AckermannTh. (ed) Wind Power in Power Systems
Подождите немного. Документ загружается.
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 221 – [197–232/36]
20.12.2004 7:41PM
10.3.3 Wind power forecast models
For forecasting wind power, Eltra uses the WPPT (Wind Power Prediction Tool)
program package developed by the Department of Informatics and Mathematical
Modelling (IMM) at the Technical University of Denmark, in cooperation with Elsam
and Eltra (Nielsen, 1999; see also chapter 17)
WPPT predicts the wind power production in Jutland–Funen using online data cover-
ing a subset of the total population of wind turbines in the area. The Western Danish
Jutland–Funen area is divided into subareas, each covered by a reference wind farm. For
each subarea, local online measurements of climate variables as well as meteorological
forecasts of wind speed and wind direction are used to predict wind power.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 50 100 150 200 250 300
Distance between wind power production sites (km)
Correlation
Onshore–onshore wind
Onshore–offshore wind
Figure 10.19 Correlation in space between estimated power production at wind farms in the
Eltra area, based on 12 months of measurements (15 May 1999 to 15 May 2000)
0
0.2
0.4
0.6
0.8
1
1.2
0 1020304050
Time (hours)
Autocorrelation
Horns Rev
Læsø
Figure 10.20 Correlation in time for estimated wind power production at two Danish offshore
sites, based on 12 months of measurements (from 15 May 1999 to 15 May 2000)
Wind Power in Power Systems 221
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 222 – [197–232/36]
20.12.2004 7:41PM
The power prediction for the reference wind farm of each subarea is subsequently
upscaled to cover all wind turbines in the subarea. Then the predictions for the subareas
are compiled to arrive at a total prediction for the entire Eltra area. Table 10.5 outlines
the forecast statistics for wind power within the Eltra area for the year 2002.
It follows from the statist ics in Table 10.5 that the miscalculation for the year 2002
corresponds to 31 %f[(1095 GWh)/(3478 GWh)] 100g of the total wind energy pro-
duction. The unpredictable wind power causes large problems with keeping the physical
balance and leads to reduced system security. Trading in the Nordic power exchange
(Elspot) must be completed at 12 noon of the day prior to the day of operation. The
daily miscalculation is thus calculated as the sum of absolute hourly deviations between
forecasts of next day’s wind power production and the actual production observed later.
The da ily miscalculations are added up in order to arrive at the yearly miscalculation.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
–8 –6 –4 –2 0 2 4 6 8
Time (hours)
Correlation
Figure 10.21 Correlation between estimated hourly production at Horns Rev and Læsø, based
on 12 months of measurements (from 15 May 1999 to 15 May 2000); positive values on the first
axis indicate the number of hours that the Læsø values were ‘delayed’ compared with the Horns
Rev values
Table 10.5 Wind power statistics within the Eltra area, 2002
Quantity Value
Installed wind power capacity:
start of year (MW) 1780
end of year (MW) 2000
Wind energy:
produced wind energy (GWh) 3478
Miscalculated wind energy (GWh) 1095
Miscalculation (%) 31
Added wind power costs:
Payment of real-time imbalance power (DKr millions) 68
Added costs of wind power (DKr/kWh) 0.02
Note: in mid-2003, 1 Danish Krone (DKr) ¼ 7:5 euros.
222 The Danish Power System
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 223 – [197–232/36]
20.12.2004 7:41PM
Owing to this miscalculation, in 2002 the added costs of handling wind power
amounted to approximately DKr68 million. The added costs correspond to approxi-
mately DKr 0.02 per kWh wind power and are related to the purchase of regulating
power from the real-time imbalance power market [added costs ¼ (DKr68 million)/
3478 million kWh)]. The added cost of DKr68 million per year is a substantial amount,
and a miscalculation of 31 % leaves great potential for improvement.
The primary reason for the relatively poor wind forecasts is low-quality meteoro-
logical data. Eltra is strongly engaged in a number of research and development
activities that aim at improving wind power forecasting.
In this respect, the concept of ‘ensemble forecasting’ looks very promising. The Ensemble
Prediction System (EPS) relies on generating several forecasts instead of just one and aims
at predicting the correct wind speed instead of providing one generally reliable weather
forecast. Initial studies have been conducted in which the meteorological model comes up
with 25 wind forecasts instead of a single forecast and then ‘selects the best one’. These
studies suggest that use of EPS may reduce Eltra’s present misforecasts by approximately
20 %. These 25 forecasts have been generated by combining five different vertical diffusion
schemes with five different condensation schemes, which results in 25 ensemble members.
In 2003–2004, Eltra will test the ‘multi-ensemble concept’ on wind power forecasts.
For the tests, Eltra will set up a dedicated in-house network (cluster) of PCs to test the
method in a real-time setting.
10.3.4 Grid connection
10.3.4.1 Wind farms connected to grids with voltages above 100 kV
Eltra has developed specifications for connecting wind farms, including offshore farms,
to the transmission network (Degn, 2000; see also Chapter 7). These specifications,
which have become a paradigm in Europe, are the counterpart of the gen eral power
station specifications for land-based production facilities.
The most important new requirement is that wind farm s – like any other major
production plant – are not allowed to lose stability or trip during short circuits in the
network that is disconnect ed by primary network protection. In more popular terms, the
turbines must be able to survive a short dead time (about 100 ms) and resume produc-
tion once the fault ha s been disconnected and the voltage starts to return.
The specifications require an increased control capability of wind farms according to
which a wind farm has to be able to reduce its production from full load to a level
between 0 % and 20 % within a few seconds.
The specifications include requirements regarding:
.
power and power control, including startup;
.
frequencies and voltages (e.g. a wind farm has to be able to operate at deviating
frequencies and voltages);
.
the wind farm’s impact on voltage quality;
.
the interaction between power system and wind farm in the case of faults in the power
system;
.
protection and communication.
Wind Power in Power Systems 223
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 224 – [197–232/36]
20.12.2004 7:41PM
10.3.4.2 New wind turbines connected to grids with voltages below 100 kV
The wind turbines that are currently connected to the distribution system were erected in
compliance with DEFU KR-111 (DEFU/DTU, 1998) This recommendation applies to
the distribution undertakings and includes, in addition to frequency trip limits, very few
overall power system requirement s.
Based on the experience from applying Degn (2000) and DEFU KR-111 (DEFU/
DTU, 1998), the two Danish TSOs have developed a new set of specifications for the
connection of wind turbines to grids with voltages be low 100 kV. These specifications
entered into force in July 2004.
The intention of these specifications is to ensure that the wind turbines have the
regulating and dynamic properties that are essential to operate the power system. The
control requirements have been set up to allow an increased penetration of wind power
and to prepare for a possible distributed control of the power system.
Since these wind turbines usually are connected to the distribution system, the
specifications have to balance the local requirements for security of supply (voltage
quality, overvoltage, islanding operation, etc.) with the overall system requirements
(fault ride-through capability, frequency control , etc.).
10.3.5 Modelling of power systems with large-scale wind power production
Today, distributed generation (DG) in the Eltra system is handled as nondispatchable
production. The wind generators produce power according to the wind, and the power
production from small-scale CHP units is governed by the local demand for district
heating. They a re not governed by market demand for power. In addition, when
producing the necessary heat, the CHP units today optimise their profits according to
statutory power tariffs. The problem is that these tariffs do not reflect the current
market value of power.
The present DG in the Eltra area covers about half of the consumption (see Section
10.1), without contributing to regulating power and other ancillary services. Therefore,
the present handling of DG causes large operational problems regarding the physical
power balance. This leads to reduced system security and increased costs of providing
the necessary services from the remaining large central production plants.
The Danish government is preparing a change to the concept of subsidising DG,
including the abandoning of its priority status. This means that small-scale CHP units in
the future will have to produce according to market signals. The change of concept is
expected to start in 2005 for units above 10 MW installed capacity.
10.3.5.1 Wind power and the power market
The variable costs of wind power are only a fraction of the variable costs of thermal
production. In this context, wind power is similar to uncontrollable hydropower (‘run-
of-river’ concepts).
Consequently, large quantities of installed wind power in the system will produce a
downward pressure on the market price of electricity in the day-ahead market (the spot
224 The Danish Power System
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 225 – [197–232/36]
20.12.2004 7:41PM
market) when wind forecasts generate very positive wind power production estimates
for the day ahead.
In order to analyse how wind power affects, among other things, market prices and
competition within the Nord Pool area, Eltra has developed a new model, MARS
(MARket Simulation; Eltra, 2003; Kristoffersen, 2003). MARS is Eltra’s new market
model for the simulation of prices, production, demand and exchanges in the power
market. The model area comprises the Nordic countries (the Nord Pool area) and
Northern Germany.
The ne w feature of this model is that prices, exch anges and so on are calculated on an
hourly basis . The model uses the same principles as Nord Pool, including the division of
the Nordic countries into price areas with price-dependent bids.
The model is designed for wind power, hydropowe r and thermal production, includ-
ing nuclear power. On the demand side, price elasticity is taken into consideration (i.e.
that demand varies according to price). Game theory was used to simulate the produ-
cers’ strategic behaviour (i.e. producers’ options for exercising market power).
MARS uses data reported to Nordel by the TSOs for, among others, carrying out
system analyses with different model tools. The data are then converted to hourly values
in relation to the available information on the distribution of consumption and so on.
The data conversion is based on information from the Nord Pool FTP server and on a
purchased database of production plants in various countries as well as on various other
sources.
As an example of how the MARS model is used, Figure 10.22 shows simulated spot
prices in the Eltra area for a week in the winter of 2005. The prices are computed for two
scenarios; (1) ‘perfect competition’ and (2) when the large producers in the Nordic
market exercise market power. It can be seen that the prices are low when there is much
wind power, and that prices are high when there is little wind power.
Price (DKr/MWh)
100
150
200
250
300
350
0 24 48 72 96 120 144 168
Time (hours)
0
600
1200
1800
2400
3000
Wind Power (MWh)
Perfect competition
Market power
Wind power
Figure 10.22 Estimated prices in Western Denmark for a week in the winter of 2005 with
incidents of high wind power production: perfect competition and market power scenarios
Wind Power in Power Systems 225
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 226 – [197–232/36]
20.12.2004 7:41PM
It is also evident that, generally, the potential for exercising market power (i.e.
pushing the market price in Western Denmark up through strategic bidding into Nord
Pool) is highest during hours with large wind power production in the Eltra area. The
reason is that, in such situations, producers can raise the price in Western Denmark to
match the price level of its neighbours and hence reap the extra profit.
10.3.6 Wind power and system analysis
10.3.6.1 The Eltra SIVAEL model, including stochastic wind power descri ption
In view of the large share of wind power in the Eltra system, Eltra has updated the SIVAEL
simulation model to include the stochastic behaviour of wind power (Pedersen, 2003).
The standard SIVAEL model (Eriksen, 1993, 2001; Pedersen, 1990; Ravn, 1992) solves
the problem of how to schedule power production in the best way. It generates the
scheduling of power production units, heat production units and CHP units on a weekly
basis. The constraints require the power and heat demand and the demand for spinning
reserves to be met every hour of the week. The scheduling includes unit commitment
(start–stop of units) and load dispatching (unit production rates of power and heat).
SIVAEL handles power and heat generation plants, heat storage systems, wind
turbines, foreign exchange, immersion heaters, heat pumps and electricity storage. The
scheduling is carried out with the objective of minimising total variable costs, which
include variable operational and maintenance costs and startup costs.
By its nature, wind power is largely unpredictable (see Sectio n 10.2.6 and Section
10.3.3). The new and improved SIVAEL model simulates wind power prediction errors
as a stochastic process. The parameters in the stochastic model description have been
estimated from observations of wind power forecast errors throughout the year within
the Eltra system. The work has been carried out in cooperat ion with IMM at the
Technical University of Denmark (DTU).
As a new feature, the updated SIVAEL model can simulate the need for upward and
downward regulation of the system, due to changes in wind power in the real-time hour
of operation. A change in wind power is by far the most important source of unforeseen
regulation in the Eltra system.
With the model it is possible to compare for each hour the need for regulating the
system with the available resources, including:
.
regulation of primary central units;
.
regulation via transmission lines to neighbouring areas;
.
regulation of local small-scale CHP units.
The present version does not include the formal optimisation of the use of the regulating
resources generated by local CHP units . The optimisation is limited to the two other
resources: primary central units and the trans mission system. However, the model may
be expanded and used for an evaluation of the regulating potential of local CHP units.
Figure 10.23 shows hourly values of simulated wind power prediction errors through-
out a year, in the future. The errors are measured as a percentage of the hourly wind
power forecasts (with positive figures when produ ction exceeds forecast).
226 The Danish Power System
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 227 – [197–232/36]
20.12.2004 7:41PM
10.3.6.2 Active use of small-scale combined heat and power for regulation purposes
As an example of how the SIVAEL model can be applied, Figures 10.24(a) and 10.24(b)
present SIVAEL simulations of the Eltra system’s demand for regulation in 2005, and
how that demand is covered. It is assumed that the demand for regu lation is determined
by the magnitude of a faulty wind forecast. The 2005 wind forecasts are assumed to be
of a similar quality to today’s predictions.
–100.00
–80.00
–60.00
–40.00
–20.00
0.00
20.00
40.00
60.00
80.00
100.00
1 Jan. 2 Mar. 1 May 30 Jun. 29 Aug. 28 Oct. 27 Dec.
Prediction error (%) Prediction error (sorted)
Date (2005)
Error (%)
Figure 10.23 Simulation of wind power prediction errors throughout a year, Eltra 2005
–1500
(a)
–1000
–500
0
500
1000
1500
2000
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Time (hours)
Demand (MWh/h)
Total demand for regulation
Demand after utilising primary units
Upward regulation
Downward regulation
Figure 10.24 Regulating demand and lack of regulating power in the Eltra area, 2005, for the
case of high prices in the Nordic market; small-scale combined heat and power (CHP) facilities
run according to (a) a three-rate tariff model (as at present) and (b) market signals
Wind Power in Power Systems 227
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 228 – [197–232/36]
20.12.2004 7:41PM
Simulations were carried out for two scenarios of operating small-scale CHP units:
.
Scenario 1: the small-scale CHP units show the same operational patterns as today,
with their operation determined by current tariffs and heat requirement (company-
specific economic dispatch).
.
Scenario 2: the present prioritising has been abandoned, and the operational patterns
of the small-scale units follow market signals (socioeconomic load dispatch).
Each of the above scenarios has been analysed relative to a low-price situation, with
market prices averaging DKr120 per MWh a year, and a high-price situation, with
prices averaging DKr220 per MWh. As an additional feature, the prices have been
assumed to vary over the days of the year and the hours of the day.
The results of the high-price situation are compiled in Figure 10.24, which shows that,
in the Eltra area, small-scale CHP production has a major potential that can be used for
wind power regulation purposes. Activating this resource by opening the market to small-
scale CHP production could reduce Eltra’s regulation difficulties arising from the unpre-
dictability of wind power. Similar conclusions can be drawn for the low-price situation.
10.3.7 Case study CO
2
reductions according to the Kyoto Protocol
Under the Kyoto Protocol, Denmark has committed itself to reducing carbon dioxide
(CO
2
) emissions by 21 % compared with 1990. This is equivalent to a total national
emission of 55 million tonnes of CO
2
by 2012.
As a result of the Kyoto target, emissions from Danish heat and power production are
expected to be less than 23 million tonnes, which corresponds to approximately 14
million tonnes for the Eltra area (Western Denmark).
–1500
(b)
–1000
–500
0
500
1000
1500
2000
500
Time (hours)
Demand (MWh/h)
0 1000 1500 2000 2500 3000 3500 4000 4500 5000
Total demand for regulation
Demand after utilising primary units and local CHP
Upward regulation
Downward regulation
Figure 10.24 (continued)
228 The Danish Power System
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 229 – [197–232/36]
20.12.2004 7:41PM
As part of the efforts to fulfil the Danish Kyoto targets, Eltra regularly studies the
technical and emission-related consequences of expanding wind power in Western
Denmark. The following sections provide a summary of the assumptions and the
findings that Eltra has reached so far. It will serve as a case study of modelling a power
system with large-sca le wind power production. Eltra’s SIVAEL model was used to
perform the simulations (see Section 10.3.6).
10.3.7.1 Assumptions
The reference scenario assumes that by 2012 there will be 2500 MW land-based turbines
and 760 MW offshore turbines in the Eltra area. Furthermore, the offshore expansion is
assumed to take the form of four 150 MW offshore wind farms, in addition to the
existing offshore wi nd farm at Horns Rev.
The Kyoto alternative assumes that by 2012 the Eltra system will include another
1000 MW of offshore wind power at Horns Rev compared with the reference scenario. It
is also assumed that immersion heaters will be installed in the district heating systems of
the central and natural-gas-fired local CHP areas. Furthermore, it is assumed that some
of the local CHP facilities will inco rporate air coolers, which would make it possible to
generate electricity without generat ing district heat at the same time.
As a general condition, it is assumed that the local (small-scale) CHP facilities will be
operated under market conditions and not, as is the case today (mid-2003), according to
fixed tariffs, which do not reflect the current market value of power. This means that the
local CH P facilities will not be in service in low-price situations. When the price of
electricity is low, peak load boilers and immersion heaters will be used to generate the
required heat. Finally, a general CO
2
shadow price of DKr120 per tonne of CO
2
is
assumed.
The an alyses have been conducted for three different market price scenarios:
.
Scenario A: a low-price scenario (average annual electricity price of DKr120 per
MWh in the Nordic countries and DKr170 per MWh in Germany).
.
Scenario B: a high-price scenario (average annual electricity price of DKr220 per
MWh in the Nordic countries and in Germany).
.
Scenario C: an extremely-high-p rice scenario (average annual electricity price of
DKr600 per MWh in the Nordic countries and DKr220 per MWh in Germany).
10.3.7.2 Results
Figures 10.25 and 10.26 show selected results of the distribution of power production
and emissions for the three price scenarios. They illustrate that an additional 1000 MW
offshore wind power would generate 4.2 TWh of power. In the low-price scenario net
imports would be reduced by approximately 3.5 TWh, and there would be a similar
increase of 2.6 TWh for net exports in the high-price scenario. In addition, another
1000 MW of wind power would lead to minor changes in central and local production.
Figure 10.26 shows that the changed production pattern resulting from an additional
1000 MW wind power would lead to only modest changes in CO
2
emissions in Western
Wind Power in Power Systems 229
//INTEGRAS/KCG/P AGIN ATION/ WILEY /WPS /FINALS_1 4-12- 04/0470855088_ 11_CHA10 .3D – 230 – [197–232/36]
20.12.2004 7:41PM
–20.0
–10.0
0.0
10.0
20.0
30.0
40.0
50.0
Ref. Alt. Alt. Alt.Ref. Ref.
Power Production (TWh)
8.7
–12.7
7.2
20.5
12.9
–4.0
5.0
11.7
8.7
–1.4
5.4
12.1
12.9
6.9
2.7
6.5
8.7
10.4
2.9
6.8
Wind power Net imports Small-scale CHP Primary production
Low prices High prices Extremely high prices
12.9
–15.0
19.8
6.8
Figure 10.25 Distribution of power production within the Eltra area in the reference case (Ref.)
and in the alternative case (Alt.), with an additional 1000 MW of offshore wind power installed by
2012, for three different power price scenarios
18.9
19.6
12.4
12.9
7.4
7.7
–1.1
–1.2
–1.6
–0.9
–1.0
–1.4
–5.0
0.0
5.0
10.0
15.0
20.0
25.0
Ref. Ref. Ref.Alt. Alt. Alt.
Emission of CO
2
(millions of tonnes)
CO
2
emissions
Import–export correction 1 Import–export correction 2
Low prices High prices
Extremely high prices
Figure 10.26 Emissions of carbon dioxide (CO
2
) within the Eltra area in the reference case (Ref.)
and in the alternative case (Alt.), with an additional 1000 MW of offshore wind power installed by
2012, for three different power price scenarios. Note: foreign CO
2
emission reductions as a result
of an additional 1000 MW of Danish wind power capacity are calculated by two methods of
correction (see text)
230 The Danish Power System