AckermannTh. (ed) Wind Power in Power Systems

Подождите немного. Документ загружается.

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 441 – [433–460/28]

17.12.2004 10:43PM

influence the NTC, even if the technical parameters remain constant. TSOs also model

the generation increa se or decrease, E, in different ways.

When determining the limits of a feasible network operation, some TSOs include

thermal limits that vary over the year and across their respective areas. Other TSOs

apply probabilistic models based on meteorological statistics. Some TSOs assume

constant ambient conditions throughout the year. The maximum allowed continuous

conductor temperature differs largely from one TSO to the other, with values ranging

from 50



C to 100



C. Some TSOs allow higher continuous current limits in (n  1)

contingency situations (i.e. the outage of a single network element). However, the

percentage of accepted overloads is also different for different TSOs. Many TSOs

tolerate higher current limits in contingency situations only if the TSO itself is able

to decrease the loading to a level below normal limits within a short period of time

(10–30 min). Voltage stability is assessed for normal system operation as well as for the

operation after (n  1) contingencies that are considered relevant for the security

assessment. Some TSOs investigate not only single failures but also certain failure

combinations. Nordic TSOs (Fingrid, Statnett, and Svenska Kraftna

t) also include

busbar failures, as possible consequences may be severe and endanger the security of

the system. The TSOs also have diverging methods for assessing uncertainty as well as

sourses of uncertainty.

Apart from determining the NTC twice a year, the available transmission capacity is

also determined on a daily or weekly basis for deay-ahead congestion forecasts. The

method follows the pattern described above, but here the base case reflects a load flow

forecast based on snapshots from the preceding day, and sometimes weather forecasts.

The system models are updated according to changes in topology and switching status.

Weather forecasts are used by some TSOs to allow for higher thermal transmission

limits. As there are less uncertainties related to the short-term horizon, transmission

reliability margins can be decreased. As a result of these factors, the actual allocation of

transmission capacity can vary substantially from the NTC values that are calculated

twice a year.

20.3.2 Determination of transmission capacity within the country

Each TSO applies to its area of responsibility complete system models for assessing the

available transmission capacity. All TSOs use simila r methods, but the security stand-

ards and base scenario assumptions again can vary considerably.

As mentioned in Section 20.2.1, thermal limits can be evaluated for each line depend-

ing on material, age, geometry, the height of the towers, the security standards that limit

the clear ance to the ground and so on. There can be different policies regarding the

tolerance to overloads during contingencies, too.

The load flow is calculated for different scenarios. These calculations are then used

for assessing the voltage stability during normal operation as well as after the most

frequent and severe contingencies. The selection of failures to be asses sed is based on the

subjective distinction between ‘frequent’ and ‘rare’ failures and between ‘severe’ and

‘minor’ consequences (Haubrich et al., 2001). As was discussed in Section 20.3.1, the

types of contigencies that are included in the assessment vary considerably.

Wind Power in Power Systems 441

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 442 – [433–460/28]

17.12.2004 10:43PM

Transient stability is analysed by performing dynamic simulations of the system

during and after the fault. The transmission limits for the respective lines are defined

by the most severe set of conditions.

The transmission limits within the country are taken into account:

assessing the securi ty for the cross- border capacity allocation;

in order to adopt corrective measures in the case of congestions, such as redispatching

generators, switching and load shedding;

in order to prepare the technical prerequisites for new generation and so on.

20.3.3 Summary

Transmission capacity does not always reflect only the technical properties of the

system. Determining the transmission capacity is a complex task and its results depend

heavily on the assumptions made by each individual TSO. Although the general scheme

for capacity allocation is similar for all European TSOs, differences in definitions and

assumptions can lead to lower calculated transmission capacities. Thus a harmonised

approach to determining the transmission capacity can increase transmission capacity

without requiring substa ntial investment. This and other methods to increase transmis-

sion capacity will be discussed in the next section.

20.4 Measures to Increase Transmission Capacity

20.4.1 ‘Soft’ measures

‘Soft’ measures may result in an increase of transmission capacity at low cost. They

mostly concern cross-border transmission capacity; however, some of them may be

applied to increase transmission capacity within a country:

Determination methods between adjacent TSOs can be harmonised, with transpar-

ency regarding the base case assumptions (Haubrich et al., 2001).

Ambient temperature and wind speed statistics can be considered when determining

the transmission capacity.

Temperature and wind speed forecasts for day-ahead capacity allocation can be used.

On the one hand, the deterministic (n  1) criterion leads under certain conditions to

unnecessarily high congestion costs for power producers and for grid operators. On

the other hand, during, for example, under unfavourable weather conditions, (n  1)

constraints may not provide sufficient security. A novel approach to online security

control and operation of transmission systems is suggested by Uhlen et al. (2000).

Their approach is based on a probabilistic criterion aiming at the online minimisation

of total grid operating costs, which are defined as the sum of expected interruption

costs and congestion costs during specified periods of operation. This approach is

currently being studied by the Norwegian TSO and by the SINTEF (Stiftelsen for

industriel og teknisk forskning ved Norges tekniske høgskole) research institute. But

even the approach currently applied by the Norwegian system operator allows flexible

442 Areas with Limited Transmission Capacity

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 443 – [433–460/28]

17.12.2004 10:43PM

transmission limits by taking into account various criteria, such as the costs for

redispatching, weat her conditions and system protection. The value of the reduced

congestion is then weighted against the expected interruption costs and/or possible

congestion management costs. If the value of the reduced congestion is substantially

higher than the increa se in expected interruption costs, the system is operated at

(n  0) reliability (Breidablik, Giæver and Glende, 2003).

Temporary overloading after contingencies may be allowed. Depending on ambient

temperature, a short-term loading of 30–50 % in excess of rated capacity is accepted

for transformers, breakers and other components without a substantial loss of lifetime

(Breidablik, Giæver and Glende, 2003).

Unjustified, rare, fault considerations can be excluded from the determination of the

transmission capacity.

A probabilistic evaluation of operational uncertainties can be carried out. In this case, risk

could be defined as the probability of having to adopt undesired measures during the

operational phase multiplied by the costs caused by such measures (Haubrich et al., 2001).

This list of measures concerns mainly improvements to the methods applied for deter-

mining transmission capacity. Extensive use of system protection schemes, especially of

automatic control actions following critical line outages, may help to relax transmission

congestions without substantial investments; measures include:

the shedding of preselected generators;

the shedding of preselected loads (industrial);

network splitting to avoid cascading events.

These schemes are implemented primarily to allow increased transmission limits in

addition to reducing the consequences of disturbances or interruptions (Breidablik,

Giæver and Glende, 2003).

20.4.2 Possible reinforcement measures: thermal limit

By taking the thermal current limit as the critical factor, reinforcement measures (apart

from the construction of new lines) can aim either at increasing the current limits of

individual lines and/or the associated equipment, such as breakers, voltage and current

transformers, or at optimising the distribution of load flows in order to de crease the

loading of the critical branches. To increase current limits of individual lines, the

following measures can be applied (Haubrich et al., 2001):

Determine the dynamic current-carrying capacity rating by means of real-time moni-

toring of line tension or sag as well as line current and weather conditions. This

method reduces the probability of line overheating comp ared with the case where

conservative weather con ditions are assumed (wind speed 0.6 m/s; temperature 15



C),

(Douglas, 2003).

Shorten insulators.

Increase the tensile stress of conductors.

Wind Power in Power Systems 443

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 444 – [433–460/28]

17.12.2004 10:43PM

Increase the height of the towers.

Replace underdimensioned substation equipment.

Install conductors with higher loadability, where old supporting structures can be

reused, (Douglas, 2003). Recently developed high-temperature low-sag conductors

may have the same diameter as the original conductors and thus the existing structure

may not have to be reinforced.

If the associated line equipment limits the capacity it can be replaced by equipment with

a higher rating.

The following measures can potentially influence the distribution of load flows:

use of phase-shifting transformers;

use of series capacitors or series reactors to adjust the impedances of the lines;

inclusion of FACTS elements (including DC links) that use power electronics to

control voltage and active – reactive power.

Phase-shifting transformers and series capacitors or series reactors are usually the

preferred options. FACTS are extremly flexible, but also very expensi ve. FACTS

devices may also negatively affect the reliability of the network by diverting power to

weaker parts of the network (Douglas, 2003).

20.4.3 Possible reinforcement measures: voltage stability limit

If voltage limits or voltage stability are the determining factor for the transmission

capacity, additional sources of reactive power (i.e. shunt capacitors, shunt reactors or

FACTS elements) can be installed at critical locations in order to smoothen the steady-

state voltage profile and to increase reserves against the loss of voltage stability. If

voltage instability is caused by power transfer over long distances, series capacitors can

be installed to decrease the impedance of the lines (see Section 20.2.2). The applicability

of the suggested reinforcement measures depends on the individual network topology.

The obvious and most effective measure to increase transmission capacity is to build a

new transmission line. However, this is a time-consuming solution as it takes about five

years as well as being an expensive option. The cost of transmission line is approxi-

mately SEK4 million per kilometre (Arnborg, 2002).

(1)

Over the past few years, as a

result of environmental concern it has become difficult to obtain permission to build

new overhead transmis sion lines.

20.4.4 Converting AC transmission lines to DC for higher transmission

ratings

One way to increase transmission capacity is to convert power lines from high -voltage

AC (HVAC) to high-voltage DC (HVDC). This allows one to increase the power

(1)

As at 7 September 2004, SEK1 ¼e 0.11.

444 Areas with Limited Transmission Capacity

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 445 – [433–460/28]

17.12.2004 10:43PM

transmission rating 2–3 times and reduce transmission losses (Ha

usler, Schlayer and

Fitterer, 1997). The convertion includes changes to the conductor arrangement on the

tower, the insulator assemblies and the configurati on of the conductor bundle. The

actual tower structure or the number of towers does not have to be changed, though.

Depending on the condition of the existing conductors, it is possible to reuse them for

the new HVDC link. In the case of multiple-system lines, some tower designs even allow

a convertion in stages, and transmission can be continued.

An advantage of converting long HVAC lines (longer than 300 km) to HVDC is that

the thermal limit ratin g can be fully used. This particulary increases the avalability of the

double bipole syst ems. In the case of an outage of one HVDC line, the remaining bipole

system can transmit double the power. As mentioned in Section 20.2.2, long HVAC

transmission lines often cannot be loaded to their thermal rating because of voltage

stability problems.

Depending on the condition of the existing system, the cost of converting the line

from HVAC to HVDC can be about 30 % to 50 % lower compared with building a new

transmission line. However, this figure does not include the construction of two con-

verter stations. The disadvantage of a conversion to HVDC is that the power systems on

both ends of the HVDC link become decoupled regarding frequency control. This may

cause problems in power systems where the controllable units are concentrated at one

end of the HVDC link.

20.5 Impact of Wind Generation on Transmission Capacity

With the integration of new generation comes an increased need for additional trans-

mission capacity. The issuses discussed above are relevant to any type of new gener-

ation, not only wind power. However, wind power has some special features that have to

be taken into account when assessing transmission capacity.

First, wind power production has to be evaluated taking into account its low utilisation

time (2000–3000 hours per year), the spacial smoothing effect and the fact that the power

output is a function of the ambient conditions (see Section 20.2.3). After that, wind power

can be treated as any conventional generation when evaluating the thermal limits. Wind

speed measurements from wind farms can even be used for the online estimation of the

current-carrying capacity of short transmission lines (see Section 20.2.1).

The induction generators that are used in wind power applications consume reactive

power. If there is no reactive power compensation, this results in a lagging power factor

at the wind farm connection point. As shown in Figure 20.4, this may decrease the

maximum power transfer from the wind farm to the network, if the limit is defined by

voltage stability considerations. Reactive power compensation of wind turbines is

usually provided by shunt capacitor banks, SVC or AC/DC/AC converters. Reactive

power compensation provided by shunt capacitor banks depends on the voltage at the

connection point and theref ore may not be sufficient for lower voltage. However, if

continuous reactive power compensation is used through AC/DC/AC converters, for

example, wind power does not affect the maximum power transfer if the limit is defined

by voltage stability considerations. Moreover, if at the wind farm connection point a

leading power factor is provided, the maximum power transfer over the considered line

Wind Power in Power Systems 445

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 446 – [433–460/28]

17.12.2004 10:43PM

could be increased, especially if it is acceptable to have a higher voltage at the wi nd farm

connection point (see Section 20.2.2).

During and after faults in the system, the behaviour of wind turbines is different from

that of conventional power plants. Conventional power plants mainly use synchronous

generators that are able to continue to operate during severe voltage transients pro-

duced by transmission system faults. Variable-speed wind turbines are disconnected

from the grid during a fault in order to protect the converter. If a large amount of wind

generation is tripped be cause of a fault, the negative effects of that fault could be

magnified (Gardner et al., 2003). This may, in turn, affect the transmission capacity in

areas with significant amounts of wind power, as a sequence of contingencies would be

considered in the security assessment instead of only one contingency. During a fault,

fixed-speed wind turbines may draw large amounts of reactive power from the system.

Thus, the system may recover much more slowly from the fault (see also Chapter 28).

This could also affect transmission capacity.

There are several reasons why the integrati on of large-scale wind power may have a

particular impact on the methods that are used for determining the available transmis-

sion capacity:

The power output of wind farms depends on wind speed, therefore TSOs should

include wind forecasts in the base case for determining the day-ahead transmission

capacity and also use wind speed statistics in the base case that is used for determining

the NTC twice a year. There may be higher uncertainties associated with prediction

errors regarding the generation distribution and this may result in an increased

transmission reliability margin, which in other word corresponds to a decrease in

transmission capacity.

Compared with conventional generation, for wind farms, less sophisticated models of

generator characteristics are used. This could make simulation results less reliable (i.e.

some TSOs may choose to increase transmission reliability margins to account for that).

Apart from the impact that wind power has on the methods for determining transmis-

sion capacity, its integration also requires greater investment regarding some of the

measures for achieving an increased transmission capacity. It may, for instance, be

significantly more expensive to provide sophisticated protection schemes for wind farms

that are distributed over a certain area than for conventional generation of an equiva-

lent capacity (Gardner et al., 2003). Wind farms are built in remote areas where the grid

reinforcements are more urgent and more exp ensive than in areas close to industrial

loads, where conventional generation is usually situated. Owing to the low utilisation

rate of the wind turbines, the energy produced per megawatt of new transmission is low.

The resulting question to what extent it is economic to increase transmission capacity

for a particular wind farm will be addressed in the next section.

20.6 Alternatives to Grid Reinforcement for the Integration of Wind Power

Transmission capacity should not be increased at any cost. The optimal balance between

extra income and the cost of additional transmission capacity has to be struck. Usually,

446 Areas with Limited Transmission Capacity

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 447 – [433–460/28]

17.12.2004 10:43PM

it is considered not optimal completely to remove a bottleneck. Furthermore, the

increase of trans mission capacit y is time-consuming. Therefore, other measures are

necessary to handle congestion and the related problems and to allow a large-scale

integration of wind power. Such measures include reducing the power production in

conventional generation units, where fast regulation is possible, and/or curtailing excess

wind energy, when the transmission system is congested.

20.6.1 Regulation using existing generation sources

One possibility to integrate wind power with only partial or without any grid reinforce-

ment is to regulate power transmission with power storage devices or with conventional

power sources, if fast regulation of the power output is possible. Examples of this

approach are the use of gas turbines, or hydro power in power systems with large hydro

reservoirs (e.g. as in Sweden and Norway). During high wind power production and

simultaneous congestion problems in the transmission system, hydro power production

at the same end of the bottleneck can be reduced and water can be saved in hydro

reservoirs (see Figure 20.7). Thus, wind power production would not be affected by the

congestion problem and it would not be necessary to curtail the production of wind

energy. The saved water can then be optima lly used by the hydro power plant (HPP)

once the bottleneck is relieved (Matevosyan and So

der, 2003). Redispatching HPPs can,

however, affect the efficiency of the hydro system. Furthermore, in areas where water

inflow and wind speeds are correlated, the capacity of the hydro reservoirs will put

additional limitations on the regulating capability.

The regulation of power transmission by hydro power can be relatively easily accom-

plished if wind farms and HPPs are owned by the same company. In this case, wind

power production forecasts can be included in the production planning of HPPs

(Vogstad, 2000). Forecast errors can be accounted for by frequent updates of the wind

power production forecasts, and hydro power production can be changed accordi ngly

so that transmission limits will not be exceeded. In the case of separate ownership,

congestion problems caused by wind power produ ction will be solved within the frame-

work of the electricity market.

20.6.2 Wind energy spillage

Another option to introduce a certain amount of large-scale wind power without

grid reinforcement is to ’spill’ wind if the transmission system is highly loaded.

Transmission is limited

Other power source

load

Hydropower

wind power

load

Figure 20.7 A congested system with hydro and wind power production on the same side of the

bottleneck

Wind Power in Power Systems 447

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 448 – [433–460/28]

17.12.2004 10:43PM

Methods have been developed for estimating the amounts of spilled energy at

various wind power penetration levels. These methods can be used at the early

stage of a prefeasibility study f or a project if wind speed measurements are already

available. These estimation methods can also be used by TSOs for defining the

technical prerequisites for the connection of a wind farm, or by the authorities for

evaluating large-scale projects.

In the following subsections, we will describe two methods: the first method gives a

simplified estimation based on several extreme assumptions, and the second estimation

is similar to a probabilistic production cost simulation (Booth, 1972). The estimation

methods are compared and the results are presented, using the Swedish transmission

system as an example. The impact of the different assumptions on the results will be

discussed at the end of this section.

20.6.2.1 Simplified method

A straightforward way to estimate the capacity of wind power that can be installed

in an area with limited transmission capability is to compare the present transmis-

sion duration curve (TDC) for the bottleneck and the expected wind power produc-

tion duration curve (WPDC), (see Figure 20.8). The WPDC is obtained by using

wind speed measurements from the studied area. This is converted into power and

scaled up to the respective installed capacity. The smoothing effect within a wind

farm is not included in the considerations, and all wind turbines are assumed to be

100 % available.

0 1000 2000 3000 4000 5000 6000 7000 8000 9000

1000

2000

3000

4000

5000

6000

7000

8000

9000

Duration (hours)

Power (MW)

WPDC+TDC

TDC

WPDC

Figure 20.8 A transmission duration curve (TDC), a wind power production duration curve

(WPDC), their sum (WPDC þ TDC) and the transmission limit (TL); the area below

WPDC þ TDC and above TL corresponds to the amount of wind power that has to be spilled

448 Areas with Limited Transmission Capacity

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 449 – [433–460/28]

17.12.2004 10:43PM

This can be considered an initial estimation, but it will give an idea of the capacity of

wind power that can be installed. The following assumptions are made:

The transmission limits are given and are constant for the studied period.

Only active power flows are considered.

The produced wind power can be consumed on the other side of the bottleneck.

Hourly variations of expected wind power production (calculated from wind speed

measurements) and hourly variations of the actual power transmission from the

studied area are assumed to be simultaneous. This means, for instance, that the

maximum power transmission occurs at the same time as the maxi mum wind power

production.

The last assumption may be unrealistic, but it represents the most extreme case that can

occur.

When wind power production is high, it may be impossible to transfer all the generated

power because of the transmission limits in the network. If the excess power cannot be

stored in hydro reservoirs or by other means it has to be spilled. Given the assumptions

listed above, the spilled energy, W

spill

, can be calculated in the following way:

spill

½P

TDC

ðtÞþP

WPDC

ðtÞP

dt; ð20:3Þ

where P

TDC

(t) and P

WPDC

(t) are the values of power, in megawatts, at time t on the

TDC and WPDC, respectively; P

is the transmission limit in megawatts; t

is the

number of hours during which the transmission limit is exceeded (i.e. when

TDC

(t) þ P

WPDC

(t)  P

The percentage expected spil led wind energy, k

spill

, is calculated as follows:

spill

 100; ð20:4Þ

where W

is the energy generated by the wind farm during the analysed period T

WPDC

ðtÞdt: ð20:5Þ

Figure 20.8 presents the results of this simplified estimation method. The results are

based on the existing TDC, corresponding to the transmission from the northern to the

southern and central parts of Sweden in 2001 (Arnborg, 2002). To obtain the WPDC, we

have used wind speed measurements from Suorva in northern Sweden (Bergstro

llstrand, 2000) and the power curve for a pitch-regulated wind turbine. The calculated

wind power output is scaled up to represent 2000 MW of wind power in northern Sweden.

The effect of the fact that the wind farms are distributed over large area is not included.

From Equation (20.3), the shaded area below WPDC þ TDC and above TL in Figure

20.8 corresponds to the amount of wind energy that has to be spilled during a year.

Wind Power in Power Systems 449

//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 21_CHA20 .3D – 450 – [433–460/28]

17.12.2004 10:43PM

20.6.2.2 Probabilistic estimation methods

Another approach to estimate the total installed capacity of wind power that can be

integrated in areas with bottleneck problems is to calculate the probability of exceeding

the transmission capacity limit. This approach is similar to production cost simulations

(Booth 1972), and the estimat ion has to be based on statistical data. The assumptions

are the same as those for the simplified method, except for the last assumption, which is

changed as follows:

Wind power output and transmitted power are independent variables. This means

that it is equally probable to have high wind power production during times of high

power transmission as during times of low transmission.

In the following subsections, X is the power flow in megawatts that is transmitted

through the bottleneck in addition to the wind power generation, and Y corresponds to

the generated wind power in megawatts.

Discrete probabilistic method

If the estimation method is based on statistical data from measurements, then X and Y

are considered discrete variables.

The distribution function for the transmitted power and the corresponding probabil-

ity mass function are:

ðxÞ¼PðX  xÞ;

ðxÞ¼PðX ¼ xÞ;

where P(X  x) is the probability of transmission X having a value less than or equal to

a level x,andP(X ¼ x) is the probab ility of transmission X being exactly x. Distribution

and probability mass functions can be obtained from long-term measurements of X.

Similarly, a distribution function and probability density function can be expressed

for the wind power output Y:

ðyÞ¼PðY  yÞ;

ðyÞ¼PðY ¼ yÞ:

Using long-term wind speed measurements, the power output Y of the planned wind

farms can be obtained from the power curve of the wind turbines. Then, distribution

and probability mass functions of Y are calculated.

Now we introduce the discrete variable Z,withZ ¼ X þ Y. Z is the desired power

transmission after wind power is installed in the area with the bottleneck problems. Its

probability mass function f

(z) is obtained from a convolution, as follows (Grimmett

and Strizaker, 1992):

ðzÞ¼

ðxÞf

ðz  xÞ¼

ðz  yÞf

ðyÞ: ð20:6Þ

450 Areas with Limited Transmission Capacity