasonable margin between the nominal wind
accumulated AEP remains unchanged compared to the single- farm current Idc and the maximum allowable current Imax is
turbine power curve. Assuming an average wind turbulence obtained (refer to Table II).
intensity and a spacial distribution of the wind turbines ac- Reference [12] presents a calculation procedure for the deter-
cording to Fig. 3, the normalized standard deviation of the mination of cable losses and temperatures, taking the tempera-
wind speed distribution σ was determined as approximately ture dependence of cable losses into account. The temperature
0.02 [10]. increase in the cable core ∆θ depends on the cable current I
as
C. Drive train losses 
I
2
∆θmax cα Imax
Drive train losses comprise mechanical losses in the gearbox ∆θ =  2 , (2)
and the bearings as well as electrical losses in the generator, cm − ∆θmax α20 ImaxI
converter, transformer and filter. Due to the fact that the wind
turbine power curve is determined by measurements of the where
power output on the high-voltage side of the wind turbine ∆θmax = θmax −θamb is the max. allowable temperature rise;
TABLE II TABLE III
DC CABLE DIMENSIONING C ABLE PARAMETERS [14]
Cross Sound Cape Wind Proposed DC cable: θmax = 70 ◦ C, θamb = 15 ◦ C, θop < 41 ◦ C
Cable [11] Cable [13] Wind Farm Voltage [kV] Imax [A] R0max [Ω/km]
Pnom [MW] 330 420 201 NOVA-L 1x630 mm2 Cu 300 935 0.0322
Vdc [kV] ± 150 ± 150 ± 150
Idc [A] 1100 1400 670 AC cable: θmax = 90 ◦ C, θamb = 15 ◦ C, θop < 70 ◦ C
Cu-cable [mm2 ] 1x1300 2x630 1x630 Voltage [kV] Imax [A] R0max [Ω/km]
Imax [A] 1500 2x935 935 TKRA 3x1x240 mm2 KQ 132 490 0.121
Margin [%] 36.4 33.6 39.6 TKRA 3x1x300 mm2 KQ 132/150 545 0.102
TKRA 3x1x400 mm2 KQ 132/150 610 0.086
TKRA 3x1x500 mm2 KQ 132/150/220 675 0.073
TKRA 3x1x630 mm2 KQ 132/150/220 745 0.062
θmax = 70 ◦ C is the maximum conductor temperature; TKRA 3x1x800 mm2 KQ 132/150/220 810 0.055
θamb = 15 ◦ C is the ambient temperature; TKRA 3x1x1000 mm2 KQ 132/150/220 870 0.049
α20 is the temperature coefficient of the conductor resistivity; TKRA 3x1x1200 mm2 KQ 132/150/220 910 0.046
cα = 1 − α20 (20 − θamb), cm = 1 − α20 (20 − θamb − ∆θmax ).
Distribution cables: Voltage [kV] Imax [A] R0max [Ω/km]
The cable losses per unit length [12] can then be calculated TKRA 3x1x95 mm2 KQ 33 305 0.26
as  2   TKRA 3x1x400 mm2 KQ 33 610 0.08
0 0 I cα α20 ∆θ
P = Pmax + , (3)
Imax cm cm
where
0 0 2 metallic shield and the steel wire armor. The dielectric losses
Pmax = Rmax Imax is the maximum ohmic cable loss.
can be found as:
2) AC cable losses: The capacitance in submarine AC U2
cables plays a major role in limiting their technically and Pd0 = 2πf C 0 tan(δ), (6)
3
economically feasible length l. The distributed capacitance C 0
causes charging currents in the cable, thus limiting the load where tan(δ) = 0.0003 is the loss angle.
carrying capability of the cable. The longer the cable and the In contrast to (3), taking the longitudinal distributions of
higher the voltage, the higher the resulting charging current current and temperature into account becomes essential when
Ic : calculating AC cable losses. The mean cable losses P 0 per unit
U length can be derived from the current distribution along the
Ic = 2πf lC 0 √ , (4) cable by integration of the ohmic losses over the cable length
3
l and adding the dielectric losses [12]:
where
f = 50 Hz is the fundamental frequency and U the cable 0
Pmax
Z l 
cα α20 ∆θ(x)

2
voltage (132 kV, 150 kV or 220 kV). P = 2
0 I (x) + dx + Pd0 , (7)
lImax x=0 cm cm
The load carrying capability of the cable can be increased
considerably by installing reactive power compensation units where
at both cable ends. Thus, the resulting maximum cable charg- 0
Pmax 0
= 3Rmax 2
Imax is the maximum ohmic cable loss.
ing current is only half the one from (4). In this benchmark, Table III presents those submarine AC cables from Nexans
reactive power compensation is assumed at both cable ends. [14] that were considered in this benchmark. The choice of
From (2), the nominal cable current Iac can be calculated from an appropriate cable arrangement was based on the following
the maximum cable current Imax and the chosen operating considerations: First, the number of parallel cables should be
temperature at nominal load. Submarine XLPE cables can be as small as possible. A single three-core cable is favored due to
loaded continuously up to a conductor temperature of θmax lower environmental impacts and costs, although two ore more
= 90◦ C. However, in order to keep the losses lower and to parallel cables offer redundancy. Second, from an economical
avoid possible thermal instability, the operating temperature is point of view, the cables should not be overdimensioned in
chosen to be limited to θop < 70◦ C. This becomes particularly order to not distort the loss calculations.
important with regard to the fact that the highest currents 3) Compensation unit losses: To ensure a maximum cable
appear at the cable ends where the cooling is most difficult. capacity, both transmission cable ends are equipped with
The available capacity to carry load current Iload is given reactive power compensation units. Usually they consist of
by (5). This cable capacity determines the number of parallel shunt reactors, of which some need to be controllable. When
cables required to transmit the nominal active power. controllable shunt reactors are used, the most common are
thyristor-controlled reactors (TCR). Filters are used to reduce
s  2
Ic
Iload = Iac −2 (5) the harmonics injected into the power system by the TCR
2 operation.
The losses in AC submarine cables are composed of dielectric The internal power loss of the compensation units was as-
losses (relatively small) and ohmic losses in the conductors, sumed to be 5 kW per MVAr of rated reactive power.
TABLE IV 1000
900
T RANSFORMER LOSS CARACTERISTICS
800

AEP (GWh)
700
50 Hz three-phase transformers
600
Pnl Pl Efficiency Loss ratio
500
33/1 kV, 3 MW 6 kW 12 kW 99.4% 0.50
400
132/33 kV, 200 MW 115 kW 650 kW 99.62% 0.18
300
150/33 kV, 200 MW 130 kW 650 kW 99.61%
220/33 kV, 200 MW 190 kW 650 kW 99.58%
PSfrag0.20
replacements
0.29
200
100
0
4 5 6 7 8 9 10 11
500 Hz single-phase transformers
Pnl Pl Efficiency Loss ratio Average wind speed Vavg (m/s)
33/3 kV, 3 MW 2.1 kW 2.5 kW 99.85% 0.84
Fig. 7. AEP of the Vestas V90-3MW depending on the average wind speed.
150/33 kV, 201 MW 47.5 kW 70 kW 99.94% 0.68

11
DFIG with HVAC transmission system
10
DFIG with VSC transmission system
4) Wind farm distribution losses: The three-phase distribu- 9 Proposed VSC transmission system

Losses (%)
8
tion losses in the wind farm are calculated according to [12]. 7
Thereby, the wind farm layout as given in Fig. 3 comprises 6

81.2 km of 33-kV three-core submarine cables. The cable 5

characteristics are given in table III. The cable PSfrag


area 95 replacements
mm2 is 4

2 3
suitable for all cable chains, whereas the cable area 400 mm 50 75 100 125 150 175 200 225
is necessary for the connection of the individual cable chains Transmission distance (km)
to the offshore platform.
Fig. 8. Transmission system losses of the three considered topologies at an
The cable and filter losses of the single-phase MF collection average wind speed of 8 m/s.
grid are assumed to be twice the losses of the corresponding
three-phase collection grid. This assumption was chosen in
order to rather overestimate the losses before exact determi- filter design with a second-order shunt termination of
nation. the collection cable can dampen the ringing independent
5) Transformer losses: The transformer core losses, also of the cable length. These investigations are essential in
called no-load losses Pnl , are caused by the magnetizing cur- order to correctly estimate the MF collection grid losses.
rent needed to energize the transformer core and are essentially - The choice of the AC cable arrangement at different
invariant with the loading of the transformer. Therefore, a conditions was simplified and intended mainly to limit
small loss ratio Pnl /Pl is favorable for power transformers. the number of parallel cables in order to get a better
No-load losses are mostly hysteresis and eddy current losses comparability with the VSC transmission systems. The
in the core laminations. optimization of the AC cable arrangement has to include
The copper or load losses Pl arise from resistance losses in the such aspects as compensation unit losses, cable losses
primary and secondary transformer windings. Load losses vary as well as initial costs and may differ considerably
according to the loading of the transformer with the square of depending on the project conditions. In addition, the
the current. Table IV shows the assumed transformer losses, calculation of the compensation unit losses has to be more
whereas the single-phase MF transformer losses are calculated detailed.
according to [15]. - According to standard IEC61400-12, 100 % availability
VI. S IMPLIFICATIONS AND A SSUMPTIONS was also assumed for the transmission systems. The
The following simplifications and assumptions should be available AEP for onward distribution however may be
kept in mind when regarding the results of this article: influenced considerably by taking the reliability of the
different transmission system components into account,
- The Vestas V90-3MW was not especially designed as
e.g. the converter stations may reduce the reliability of
offshore wind turbine. The power curve of an ideal
the VSC transmission systems. But this aspect is out of
offshore wind turbine may therefore slightly differ from
the scope of this article.
the power curve in Fig. 5. Moreover, the V90-3MW has
been adapted for the DFIG and may not be an optimal
solution for the full-scale converter of the proposed VSC VII. R ESULTS
transmission topology.
A. AEP
- The single-phase MF transformer losses are only based
on analytical calculations and may turn out to be higher The accumulated estimated AEP of the wind turbines
in reality. without considering any distribution or transmission losses
- The single-phase MF collection grid has to be further is approximately proportional to the average wind speed, as
investigated regarding the ringing of the square-wave shown in Fig. 7. It can be seen that the AEP gain above the
voltage caused by the cable resonance. An appropriate rated wind speed of 9 m/s is slowly starting to decrease.
12 12 12

10 10 10

8 8 8

Losses (%)
Losses (%)

Losses (%)
placements PSfrag replacements PSfrag replacements
6 6 6

AC cable losses DC cable losses


4 4 4
DC cable losses

VSC losses VSC losses


2 Compensation unit losses 2 2

Distribution losses (Main transformer and collection grid) Distribution losses (Main transformer and collection grid) Distribution losses (Main transformer and collection grid)
0 0 0
50 75 100 125 150 175 200 225 50 75 100 125 150 175 200 225 50 75 100 125 150 175 200 225
Transmission distance (km) Transmission distance (km) Transmission distance (km)

(a) DFIG with HVAC transmission system (b) DFIG with VSC transmission system (c) Proposed VSC transmission system

Fig. 9. Loss distribution in the different components of the three considered transmission systems at an average wind speed of 8 m/s.

9
B. Losses vs. transmission distance
8 DFIG with HVAC transmission system
DFIG with VSC transmission system
Proposed VSC transmission system

Losses (%)
7
Fig. 8 shows the transmission system losses in percent of
the AEP for the three considered topologies as a function 6

of the transmission distance. The transmission losses of the 5

proposed VSC transmission system are decreased PSfragby replacements


0.62 % 4

(of the AEP), corresponding to the power gain due to lower 3


4 5 6 7 8 9 10 11
drive train losses at an average wind speed of 8 m/s (according Average wind speed Vavg (m/s)
to Fig. 6). It can be seen that the proposed VSC transmission
system offers approximately 1 % (of the AEP) lower losses Fig. 10. Transmission system losses for the three considered topologies at
than a conventional VSC transmission system at all trans- a transmission distance of 100 km.
mission distances. However, conventional HVAC transmission
systems generate least losses up to a transmission distance of
C. Losses vs. average wind speed
approximately 75 km. Above 125 km, both VSC transmission
systems generate lower losses than the HVAC transmission Fig. 10 shows the transmission system losses in percent
system. It is interesting to know, that a single 220 kV three- of the AEP as a function of the average wind speed. The
core HVAC cable with a conductor diameter of 1000 mm2 can transmission distance is 100 km, which is often considered to
transmit the rated power upp to a distance of 150 km. be a critical length for AC transmission systems [2]. Further-
Fig. 9 shows the loss distribution in the different transmission more, at a transmission distance of 100 km, the transmission
system components for the three considered topologies. For losses of all three considered topologies are approximately
the DFIG system with HVAC transmission in Fig. 9(a), the in the same range. The transmission losses of the proposed
cable and compensation unit losses dominate over the main VSC transmission system are again decreased by the power
transformer and collection grid losses, especially at increasing gain from the lower drive train losses according to Fig. 6. The
transmission distances. It can be noticed that the compensation transmission losses of the proposed system are below 4.5 % (of
unit losses decrease somewhat at a transmission distance of the AEP) at all average wind speeds and thus the smallest for
175 km, where the single 220 kV cable has to be replaced the three compared transmission systems. Especially at low
by two 132 kV cables to be able to transmit rated power. average wind speeds, the proposed system is advantageous
Fig. 9(b) shows the loss distribution for the DFIG system due to its reduced VSC losses and comparably low no-load
with VSC transmission, where the VSC losses dominate. and drive-train losses. Both DFIG systems with either HVAC
They are approximately 3.4 % of the AEP and independent or HVDC transmission systems have increasing losses at low
of the transmission length. This makes an application of average wind speeds.
VSC transmission systems ineligible for shorter transmission Fig. 11(a) shows the loss distribution in the different trans-
distances. It can be noticed that the DC cable losses are lower mission system components of the DFIG system with HVAC
compared to AC cable losses, especially at long transmission transmission. It can be seen that both the AC cable and
distances. Fig. 9(c) shows the transmission loss distribution distribution losses in percent are approximately constant and
for the proposed VSC transmission system. As expected, the independent of the average wind speed. The compensation
soft-switched single-phase VSC contributes to decrease the unit losses in percent however are increasing significantly
VSC losses. The distribution losses are however higher due at low average wind speeds. This is due to the fact that
to the increased collection grid losses that outweight the loss these losses are constant (only depending on the transmission
reduction from the single-phase MF transformer. distance and the cable voltage level) and that their significance
9 9
9

8 8
8

7 7
7

6 6
6

Losses (%)
Losses (%)

placements PSfrag replacements PSfrag replacements

Losses (%)
5
5 DC cable losses 5

4
DC cable losses
4 4

AC cable losses 3 VSC losses


3 3
VSC losses
2
Compensation unit losses 2 2

1 1
1
Distribution losses (Main transformer and collection grid) Distribution losses (Main transformer and collection grid) Distribution losses (Main transformer and collection grid)
0 0
0 4 5 6 7 8 9 10 11 4 5 6 7 8 9 10 11
4 5 6 7 8 9 10 11
Average wind speed Vavg (m/s) Average wind speed Vavg (m/s) Average wind speed Vavg (m/s)

(a) DFIG with HVAC transmission system (b) DFIG with VSC transmission system (c) Proposed VSC transmission system

Fig. 11. Loss distribution in the different components of the three considered transmission systems at a transmission distance of 100 km.

increases when the AEP decreases. The loss distributions for wind turbine with HVAC transmission is very competitive for
the DFIG system with VSC transmission in Fig. 9(b) and shorter transmission distances due to the absence of converters
the proposed VSC transmission system in Fig. 9(c) show a and the high turbine efficiency.
similar tendence. Thereby, the DC cable losses in percent
ACKNOWLEDGMENT
are increasing approximately proportional to the average wind
speed. As expected, the VSC losses in percent are higher at The authors would like to express their gratitude to Vind-
low average wind speeds due to the impact of the no-load Forsk and the Swedish Energy Agency for financial support. A
losses. special thank goes to ABB [11] and Nexans [14] for providing
the underlying data of this benchmark.
VIII. C ONCLUSIONS R EFERENCES
This paper presented a benchmark of the estimated AEP [1] S. Shaw, M. J. Cremers, and G. Palmers, “Enabling Offshore Wind
as a function of the transmission distance and the average Developments,” EWEA - Brussels, 2002.
[2] N. M. Kirby, L. Xu, M. Luckett, and W. Siepmann, “HVDC transmission
wind speed for three different wind farm topologies. The main for large offshore wind farms,” in Power Engineering Journal, vol. 16,
conclusions from this work are: pp. 135–141, June 2002. Issue 3.
[3] S. Norrga, “Novel Soft-switching Isolated Three-phase Bidirectional
- The critical length for HVAC transmission systems re- AC/DC Converter,” in Proceedings of the Nordic Workshop on Power
garding the losses lies in the range of 100 km. At and Industrial Electronics, Norpie ’02, Stockholm, Sweden, August
such long transmission distances, the cable dimensioning, 2002.
[4] S. Meier, S. Norrga, and H.-P. Nee, “New topology for more efficient
arrangement as well as the chosen voltage level are AC/DC converters for future offshore wind farms,” in Proceedings of the
important design criterias in order to reduce the AC cable 4th Nordic Workshop on Power and Industrial Electronics, Norpie ’04,
losses. Trondheim, Norway, 2004.
[5] A. K. Skytt, P. Holmberg, and L. E. Juhlin, “HVDC Light for Connection
- Distribution and VSC losses in percent of the AEP of Wind Farms,” in Second International Workshop on Transmission
are nearly independent from the transmission distance, Networks for Offshore Wind Farms, Royal Institute of Technology,
which makes VSC transmission systems ineligible at Stockholm, Sweden, March 29-30, 2001.
[6] S. Meier, S. Norrga, and H.-P. Nee, “New Voltage Source Converter
short transmission distances. Topology for HVDC Grid Connection of Offshore Wind Farms,” in
- The DC cable losses in percent of the AEP are increasing Proceedings of the 11th International Power Electronics and Motion
proportional to the transmission distance wheras AC Control Conference, EPE-PEMC ’04, Riga, Latvia, 2004.
[7] “Grid code for high and extra high voltage.” E.ON Netz GmbH.
cable losses are increasing more than proportional due [8] “International Standard IEC 61400-12: Wind Turbine Generator Systems
to charging currents. - Part 12: Wind Turbine Power Performance Testing,” First edition,
- At low average wind speeds, the tranmission system February 1998.
[9] “www.vestas.com.”
has to be designed with special regard on the no-load [10] P. Nørgaard and H. Holttinen, “A Multi-Turbine Power Curve Ap-
losses. In particular the compensation unit and VSC proach,” in Nordic Wind Power Conference, Gothenburg, Sweden, March
losses become very dominant at low average wind speeds. 1-2, 2004.
[11] “www.abb.com.”
The proposed VSC transmission system offers approximately [12] H. Brakelmann, “Kabelverluste und Verlustkosten in Windparks,” in
1 % (of the AEP) lower losses than conventional VSC trans- Bulletin SEV/VSE 24/25, 2002.
[13] www.capewind.org, “Transmission Issues for Offshore Wind Farms with
mission systems and has the lowest losses of all three con- Specific Application to Siting of the Proposed Cape Wind Project,” ESS
sidered topologies above a transmission distance of 100 km Inc., 2003.
at all average wind speeds. However, the application of the [14] “www.nexans.com.”
[15] T. Kjellqvist, S. Norrga, and S. Östlund, “Design Considerations for
proposed VSC transmission system may already be advan- a Medium Frequency Transformer in a Line Side Power Conversion
tageous for far shorter transmission distances depending on System,” in Proceedings of the 35th Annual IEEE Power Electronics
the specific project conditions. It is also clear that the DFIG Specialists Conference, PESC ’04, Aachen, Germany, 2004.

Вам также может понравиться