2018 4th International Conference on Green Technology and Sust.docx

2018 4th International Conference on Green Technology and
Sustainable Development (GTSD)
130
�
Abstract - The Vietnamese government have plan to develop the
wind farms with the expected capacity of 6 GW by 2030. With
the
high penetration of wind power into power system, wind power
forecasting is essentially needed for a power generation
balancing in power system operation and electricity market.
However, such a tool is currently not available in Vietnamese
wind farms as well as electricity market. Therefore, a short-term
wind power forecasting tool for 24 hours has been created to fill
in this gap, using artificial neural network technique. The neural
network has been trained with past data recorded from 2015 to
2017 at Tuy Phong wind farm in Binh Thuan province of Viet
Nam. It has been tested for wind power prediction with the
input
data from hourly weather forecast for the same wind farm. The
tool can be used for short-term wind power forecasting in
Vietnamese power system in a foreseeable future.
Keywords: power system; wind farm; wind power forecasting;
neural network; electricity market.
I. NECESITY OF WIND POWER FORECASTING

Today, the integration of wind power into the existing
grid is a big issue in power system operation. For the system
operators, power generation curve of wind turbines is a
necessary information in the power sources balancing. From
the dispatchers’ point of view, wind power forecast errors
will impact the system net imbalances when the share of
wind power increases, and more accurate forecasts mean less
regulating capacity will be activated from the real time
electricity market [1]. In the deregulated market, day-ahead
electricity spot prices are also affected by day-ahead wind
power forecasting [2]. Wind power forecasting is also
essential in reducing the power curtailment, supporting the
ancillary service. However, due to uncertainty of wind speed
and weather factors, the wind power is not easy to predict.
In recent years, many wind power forecasting methods
have been proposed. In [3], a review of different approaches
for short-term wind power forecasting has been introduced,
including statistical and physical methods with different
models such as WPMS, WPPT, Prediktor, Zephyr, WPFS,
ANEMOS, ARMINES, Ewind, Sipreolico. In [4], [5], the
methods, models of wind power forecasting and its impact on
*Research supported by Gesellschaft fuer Internationale
Zusammenarbeit GmbH (GIZ).
D. T. Viet is with the University of Danang, Vietnam (email:
[email protected]).
V. V. Phuong is with the University of Danang, Vietnam (email:
[email protected]).
D. M. Quan is with the University of Danang, Vietnam (email:
[email protected]).

A. Kies is with the Frankfurt Institute for Advanced Studies,
Germany
(email: [email protected] uni-frankfurt.de).
B. U. Schyska is with the Carl von Ossietzky Universität
Oldenburg,
Germany (email: [email protected]).
Y. K. Wu is with the National Chung-Cheng University, Taiwan
(email:
[email protected]).
the electricity market and power systems have been
presented. The practice and experience of short-term wind
power forecasting accuracy and uncertainty in Finland has
also been investigated [1].
In general, the equation for wind power P (W) of each
wind turbine is given by the formula (1):
P = (1/2)ρ×A×Cp×Ng×Nb×V
3 (1)
where ρ: air density (kg/m3), A: rotor swept area (m2), Cp:
coefficient of performance, V: wind speed (m/s), Ng:
generator efficiency, Nb: gear box bearing efficiency [6].
Unfortunately, many multiplication factors in the formula
(1) are uncertain. It leads to uncertainty in relationship
between wind speed and wind power of each wind turbine
[7].
II. WIND POWER FORECASTING IN VIETNAM
A. Wind power in Vietnam

Vietnam is considered to have high potential for wind
energy. The wind energy potential of Vietnam is shown in
Table 1, Fig. 1 and Fig. 2 [8]:
TABLE 1. WIND ENERGY POTENTIAL OF VIET NAM AT
80 M ABOVE
GROUND LEVEL
Average
wind
speed
(m/s)
<4 4-5 5-6 6-7 7-8 8-9 >9
Area
(km2)
95,916 70,868 40,473 2,435 220 20 1
Area
(%)
45.7 33.8 19.3 1.2 0.1 0.01 0
Potential
(MW)
956,161 708,678 404,732 24,351 2,202 200 10
The development of wind power has been paid attention
by both the Vietnamese government and investors. The
national renewable energy development strategy by 2030,
which was approved by the Vietnamese Prime Minister,

emphasizes the role of wind power in particular. Expectations
about installed wind power capacity are 800 MW in 2020;
2,000 MW in 2025 and around 6,000 MW by 2030 [9], [10].
By 2017, 160 MW of wind power capacity has been
installed, some large wind farms with capacity and year of
beginning operation are listed - Tuy Phong: 30 MW (2009);
Bac Lieu: 16 MW (2013), and 99.2 MW (2016); Phu Lac: 24
MW (2016); Phu Quy: 6 MW (2013) [8].
A Short-Term Wind Power Forecasting Tool for Vietnamese
Wind
Farms and Electricity Market*
Dinh Thanh Viet, Vo Van Phuong, Minh Quan Duong,
Alexander Kies,
Bruno U. Schyska and Yuan Kang Wu
978-1-5386-5126-1/18/$31.00 ©2018 IEEE
131
Figure 1. Wind resource map of Vietnam at the height of 80 m
Figure 2. Representative wind profile for the three regions in
Vietnam

B. Wind power forecasting in Vietnam
At present, there is no effective tool for predicting wind
power in Vietnam. With the increasing integration of wind
energy into the Vietnamese power system, the projected
capacity of wind power plays an important role in supporting
the optimal operation of wind power plants as well as the
electricity market.
The forecast error of the whole wind farm will be much
affected by the forecast errors from all wind turbines as a
sum. For the electricity market operators, the predicted power
of the whole wind farm at the point of coupling into the
power grid is needed, rather than the sum of predicted powers
of all turbines. In this paper, the approach of wind power
forecasting for the whole wind farm will be investigated.
III. SHORT-TERM WIND POWER FORECASTING
USING NEURAL NETWORK
A. Neural network
A neural network is a multi-input, multi-output system,
consisting of an input layer, one or two hidden layers and an
output layer. Each class uses a number of neurons, and each
neuron in a layer is connected to neurons in the adjacent
layers with different weights. The architecture of the typical
neural network is shown in Fig. 3 [11], [18].
Figure 3. Neural network structure
where input X = (x1, x2, ..., xd), output O = (o1, o2, ..., on).
The
signal is fed into the input layer, passing through the hidden

layer and to the output layer. In a neural network, each
neuron (except neuron at the input layer) receives and
processes stimuli (inputs) from other neurons. Each input is
first multiplied by the corresponding weight, then the
resulting products are added to produce a weighted sum,
which is passed through a neuron activation function to
produce the output of the neuron [11], [12].
B. Feedforward neural network
A feedforward neural network usually has one or more
hidden layers of sigmoid neurons followed by a linear
neurons output layer. The paper uses the model of a
feedforward neural network as described in Fig. 4. The input
layer consists of 3 neurons of historical wind speed,
temperature and wind power. The neural network has 20
neurons in the hidden layer and 01 neuron in the output layer.
The neural network may be used as a general function
approximator. With enough neurons in the hidden layer, any
function with a specific number of discontinuities arbitrarily
can be approximated well. The algorithm for building the
neural network - wind power forecasting model is shown in
132
Fig. 5. The historical data from wind farm, including wind
power, wind speed and temperature, are loaded into the
program and stored as a matrix of forecasting variable.

Because of different operation scenarios, the historical data
may not correctly reflect the relationship between wind
power to wind speed and temperature. In few cases, historical
data in practice may show negative values of wind power
during generator starting time. Therefore, a data
preprocessing is needed for reducing forecasting error. The
historical data is used for training the forecasting neural
network.
Figure 4. Feedforward neural network.
C. Wind power forecasting model
From the formula (1), it is obvious that generating power
of each wind turbine depends largely on the wind speed. The
temperature of environment is chosen as second input,
affecting on the output power [13], [14], sum of output
powers from the turbines serves as another input for neural
network wind power forecasting model (Fig. 6). The input
data is used to forecast the generating power of the wind
farm. This model is designed for Vietnamese wind farms’
power forecasting (short name: VWPF).
IV. CASE STUDY: WIND POWER FORECASTING FOR
TUY PHONG WIND FARM
A. Simulation data
Tuy Phong is the first large-scale wind farm in Vietnam
with a total capacity of 30MW, including 20 turbines of
Fuhrländer, each turbine has a height of 85m, a blade
diameter of 77m, and capacity of 1.5MW. The research is
based on real data on wind speed and wind power production
at Tuy Phong wind farm for 3 years. Collected data from
January 1, 2015 to December 31, 2017 was used for wind

power forecasting.
The data in the forecasting model VWPF includes wind
speed, environmental temperature and wind power, which are
collected every hour, represented by 24 lines per day.
Example of data on October 7, 2017 for wind power
forecasting is shown in Table 2.
The data set is divided into two data subsets:
� Data subset 1 from 01/01/2015 to 30/10/2017 is used
to train the neural network. This is a database with
data collected in 24,816 hours, almost 3 years, large
enough for the neural network training purpose.
� Data subset 2 from 01/11/2017 to 31/12/2017 is used
to compare the forecast results with the collected
actual data for the error evaluation purpose.
Figure 5. Algorithm for building the neural network - wind
power
forecasting model VWPF.
Figure 6. Wind power forecasting model VWPF
133

TABLE 2. EXAMPLE OF DATA ON OCTOBER 7, 2017
Date Hour
Tempera-
ture ( )
Wind
speed
(m/s)
Wind
power (W)
07/10/2017 1 24 3.56 272,299
07/10/2017 2 24 3.51 268,449
07/10/2017 3 24 3.52 266,840
07/10/2017 4 25 3.93 596,442
07/10/2017 5 25 3.9 572,175
07/10/2017 6 27 4.05 767,869
07/10/2017 7 27 3.96 772,839
07/10/2017 8 32 5.54 2,067,089
07/10/2017 9 32 5.47 2,137,362
07/10/2017 10 32 5.35 2,071,529
07/10/2017 11 32 5.4 2,044,341

07/10/2017 12 32 5.28 2,237,072
07/10/2017 13 32 5.25 2,105,929
07/10/2017 14 32 5.34 2,143,538
07/10/2017 15 30 4.68 1,683,934
07/10/2017 16 29 4.04 767,723
07/10/2017 17 24 4.53 1,632,532
07/10/2017 18 24 5.29 2,165,775
07/10/2017 19 24 5.77 2,943,629
07/10/2017 20 24 5.89 2,947,676
07/10/2017 21 24 6.13 3,200,192
07/10/2017 22 24 5.98 3,200,290
07/10/2017 23 25 5.75 3,084,564
08/10/2017 0 25 5.73 3,065,696
B. Error evaluation
The error measures - Mean Absolute Percent Error
(MAPE), Mean Absolute Error (MAE) and Root Mean
Square Error (RMSE) are used to evaluate the accuracy of the
forecasting model [15], [16].
The MAPE represents the accuracy of the model as a
percentage of the error, calculated according to formula (4):

(4)
where:
Preal: actual power output of the wind power plant,
Ppred: generation power according to forecasting model,
N: number of forecasting data.
The MAE shows the accuracy of the model in the same
unit of measure as the predicted data. This index is used to
evaluate the margin of error and is calculated according to the
formula (5):
(5)
In order to evaluate MAE in percentage for comparison
between different models, we can use Normalized Mean
Absolute Error (NMAE) (6):
(6)
where Pinst is the wind farm installed capacity.
The RMSE is the standard deviation of the prediction
errors (residuals). This is also a frequently used measure of
the differences between values forecasted by a model and the
values actually observed. The RMSE is calculated by the
formula (7):
(7)
Similarly, we can use Normalized Root Mean Square
Error (NRMSE) in percentage (8):

(8)
C. Result
By putting Tuy Phong wind farm data into the model and
implementing neural network training, the forecasted model
has been received and shown in Fig. 7, where the blue line
represents the predicted wind power, the red line represents
the actual generated wind power from the data subset 1. Fig.
7 shows a part of the snapshots for series of predicted time,
the total number of snapshots is 23,856 hours (from 1/1/2015
to 20/9/2017).
The forecasting model VWPF is validated with data from
any date in the data subset 2 for 92 days during period from
01/10/2017 to 31/12/2017. The real data of previous day is
updated and included into the training database for the next
day wind power forecast. As an example, the forecast on
November 26, 2017 (the lowest wind speed was 8.39 m/s and
the highest was 15.44 m/s with the steady change of wind
speed throughout the day) is shown in Fig. 8. The error in the
graph represents difference between real wind power and
predicted wind power. The measures for the accuracy of the
forecast result on November 26, 2017 are: MAPE=4.72%;
NMAE=4.66%; NRMSE=5.66%.
Forecast results for December 17, 2017 with the speed
changes from 6.15 m/s to 14.47 m/s are shown in Fig. 9. The
measures for the accuracy of the forecast result on
17/12/2017 are: MAPE=5.44%; NMAE=4.74%;
NRMSE=6.24%.
The wind power day-ahead forecast results in one week
from 25/12/2017 to 31/12/2017 and relevant forecast error
are shown in Fig.10. The forecast error or difference between

the observed and the forecast wind power for one week from
25/12/2017 to 31/12/2017 is evaluated by the following
values: MAPE=8.78%; NMAE=4.14%; NRMSE=4.58%
134
Based on the results of the forecasted wind power from
VWPF, we have a practical range of errors as in Table 3:
TABLE 3. RANGE OF ERRORS FROM PRACTICAL
TESTING
WITH THE MODEL VWPF
No MAPE (%) NMAE (%) NRMSE (%)
1 8.05 5.52 7.28
2 4.72 4.66 5.66
3 5.13 5.13 6.20
4 10.06 5.98 7.32
5 7.03 5.59 7.70
6 6.07 4.07 4.09
7 5.44 4.74 6.24

8 9.64 5.95 8.00
9 5.54 5.94 7.70
Average 6.85 5.29 6.69
Figure 7. Wind power forecasting result after neural network
training.
Figure 8. Forecasted versus actual wind power on 26/11/2017
Figure 9. Forecasted versus actual wind power on 17/12/2017
Summary of error range from Table 3: MAPE = 4.72-
10.06%, NMAE = 4.07-5.98%, NRMSE = 4.09-8.00%.
135
Figure 10. Forecasted versus actual wind power during one
week from
25/12/2017 to 31/12/2017
TABLE 4. COMPARISON THE FORECAST ERROR INDICES
OF
THE PROPOSED MODEL (VWPF) WITH SOME OTHER

MODELS
Forecasting Model
Error indices
MAPE
(%)
NMAE
(%)
NRMSE
(%)
Persistence 14.43 6.18 7.99
BPNN 14.35 5.98 7.53
RBFNN 12.73 5.94 7.40
ANFIS 14.92 6.24 8.03
NNPSO 11.51 5.35 6.59
WT+BPNN 12.19 5.77 7.18
WT+RBFNN 11.18 5.62 6.95
WT+ANFIS 12.58 5.86 7.67
WT+NNPSO 8.19 4.86 6.28
VWPF 6.85 5.29 6.69
Table 4 showed comparison between the forecast errors
of the proposed model with some other published models
[17]. The error indices in different seasons [17] were
recalculated as the average values. From Table 4, we find that
the error indices between of the VWPF model is relatively
smaller in comparison with most of the published wind
power forecasting models. It proves that the VWPF model
provides reliable forecasting results.

V. CONCLUSION
In the paper, a model of the wind power forecasting
(VWPF) is developed for this need in Vietnam. The power
system operators are usually interested in the forecasting of
the whole wind farm’s power, which is generated into the
power system, rather than forecast power of each wind
turbine. It shows advantage and effectiveness of the
developed model in power prediction for the whole wind
farm, which is well appropriate for dispatcher working as
well as electricity market operator. The neural network
prediction model can be used for short-time wind power
forecasting (hour-ahead, day-ahead, and week-ahead). The
forecasting model has been applied for estimating the wind
power output of the Tuy Phong wind power plant in Binh
Thuan province, Vietnam. The predicted results were
evaluated with the average forecast error indices
MAPE=6.85%, NMAE=5.29%, NRMSE=6.69%. The
forecast error indices, showing the high accuracy of the
model, are relatively smaller in comparison with most of
similar research models (Table 4). Application of artificial
intelligence technique at the connected point of the wind
farm to the power grid proved effectiveness of this approach.
This wind power forecasting tool can be applied not only for
Tuy Phong wind farm, but also for the others in Vietnam.
ACKNOWLEDGMENT
This work is part of the R&D Project “Analysis of the
Large Scale Integration of Renewable Power into the Future
Vietnamese Power System”, financed by Gesellschaft fuer
Internationale Zusammenarbeit GmbH (GIZ, 2016-2018).
REFERENCES

[1] H. Holttinnen, J. Miettinen, S. Sillanpää “Wind power
forecasting
accuracy and uncertainty in Finland”, VTT Technology 95-320,
2013.
[2] T. Jónsson,; P. Pinson, H. Madsen, “On the market impact of
wind
energy forecasts”, Energy Economics, Volume 32, Issue 2,
March
2010, pp. 313-320.
[3] X. Wang, P. Guo, X. Huang, “A Review of Wind Power
Forecasting
Models”, Energy Procedia, vol. 12, pp. 770 – 778, 2011.
[4] G. Kariniotakis, Renewable Energy Forecasting: From
Models to
Applications, 1st ed., Woodhead Publishing, 2011.
[5] G. Giebel, R. Brownsword, G. Kariniotakis, M. Denhard, C.
Draxl,
“The State of the Art in Short-Term Prediction of Wind Power:
A
Literature Overview”, 2nd ed., 2011.
[6] A. Sarkar, D. K. Behera, “Wind Turbine Blade Efficiency
and Power
Calculation with Electrical Analogy”, International Journal of
Scientific and Research Publications, vol. 2, Issue 2, February
2012.
[7] D.M. Quan, E. Ogliari, F. Grimaccia, S. Leva, M. Mussetta,
“Hybrid
model for hourly forecast of photovoltaic and wind power”,
2013

IEEE International Conference on Fuzzy Systems, p.p 1-6,
2013.
[8] N. Q. Khanh, “Analysis of future generation capacity
scenarios for
Vietnam”, Green Innovation and Development Centre
(GreenID),
Vietnam, 2017.
[9] The Vietnamese Prime Minister, “Approving the
development
strategy of renewable energy of Vietnam by 2030 with a vision
to
2050”, Decision No. 2068/QD-TTg dated November 25, 2015.
[10] A. Kies, B. Schyska, D. T. Viet, L. Bremen, D. Heinemann,
S.
Schramm, “Large-Scale Integration of Renewable Power
Sources into
the Vietnamese Power System”, Energy Procedia, vol. 125, pp.
207–
213, 2017.
[11] S. Samarasingh, Neural Networks for Applied Sciences and
Engineering: From Fundamentals to Complex Pattern
Recognition,
CRC Oress, Taylor & Francis Group, Boca Raton, 2007.
[12] The MathWorks Inc, Neural Network Toolbox User’s
Guide, 2014.
[13] A. Singh ; K. Gurtej ; G. Jain ; F. Nayyar ; M. M.
Tripathi, Short
term wind speed and power forecasting in Indian and UK wind
power
farms, 2016 IEEE 7th Power India International Conference

(PIICON),
[14] Salih Mohammed Salih, Mohammed Qasim Taha,
Mohammed K.
Alawsaj, “Performance analysis of wind turbine systems under
different parameters effect”, International Journal of Energy
and
Environment, vol. 3, Issue 6, pp.895-904, 2012.
[15] D. B. Alencar, C. M. Affonso, R. C. L. Oliveira, J. L. M.
Rodríguez, J.
C. Leite and J. C. R. Filho, “Different Models for Forecasting
Wind
Power Generation: Case Study”, Energies, 2017.
[16] X. Zhao, S. Wang, T. Li, “Review of Evaluation Criteria
and Main
Methods of Wind Power Forecasting”, Energy Procedia, vol. 12,
pp.
761 – 769, 2011.
[17] P. Mandala, H. Zareipourb, W. D. Rosehart, Forecasting
Aggregated
Wind Power Pro duction of Multiple Wind Farms Using Hybrid
Wavelet-PSO-NNs, International Journal of Energy Research,
Vol.38, Issue13, pp. 1654-1666, 2014.
[18] S. Haykin, Neural Networks and Learning Machines, 3rd
ed., Pearson
Education Inc, 2009.
IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL.
3, NO. 4, OCTOBER 2012 809

Comparison of Wind Energy Support Policy
and Electricity Market Design in Europe,
the United States, and Australia
Néstor Aparicio, Member, IEEE, Iain MacGill, Member, IEEE,
Juan Rivier Abbad, Member, IEEE, and
Hector Beltran
Abstract—This paper is intended to fill a gap in the current lit-
erature comparing and contrasting the experience of a number
of
Europeancountries,U.S. states, andAustraliawithregard towind
energysupportpolicyandelectricitymarketdesign.Aswindpene-
trations increase, thenatureof thesearrangementsbecomesan in-
creasingly importantdeterminantofhoweffectivelyandefficiently
thisgeneration is integrated into theelectricity industry.Thejuris-
dictions considered in this paper exhibit a range of wind support
policy measures from feed-in tariffs to green certificates, and
elec-
tricity industry arrangements including vertically integrated
utili-
ties,bilateral tradingwithnetpools,aswellasgrosswholesalepool
markets. We consider the challenges that various countries and
states have faced as wind generation expanded and how they
have
responded. Findings include the limitations of traditional feed-
in
tariffsathigherwindpenetrationsbecausetheyshieldwindproject
developersandoperators fromthe implicationsof theirgeneration
on wider electricity market operation. With regard to market de-
sign, wind forecasting and predispatch requirements are particu-
larly important for forward markets, whereas the formal
involve-
ment of wind in scheduling and ancillary services (balancing
and

contingencies) is key for real-time markets.
Index Terms—Balancing markets, electricity market design, re-
newable energy policy, wind energy.
I. INTRODUCTION
P OLICY measures to support greater wind energy
havehadademonstrated impacton itsdevelopment indifferent
jurisdictions around the world. Experience to date suggests
that feed-in tariff (FIT) policies have been the most successful
approach in rapidly expanding wind generation capacity, as
demonstrated incountries includingDenmarkandSpain,which
now have world leading wind energy penetrations [1]. This ap-
proach, however, may cause increasing integration challenges
for the electricity industry as wind penetrations continue to
Manuscript received August 30, 2011; revised June 08, 2012;
accepted July
06, 2012. Date of publication September 10, 2012; date of
current version
September 14, 2012. This work was supported in part by the
Universitat
Jaume I under Grant P1·1A2008-11. N. Aparicio’s research visit
to the Centre
of Energy and Environmental Markets, which kindly offered
him a visiting
position, was supported by the Universitat Jaume I under Grant
E-2008-06.
N. Aparicio and H. Beltran are with the Area of Electrical
Engineering, Uni-
versitat Jaume I, 12071 Castelló de la Plana, Spain (e-mail:
apari[email protected]).
I. MacGill is with the School of Electrical Engineering and
Telecommuni-

cations and Centre for Energy and Environmental Markets,
University of New
South Wales, Sydney 2052, Australia (e-mail:
[email protected]).
J. Rivier Abbad is with Iberdrola Renovables, 28033 Madrid,
Spain (e-mail:
[email protected]).
Color versions of oneormore of thefigures in this paper are
available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TSTE.2012.2208771
rise. The value of electricity within a power system varies over
time, by location and subject to uncertainties reflecting, in ag-
gregate, the changing costs and benefits of all generations and
end-users. There have been worldwide moves over the last two
decades to restructure electricity industries so that generators
and end-users see price signals that more appropriately reflect
these underlying industry economics. In their simplest form,
FIT schemes can effectively shield project developers from
such energy market signals through a fixed payment for each
MWh of renewable generation independent of the value it ac-
tually provides for the industry at that time and location within
the network [2]. Simplified tendering processes awarded to
projects on the basis of lowest required government payments
per MWh of renewable generation, which were adopted in
countries such as Ireland and China, can have similar impacts.
Other policy approaches such as renewable electricity pro-
duction taxcreditsas seen in theU.S., and tradablegreencertifi-
cates as seen in a number of European countries and Australia,
provide another approach for supporting wind energy. By com-
parison, these can ensure that wind farm developers and opera-

tors are still incentivizedbyelectricitymarket “signals” tomax-
imize overall industry value.
The last few years has seen important developments in a
number of countries that can help us better understand these
issues. For example, Denmark and Spain have moved from
a conventional FIT to a tariff premium above the electricity
market price, the latter with additional arrangements that cap
potential incomes to wind generators. The UK Renewables
Obligation scheme now appears to be driving greater industry
development, especially in offshore projects. The U.S. Federal
production tax credits and state-based renewableportfolio stan-
dardshavealsodrivenvery significant if sometimesboom–bust
winduptake,particularly inTexaswithaquarterof thatnation’s
installed capacity. Table I shows the total wind energy installed
capacity at the end of 2010 in the regions considered in this
paper together with their proportion of electricity consumption
now supplied by wind energy.
Wind generation penetrations have now reached significant
levels (from 10%–20%) in countries such as Denmark and
Spain, and states such as South Australia and Iowa. This, in
turn, has driven changes in electricity market design and wider
policyarrangements in thesecountries inorder tobettermanage
the major contributions of highly variable and only somewhat
predictablewindgenerationwithin theirpowersystems.Formal
participation by wind generation in electricity market dispatch
and ancillary services may be limited to day-ahead markets
1949-3029/$31.00 © 2012 IEEE
810 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY,
VOL. 3, NO. 4, OCTOBER 2012

TABLE I
TOTAL INSTALLED WIND ENERGY CAPACITY AT THE
END OF 2010 AND
THE PROPORTION OF ELECTRICITY CONSUMPTION
SUPPLIED IN 2010
FOR SELECTED COUNTRIES AND REGIONS
or include real-time markets and even ancillary services such
as voltage and frequency control. Improved wind forecasting
systems have reduced prediction errors, whereas delayed gate
closures and active demand participation have decreased the
potential energy imbalances that have to be resolved by the
industry. Other electricity market arrangements that may affect
wind energy are charges due to imbalances settlement, addi-
tional income due to capacity recognition and rewards when
wind generators reduce their output following orders from
transmission system and market operators.
This paper draws together someof thekeyexperiences, chal-
lenges, and responses to growing wind penetrations in selected
jurisdictions within Europe, the United States, and Australia.
These are by no means the only countries from which lessons
might be drawn, or where significant wind industry develop-
ment is underway. However, they do represent important and
interesting examples of some key current and possible future
wind markets, and the range of support policy approaches and
electricitymarket arrangements thatmaybeemployed to facili-
tate highwindpenetrations.The selected states inAustralia and
theU.S.are thosewith thehighest installedwindcapacities.The
paper is divided into four further sections. Section II presents
the main support policies in place for wind in these selected ju-
risdictions. Section III provides an overview of their different
electricity market arrangements. Section IV discusses the inter-
actions between wind energy and the support policies and elec-
tricity markets considered in the previous two sections. Finally,

conclusions are presented in Section V.
II. SUPPORT POLICIES
A wide range of policy mechanisms to support wind energy,
or renewable energy more generally, have been used by dif-
ferent jurisdictions over recent decades. A general assessment
of the available support policy mechanisms and their potential
strengthsandweaknessescanbefoundin[3].Fourgeneralwind
energy support policy mechanisms are considered here. The ju-
risdictionscovered in thispaper thathaveoptedforeachof these
four approaches, together with their particular characteristics,
are described below and summarized in Table II.
TABLE II
RENEWABLE ENERGY SUPPORT POLICIES IN DIFFERENT
REGIONS
A. Tender Schemes
The theoretical basis of tender schemes is highly
promising—governments can set a target of installed capacity
or total public expenditure, and invite prospective project de-
velopers to submit project tenders that specify the government
support—capital $ or $/MWh—required to proceed. Gov-
ernments can then choose the lowest cost project providers.
Unfortunately, the experience to date with tender-based ap-
proaches is mixed. For example, a number of countries opted
for this approach in order to drive initial deployment of re-
newable energies but abandoned it some years later. In 1990,
the UK introduced the Non Fossil Fuel Obligation mainly as a
policy to support the nuclear industry although it also drove the
installation of a number of wind farms. Ireland introduced its
tender scheme in 1996 based on the UK model, but abandoned
it ten years later because it failed to reach its set targets [4]. It
is also notable that China adopted a franchise tender program

in 2003 that was abandoned in 2009 for onshore projects.
What proved to be irrationally low bids offered by competing
developers led to the Government selecting projects with such
lowrequired support prices that the“winning”proponentswere
later unwilling or unable to actually undertake their projects
[5]. However, China kept a tender scheme for offshore wind
APARICIO et al.: COMPARISON OF WIND ENERGY
SUPPORT POLICY AND ELECTRICITY MARKET DESIGN
811
farms and Denmark opened a scheme in2005, also for offshore
wind, with very good results [6].
A number of U.S. states use tender-based processes for their
renewable portfolio standard although these may also be based
around the use of tradable certificates as outlined in the next
section.The jurisdictionsusing such tendershavealsoachieved
mixed success [7].
B. Feed-in Tariffs
Feed-in tariff schemes adopted by Denmark, Germany, and
Spain have without doubt been the primary drivers of their
significant wind energy deployment over the last two decades.
Initial policy settings in these three countries, announced in the
1990s,were similar and relatively simple schemeswith a single
tariff for all renewable producers. These FIT schemes provided
an electricity consumer funded fixed price for each MWh of
generation over a given time period. Any project meeting
the scheme requirements was eligible for this payment. As
wind installed capacity started to rise, however, these policies
have been significantly amended. Each country has followed
different strategies.

Germany has decided to make important changes in the FIT
scheme, includingfixeddegression,anequalizationschemethat
tries to compensate the differences in wind resources between
regions, and higher tariffs for repowering and offshore wind
farms.The latest amendment of theGermanRenewableEnergy
Act, in forcesinceJanuary1,2012,has increased thedegression
for both onshore and offshore projects. However, the reduction
inoffshore tariffswillnotbeapplicableuntil 2018 insteadof the
originally proposed 2015. An optional accelerated repayment
model which offers a higher initial tariff for a reduced number
of years has also been introduced for offshore wind farms. This
amendmentalso introducedanewmarketpremium,openingthe
possibility fordirect selling(ordirectmarketing),whichhas im-
portant implications for the generators that participate, as it is
shown latter in the paper. German amendments have managed
tokeepannual increases in installedcapacity relativelyconstant
over the past decade as Fig. 1 shows.
Denmark phased out its FIT scheme in 2000. After a transi-
tionperiod, it adoptedaschemewhere the“feed in” tariff isnow
afixedpremiumpaymentaboveandbeyondwhat thewindfarm
projects earn from the electricity market. In Denmark’s case,
wind generators connected to the grid after January 2003 must
sell their production to the electricity market. Fig. 1 shows that
new installed capacity rapidly declined following this change.
However, andasnotedabove, the tenderingprocess foroffshore
wind farms has driven more than 200 MW of new capacity in
both 2009 and in 2010, and in 2013 the Denmark’s largest off-
shore wind farm with 400 MW will start operation and is ex-
pected to supply 4% of the country’s demand. This will help to
meet the Danish target of 50% from wind by 2020.
Spain introduced theoptionofwind farms takinganFITpre-
mium in addition to energy market prices in 1998, earlier than
Germany, but also kept the conventional FIT mechanism. Thus

wind energy producers are able to choose between both remu-
neration schemes and switch between them every 12 months.
No wind energy producers initially decided to participate in the
Fig. 1. Annual increases in installed wind capacity in Spain,
Germany, and
Denmark from 2002 to 2011. Sources: Respective national wind
associations.
electricity market so an amendment in 2004 provided extra in-
centives to switch. By the end of 2006, around 90% of wind
generationcapacitybid in themarketas thosearrangementspro-
vided higher revenues than the conventional FIT. Finally, fur-
ther amendments, announced in 2007, modified the tariffs in-
troducing a cap and a floor in the sum of market price plus pre-
mium, variable degression depending on inflation, and lower
tariffs once a technology target has been reached. Many wind
energy developers accelerated the installation of their projects
in order to complete them before this amendment came into ef-
fect at the beginning of 2008. The year 2007, therefore, saw a
record increase in installed capacity, as shown inFig. 1.Thede-
crease in new installed capacity over the last two years is due
to the“pre-assignation” register introducedby theSpanishgov-
ernment in 2010. A limited number of projects are approved in
order to ensure Spain does not surpass its targets for the wind.
Moreover, in January 2012, the government announced a tem-
porary moratorium that freezes policy support for any new re-
newable energy project due to the impacts of the Global Finan-
cial Crisis.
Ireland and China have now replaced their tender schemes
withFITs.InIreland,FITshavebecomethemainmechanismfor
supporting wind energy. Offshore wind farms have had higher
tariffs since2008 [4]. InAugust 2009,China announced itsfirst
FIT scheme for onshore wind with different tariffs that depend
on thewind resources and investment conditions in eachof four

regions [8].
C. Quota System/Tradable Certificates
Aquota systemisusually relatedwith tradable renewableen-
ergy certificates (or credits) and has different names depending
on the country. In the UK, the scheme that came into force in
2002 is known as Renewables Obligation (RO). Wind genera-
tors receivedaRenewableObligationcertificate (ROC)foreach
MWh of electricity generated. A 2008 amendment introduced
the concept of ROC banding in order to give additional ROCs
to emerging technologies. Thus, onshore wind, included in the
“reference” band, receives 1.0 ROC per MWh, while offshore
wind, which is included in the “postdemonstration” band, ini-
tially received 1.5 ROCs per MWh. The 2009 budget raised it
to2ROCsfor2009/10and1.75 for2010/11.TheROCsaresold
to electricity suppliers in order to fulfil a mandated obligation
placed upon them by the government according to its national
renewable energy quota or target. Suppliers can either present
enough ROCs to cover their obligations or they can pay for any
shortfall into a buyout fund. The Renewables Obligation Order
2009 introduced significant changes. It requests the Secretary
of State to announce the obligation level six months preceding
an obligation period. This obligation level is the greater value
of either the number of ROCs needed to meet a fixed target of
ROCs/MWh or a headroom that is calculated as the ROCs ex-
pected to be issued according to the amount of renewable elec-
tricity expected to be generated, uplifted by 10%. This “guar-
anteed headroom” mechanism sets an effective floor for ROC
prices once the obligation is reached. In the case where ROC
supply exceeds theobligateddemand, therefore, theROsystem

will then effectively operate in a similar manner to an FIT pre-
mium mechanism.
TheAustralianGovernment’sMandatoryRenewableEnergy
Target (MRET) commenced operation in 2001 as the world’s
first renewable energy certificate trading scheme [2]. It requires
all Australian electricity retailers and wholesale electricity cus-
tomers to source an increasing amount of their electricity from
newrenewablegeneration sources.The liablepartieswithin the
Australian electricity market are electricity retailers and those
large consumers who purchase directly from the wholesale
market. The “additional renewable electricity” that the liable
parties are required to acquire was originally intended to be
equivalent to 2% of their electricity purchases by 2010. The
Renewable Energy Target (RET) announced in 2009 raised the
requirement to 20% by 2020. Targets to date have been easily
met and the costs seem reasonable by international standards
[2].
Some State Governments have also set jurisdictional renew-
able goals although they are not backed with specific obliga-
tions. For example, in 2007, South Australia set a 20% target
for2014. In June2011, it hasbeenalreadymet.Thestate, there-
fore, set a new goal of 33% by 2020 [9]. Victoria, which has
the second highest installed wind capacity, has a 2020 goal of
25%, with a minimum of 20% for wind energy. However, it is
intended that the Federal RET provide almost all wind project
support in Australia. A number of changes have been made to
the schemeover its decadeof operationbeyondagreater target,
including separate arrangements for large-scale and small-scale
renewable energy systems [2].
Currently, there isnoafederalquotaschemein theU.S.How-
ever, 29 states, the District of Columbia, and Puerto Rico have
a Renewable Portfolio Standard (RPS), and eight states have a
renewable portfolio goal.

Texas adopted both an RPS and a renewable energy credit
(REC) trading program in 1999 [10]. The RPS target was
2000 MW of new renewable generation by 2009, in addition
to the 880 MW installed at the time. It was raised in 2005 to
5880 MW by 2015, where 500 MW must be resources other
than wind, and to 10000 MW by 2025. According to the 2009
compliance report, Texas had already surpassed its 2025 target
by 2009. The Electric Reliability Council of Texas (ERCOT)
acts as the program administrator of the REC trading program.
Iowa passed one of the earliest renewable energy laws in
the U.S. in 1983. It allocated 105 MW of renewable gener-
ating capacity between the two Iowan investor-owned utilities:
Mid-AmericanEnergyCompany (MEC)andAlliantEnergy In-
terstate Power and Light (IPL) [10]. As Table I shows, the re-
quirementhasbeenclearly surpassed sobothutilitieshavebeen
authorized to export RECs by participating in Midwest Renew-
ableEnergyTrackingSystem, Inc. (M-RETS).Renewablegen-
eratorsusedformeetingtheRPSarenotallowedtoexportRECs
in order to avoid double-counting [11]. A voluntary goal of
1000MWofwindenergycapacityby2010,establishedin2001,
has also been easily exceeded. Section 476.53 in Code of Iowa
(2009) provides that it is the intent of the general assembly to
attract the development of electric power generating facilities
within the state. Thus, when eligible new electric generation is
constructed by a rate-regulated public utility, the Iowa Utilities
Board, upon request, must specify in advance the ratemaking
principles that will apply when the costs of the new installa-
tionare included in electricity rates. InMarch2009, pursuant to
section 476.53, MEC filed an application for determination of
advance ratemakingprinciples for up to1001MWofnewwind
generation to be built in Iowa from 2009 through 2012. In De-
cember 2009, the Board took up MEC’s proposal. As a result,
MEC has installed significant wind generation capacity and is
expected to have a total 2284 MW by the end of 2012.

Minnesota introduced twoseparateRPSpolicies in2007,one
for theutilityXcelEnergyand the second for other electric util-
ities. The latter includes public utilities providing electric ser-
vice, generation, and transmission cooperative electric associa-
tions, municipal power agencies, and power districts operating
in the state [10]. The RPS for Xcel Energy requires 30% of its
total retail electricity sales in Minnesota to come from renew-
able sources by 2020. It included a minimum of 25% for wind
energybut aStateSenateBill passed in2009addedamaximum
of 1% from solar to this requirement. Thus, at least 24% must
come from windenergy, up to1% maycome from solar energy,
and the other 5% may come from other eligible technologies.
TheRPSforotherutilities requires25%of their total retail elec-
tricity sales in Minnesota to come from renewable sources by
2020 without any technology minimums. Minnesota has been
included in the M-RETS since 2008. This tracking system pro-
gram ascribes the same amount of credits to all eligible tech-
nologies independently of the state where electricity is gener-
ated.XcelEnergyisnotallowedtosellRECstootherMinnesota
utilities for RPS-compliance purposes until 2021.
813
California launched an RPS in 2002 with a target for its
electric utilities to have 20% of their retail sales derived from
eligible renewable energy resources in 2010 [10]. Senate Bill
X1-2, enacted in 2011, raised the requirement to 33% by
2020. The Bill also established three categories, also known
as buckets, of RPS-eligible electricity applicable to contracts
executed from June 2010. The decision adopted in December
2011 by the California Public Utilities Commission provides

detailed requirements for the three categories. Category one is
for electricity that is from an RPS-eligible generation installa-
tion that has its first point of interconnection with a California
Balancing Authority (CBA); scheduled from an RPS-eligible
generation installation into a CBA without substituting elec-
tricity; or dynamically transferred to a CBA. Category two is
for electricity that is firmed and shaped, providing incremental
electricity scheduled into a CBA. Category three is for those
transactions that do not meet the criteria of any of the two
previous categories, including unbundled RECs. There are
limitations on the amount of generation procured in categories
two and three that become increasingly narrow over time.
Categoryonegeneration is required tobeaminimumof50%of
the total for the compliance period ending in 2013, 65% for the
compliance period ending in 2016, and 75% thereafter. Until
the portfolio content categories become sufficiently clear, the
utilities are preferencing power purchase agreements (PPAs)
with installations that definitely belong to category one. Given
California’s increasing target, it is envisaged that PPA opportu-
nities in categories two and three will expand in time, and with
greater clarity on the arrangements.
The Western Renewable Energy Generation Information
System (WREGIS) tracks the renewable energy generated in
the region covered by the Western Electricity Coordinating
Council (WECC), California included. WREGIS issues cer-
tificates for every REC generated, which can be used to verify
compliance with state RPS. However, currently WREGIS is
not able to track the three new portfolio content categories.
D. Tax Credits
The U.S. Federal production tax credit (PTC) for wind has
had a checkered history over the past decade. The latest exten-
sionof2009providesan incometaxcreditof2.2¢/kWhuntil the
endof2012while addinganumberofprovisions.Taxpayers el-

igible for the PTC are allowed to take a business energy invest-
ment tax credit (ITC) equal to30% of the construction costs for
the installation or to receive a cash grant of equivalent value if
construction began by the end of 2011. Before 2009, installa-
tion owners that did not generate enough taxable income were
unable to utilize PTC so they had to monetize the PTC through
tax equity investors.
The PTC, which applies for the first ten years of electricity
production [12], has been remarkably successful in supporting
wind deployment when it has been in place. However, during
the periodic lapses of the PTC prior to congressional renewal,
the state-based RPS mechanisms alone were not able to sustain
the growth of wind power [13]. As it represents a credit against
passive income, the PTC has a significant resemblance to an
FIT premium. In fact, both are a fixed cash incentive provided
to each kWh generated by wind.
Finland is the only EU country which uses tax incentives as
the main support scheme for renewable energies. This policy,
however, has not been effective for wind development [6].
III. OVERVIEW OF ELECTRICITY MARKETS
AROUND THE WORLD
Jurisdictionsaround theworldhave takenawide rangeofap-
proaches to electricity industry restructuring over the past three
decades. Care must, therefore, be taken when comparing in-
dustry approaches and performance. Generally, countries with
restructured electricity industries have both forward markets
and real-time markets. In the forward markets, electricity is
traded either centrally on a power exchange or bilaterally di-
rectly between market participants. A key role for the forward
markets is to support unit commitment of those thermal genera-
tors that require generation scheduling aday prior to energyde-
livery. In theday-aheadmarkets,electricity is traded in intervals

(settlementperiods) thatmaybeonehour longor lessdepending
on the market design. A key challenge for these arrangements
is that unexpected generator outages or changes in demand can
take place between the closing of the day-ahead market and de-
livery thenextday. Intradaymarketspermitmarketparticipants
to trade closer to thedelivery time, upuntil just prior to thegate
closure. Intraday markets are commonly a continuous trading
market that operates with a gate closure set one hour ahead of
the settlement period [14]. After gate closure, it is no longer
possible to change bids and offers for the settlement period.
Fromgateclosureuntil real-time, thedifferencebetweensupply
and demand is continuously balanced through real-time mar-
kets,where transmissionsystemoperators (TSOs)—orindepen-
dent system operators (ISOs) or regional transmission organi-
zations (RTOs)—purchase the energy needed to match supply
to demand and to solve any network constraints. These orga-
nizations also have a responsibility for imbalance settlement.
Chargesapply togenerators if theirenergydeliveriesdiffer from
theoffers submitted to themarket.Thepayment dependson the
price system. This may be a two-price system, which has one
price for imbalanceswith thesamesignas thesystemnet imbal-
ance (prejudicial for the system since they contribute to the net
imbalance)andadifferentprice (lower) for imbalanceswithop-
posite sign (beneficial since they counteract the net imbalance);
or a one-price system, which has only one price for all imbal-
ances. Normally, the two-price system is used, since it encour-
ages market participants to limit their imbalances according to
whether they add to, or subtract from, net system imbalances.
Toensurepower systemsecurityand reliability, ancillaryser-
vices are needed, including frequency and voltage control and
black-start capability. TSOs or their equivalent organizations
purchase ancillary services from service providers. Frequency
control ancillary services are commonly tradedonamarket that
has marked similarities to the electricity market.
Gate closure, the duration of settlement periods, and imbal-

ance settlement arrangements are all potentially very important
for wind energy integration. The brief descriptions of elec-
tricity market arrangements for each of the jurisdictions that
we are considering here, therefore, pay special attention to
these features.
A. Denmark
The four countries comprising the Nordic region (i.e., Den-
mark,Sweden,Norway,andFinland)areamongthefirst tohave
restructured their electricity industries. In 1993, they connected
their individual markets creating Nord Pool Spot, which was
the world’s firstmultinational power exchange. Nord PoolSpot
trades 74% of the electricity generated in the Nordic region.
The rest is traded through bilateral contracts. It also operates a
day-ahead market called Elspot, and an intraday market called
Elbas. Elspot trades in one-hour intervals and closes at 12:00.1
It introduced different area prices in order to deal with network
congestion between countries and within Norway [15]. Elbas
is a continuous intraday market that covers the Nordic Region,
Germany, and Estonia with a gate closure one hour before de-
livery. This intraday market is becoming increasingly signifi-
cant as more wind energy enters the grid given that imbalances
between its day-ahead contracts and produced volumes often
needtobeoffset [15].Asdiscussedbelow, thesemarketarrange-
ments are now being changed in ways that affect wind energy
directly.
Since April 2011, the gate closure for trading in Germany

has been reduced to 30 min. Negative prices have also been
introduced inallmarket areas sincemid-February2011 inorder
to price oversupply [16].
Each local TSO has a different set of ancillary services costs
according to their procurement arrangements and reserve re-
quirements. Energinet.dk is the Danish TSO and purchases dif-
ferent ancillary services inWesternandEasternDenmarkas the
former is synchronouslyconnected to theUCTEsystemand the
latter to the Nordel system. Denmark settles imbalances using
a two-price system. Norway used one-price system until 2008.
Since then, all Nordic countries have used a two-price system
forproduction imbalances andaone-price systemforconsump-
tion imbalances [17].
B. Spain
Spainbelongs,withPortugal, toMercado Ibéricode laElect-
ricidad (MIBEL)—the IberianElectricityMarket.Eachcountry
may have different prices (splitting) in the case of transmis-
sion restrictions. All market participants can either arrange bi-
lateral physical contracts or participate in a day-ahead market.
As with the Danish market, this day-ahead market is divided
into one-hour settlement periods and closes at 12:00. After the
day-aheadscheduling,generatorpositionscanbeadjusted in the
intradaymarket.Rather thanbeingacontinuousmarket, it is di-
vided into six sessions.Eachsessionhasadifferentgate closure
(around2hours)aswell asadifferent timeofdeliveryscope(up
to 9 hours in the sixth session).
The provision of ancillary services follows the general rules
common in theUCTEsystem.Theprimarycontrol is a compul-
sory service shared across all generators and without remuner-
ation. The secondary control is performed within control areas
according to the requirements, in MW, set by the Spanish TSO
(Red Eléctrica de España). The generators willing to offer this

1This time of day and the following ones are in 24-h notation.
service bid their power available and a market clearing process
calculates the price per MW. The price is paid even if their ser-
vicesarenot required(availability).The tertiarycontrol restores
the secondary control reserves under emergencyconditions and
while it is optional for generators to formally participate, the
TSO can call upon any generator should it be required. By con-
trast with secondary control reserves, tertiary control services
areonlypaid for ifused. Imbalancesare settledwitha two-price
system.
C. Germany
Most wind generation is not scheduled and, instead, is intro-
duced into the electricity market through one of the country’s
fourTSOs.Thedistributionnetworkoperators transfer thewind
generationforafixedprice to their respectiveTSO,which trans-
forms the load fluctuating profiles into standard load profiles
which are sold to all utilities [18]. Customers pay an average
tariff to utilities. According to [18], this profile transformation
mechanism isnot fully transparent andnot cost-optimized.Fur-
thermore, as the costs of the profile transformation are com-
pletelypassedthroughtothenetworkcustomers, there isnoeco-
nomic incentive to minimize these costs. This mechanism has
thereforebeenargued to represent an importantweaknessof the
GermanFITscheme.However, theamendmentapplicable from
January2012 [19]mayhelp inaddressing thesedrawbacks.Re-
newablegenerators candecideonamonthlybasis to change the
remunerationmechanismsandparticipate indirect selling in the
electricity market. They have to forecast their production and
are directly charged for their imbalances. The additional costs
are covered by an extra premium known as management pre-
mium.
TSOssharewindenergy imbalances,which reduces theneed

for reserves. In order to manage this equitably, it is allocated
proportionally to theTSOs’according to their consumption,not
their installed wind power.
D. United Kingdom
The British Electricity Trading and Transmission Arrange-
ments (BETTA) cover England, Wales, and Scotland. Northern
Ireland has been part of the Single Electricity Market together
with the Republic of Ireland (see Section III-E) since 2007. In
BETTA, over 90% of electricity is traded through unrestricted
bilateral contracts. A power exchange permits market partici-
pants tofine tune their contractedpositions.Gate closure is cur-
rently set one hour ahead of each half hourly settlement period.
ELEXON, the Balancing and Settlement Code Company, uses
a two-price system [20].
E. Ireland
The Single Electricity Market (SEM) commenced trading in
IrelandandNorthern Irelandonanall-islandbasis inNovember
2007. SEM is a mandatory power exchange where all Ireland’s
electricity must be traded. It has only a day-ahead market with
a gate closure at 10:00. Energy is settled weekly. SEM plans to
develop an intraday market [21].
815
F. Australia
The Australian National Electricity Market (NEM) includes

all states and territories other than Western Australia and the
NorthernTerritory. Itscenterpiece isasetof regionalgross-pool
spot energy and ancillary services markets that solve a secu-
rity-constrained dispatch every 5 min. The Australian Energy
MarketOperator (AEMO) is thewholesalemarket operator and
TSO for the entire system. Regions are currently located at all
borders between states within the NEM. All generating plants
of greater than30-MWcapacity (except intermittent generation
including wind) are required to participate as scheduled gen-
erators and submit offers to sell or bids to buy energy (and/or
ancillary services) in the NEM dispatch process. The predis-
patch processes forecasts up to 40 hours ahead of real time and
provides public forecasts of energy and ancillary service prices
and (privately to each dispatchable participant) dispatch levels
based on participant bids and offers, the demand forecasts and
the estimated effects of dispatch constraints. Demand is per-
mitted to participate directly in the wholesale market; however,
nearly all end-users interface with the market through an elec-
tricity retailer [2].
There are eight Frequency Control Ancillary Services
(FCAS) markets to provide load following (raise and lower)
and three contingency responses of different speed (raise and
lower) between the 5-min energy dispatches. Market dispatch
co-optimizes energy and FCAS bids and offers to establish re-
gional prices for both energy and FCAS for each 5-min period.
Commercial trading is based on these prices averaged over
30 min. Locational pricing within regions is achieved using
averaged loss factors. Importantly all generators are permitted
to change their offers (rebid) just prior to each 5-min dispatch.
Furthermore, the only commercially significant prices in the
NEM are these averaged30-minprices—the predispatchprices
are advisory only. Note also that the NEM is an energy-only
market and participants are required to manage their own unit
commitment and other intertemporal scheduling challenges
(within a range of technical dispatch constraints).

G. Texas
ERCOT manages the electricity industry arrangements sup-
plying 85% of Texas demand and covering 75% of state land
area. The ERCOT control area is not synchronously connected
to either the Eastern or Western Interconnection. However, it
can exchange about 860 MW through dc links. In December
2010, ERCOT switched from a zonal market to a nodal market
in order to improve price signals and dispatch efficiencies and
assign localcongestiondirectly[22].Theday-aheadmarketem-
ploys a co-optimization engine that uses both energy and ancil-
lary services offers to calculate the energy schedules and ca-
pacity awards. ERCOT closes this market at 14:30.
The real-timemarket is called security constrainedeconomic
dispatch(SCED).ERCOTgenerally runs theSCEDevery5min
using offers by individual resources and actual shift factors by
each resourceoneach transmissionelement.The settlementpe-
riods are 15 min long.
H. Minnesota and Iowa
Minnesota and Iowa belong to the Midwest ISO (MISO),
which operates a day-ahead market and a real-time and op-
erating reserves market. They coexist with both financial and
physical bilateral transactions between industry participants.
The day-ahead market simultaneously clears energy and op-
erating reserves on a co-optimized basis for every one-hour
settlement period. Security constrained unit commitment
(SCUC) and SCED algorithms ensure the scheduling of ade-
quate resources [23].
The real-time and operating reserves market uses an SCED
algorithm to simultaneously balance supply and demand and
to meet operating reserves requirements amongst other actions.

The gate closure is set to only 30 min ahead of delivery.
InMarch2011,an importantchangecame intoeffectwith the
creation of a new category of resources called Dispatchable In-
termittentResources (DIRs).Thiscategoryonlyapplies towind
farmsandallows themtovoluntarilyparticipate in the real-time
market fromJune2011,where theyareeligible tosupplyenergy
but not operating reserves. From September 2011, DIR imbal-
ances are settled similarly to conventional generators although
only when an 8% tolerance band is exceeded, with a minimum
of 6 MWh and a maximum of 30 MWh, for four or more con-
secutive 5-min intervals within an hour. Wind generators are
exempt of imbalance charges in cases of force majeure, such as
extreme winds.
I. California
The California ISO (CAISO) has three day-ahead processes:
a market power mitigation determination, integrated forward
market, and residual unit commitment. A bid from a market
participant that fails the market power test is automatically re-
duced to the reference level price of that participant, and the
system determines the minimal and most efficient schedule of
generation to address local reliability. The integrated forward
market simultaneously analyzes the energy and ancillary ser-
vices market to determine the transmission capacity required
(congestion management) and confirm the reserves that will be
needed to balance supply and demand based on supply and de-
mand bids. It ensures generation meets load and that all final
schedules are feasible with respect to transmission constraints
as well as ancillary services requirements. When forecast load
is not met in the integrated forward market, the residual unit
commitment process enables CAISO to procure additional ca-
pacity by identifying the least cost resources available [24].
The real-time market produces energy to balance instanta-

neous demand, reduce supply if demand falls, offer ancillary
services as needed and, in extreme conditions, curtail demand.
The gate closure is set 75 min ahead of delivery. The market
has two unit commitment mechanisms. Real-time unit commit-
ment assigns fast- and short-start units in 15-min intervals and
looks forward 15 min, while short-term unit commitment as-
signs short- and medium-start units every hour and looks for-
ward three hours beyond the settlement period every15min. In
real-time, the economic dispatch process dispatches imbalance
volumes and energy from ancillary services. It runs automat-
ically and dispatches every 5 min for a single 5-min interval.
Under certain contingency situations, CAISO may dispatch for
a single 10-min interval.
IV. WIND ENERGY PARTICIPATION IN ELECTRICITY
MARKETS
A compilation of some key themes for wind energy integra-
tion coming out of the individual jurisdictional experiences is
presented as follows.
A. Wind Energy Forecasts and Generation Scheduling
Windgeneratorsparticipatinginday-aheadmarketsmustpre-
dict theiroutputbeforemarketclosetime(varyingbetween9.5h
in ERCOT to 40 h ahead of delivery in some other markets).
Theday-aheadand longerwind forecasts required for suchgen-
eration scheduling are not sufficiently precise and two possible
alternatives have been put forward to help manage this. In the
first case, adopted in Germany (not for renewable generators

that participate in direct selling in the electricity market), Aus-
tralia, MISO, ERCOT, and CAISO, the output of all wind gen-
erators is predictedover a rangeof timehorizons throughacen-
tralized forecasting system. In the second approach, adopted in
Denmark, Spain, and the UK, numerous prediction companies
compete to provide forecasts for wind farm clients [25]. Note
that such wind energy forecasts have value to all market partic-
ipants, not just the wind farms—another argument for centrally
provided forecasting services.
Even in industries where wind is not required to partici-
pate in forward markets there is considerable value in wind
farms and other market participants having useful day-ahead
forecasts. These can play a role in derivative market trading
around the future spot price, maintenance scheduling of wind
farms, unit commitment strategies of thermal plant, and pro-
duction scheduling of hydro. There is also considerable value
in useful short-term forecasts that assist in generator bidding
in the real-time spot markets. For example, recent changes
in the Australian market arrangements have more formally
incorporated wind farms into market scheduling and ancillary
services arrangements through a semischeduled classification.
There are similarities with the new DIR category within the
MISO arrangements that was described in the previous section.
B. Imbalance Settlement
Imbalance settlement is probably the aspect of electricity
markets design that has highest impact on wind energy [17],
[26]. Ingeneral, systemandmarketoperatorscalculate such im-
balances and settle them according to either a one- or two-price
system. This determines how the total balancing costs are
distributed and how incentives are given to market participants
[27]. With the one-price system used for balance settlement
in Norway before 2008, the balance costs of wind energy
were negligible as long as its random variability assured that

positive imbalances from some wind generation were compen-
sated by negative ones [28]. Since 2008, all Nordic countries
have used a two-price system for production imbalances and
a one-price system for consumption imbalances [17]. Spain
originally applied a different one-price system that charged for
all imbalance volumes independently of their direction. Now it
uses this two-price system: zero costs for producers that do not
contribute to the system net imbalance and a penalty for those
that do. Denmark, the UK, and Australia (only with Regulating
FCAS)also runa“causerpaysprocedure” inorder to assign the
cost of the regulating power to those market participants who
are responsible for the imbalance. Note, however, that there
are inevitably difficulties in assigning such responsibilities
given the complex nature of electricity industry operation. This
kind of arrangement may impose significant charges on wind
energy so some tolerance in energy imbalance is often applied
and the prices paid by wind are often lower than those paid by
conventional generation.
In Germany, full responsibility for balancing wind genera-
tion is assumedby theTSOs. Indeed, they are in chargeof fore-
casting, scheduling, and balancing. In the U.S., Federal order
890assists intermittent generationwithmoreflexiblebalancing
settlement [29]. For example, DIRs in MISO are only charged
if an 8% tolerance band is exceeded for four or more consec-
utive 5-min intervals within an hour, whereas CAISO has the
Participating IntermittentResourceProgram(PIRP).Windgen-
erators that participate in the PIRP have better arrangements.
Energy imbalances are netted on a monthly basis and settled at
a monthly weighted market-clearing price.
InAustralia, thehybrid5/30mingross spotmarketwithasso-
ciated frequency control ancillary services (FCAS) seems rea-
sonably supportive of wind integration [2].
Reduced gate closures permit rebidding up to few minutes

from delivery whereas in many intraday markets it is possible
to reschedule with updated forecasts of only one hour (or less
as in MISO). However, this can reduce liquidity; especially in
the case of market power (CAISO has market power mitiga-
tion). If themarket is openuntil very close togate closure,wind
generators are able to better forecast and manage the energy
that they actually produce. However, the closer to real time,
the fewer the conventional generators that may be available to
help wind generators to correct their position. Continuous mar-
kets permit changes to bids closer to real time (for example, in
the UK until one hour before). In Spain, by contrast, the six in-
traday sessions have delivery scopes up to 9 hours. There are
higher forecast errors; however, there is also a greater willing-
nessand interestamongst thegenerators tochange theirposition
(including all wind generators), so the market is likely to have
greater liquidity [18]. Experience to date suggests that intraday
pricesdiffer little fromtheday-aheadprices.So thismore liquid
market permits generators to improve their bids and offers at
littlecostwhilealsoreducingregulationrequirements.Note that
gross pool arrangements such as those of the Australian NEM
resolveshort-termsupply–demandbalancewith thecompulsory
involvement of all generation and load.
Wind imbalances are significantly reduced by aggregating
the bids of wind farms over geographically dispersed locations
and large areas [26]. Active demand participation is also useful
in reducing imbalances. FERC Order 719 considers electricity
market accepting bids from demand response resources, on a
basis comparable to any others, for ancillary services that are
acquired in a competitive bidding process.

817
C. Curtailment
An excess of wind generation may cause system operating
problems such as transmission line overloading or insufficient
regulation reserves that force TSOs or their equivalent organi-
zations toorder real-timecurtailment towindgenerators during
normal operation. Wind generators that reduce power may be
rewarded for this, depending on the electricity market arrange-
ments. When rewards are given, wind generators typically earn
a percentage of what they could have generated. In Ireland it
is 100% [30] while in Spain it is just 15%. Another possibility
to reduce over-production is to permit negative wholesale
market prices [27]. Prices in electricity markets typically have
a zero-floor limit. Elbas has accepted negative prices since
2009 while Australia and some U.S. electricity markets have
permittednegativepriceswellbefore this.CAISOhasproposed
to lower the bid floor from $30/MWh to $150/MWh, then
to $300/MWh. Spain may consider accepting it only for
downward regulation provision.
V. CONCLUSION
Wind generation penetrations have reached significant levels
in some countries and states around the world. In almost all
cases this has required changes in both policy support mech-
anisms and electricity market design in order to better manage
wind energy. Tender schemes have been successful only with
offshore wind. FIT tariff premiums are now the predominant
scheme as a market oriented transition from conventional FITs.
Denmark made the transition mandatory whereas Spain, and
Germany since 2012, introduced incentives. For the case of
Spain, theypersuadedaround90%ofwindgenerators to switch
from FITs. The “guaranteed headroom” mechanism introduced
in the UK is a way to transform the Renewables Obligation

mechanism into FIT premiums given an oversupply of ROCs.
The PTC can also be considered as a form of FIT premiums.
Quota-based approacheshavehadmore limited applicationand
mixed success to date.
With regard to electricity market arrangements, imbalance
settlement is probably the element of electricity market design
thathas thehighest impactonwindgeneration. Itdependsonthe
price system, the specificarrangements for intermittent sources,
possibilities to aggregate bids of wind farms over geographi-
cally dispersed sites and large areas, active demand participa-
tion, and gate closure. Small gate closures permit scheduling
with reduced forecast errors. However, this reduces liquidity.
In conclusion, there are complex and changing interactions
between a) the desired policy objectives of increasing wind
energy generation or renewable energy more generally, b) the
chosen policy approaches applied to facilitate greater deploy-
ment, and c) the commercial and regulatory arrangements that
govern how such wind energy is integrated into existing elec-
tricity industries. It is evident that the challenges of appropriate
policy and electricity market arrangements grow as wind pen-
etrations increase. Some clear trends have emerged with those
countries now experiencing high penetrations. These include
the need to have wind farms more formally participating in
the electricity market mechanisms that manage supply-demand
balanceover the immediate to longer-term.While thismight be
seen as an impediment to greater wind deployment, it is better
understood as the inevitable process of wind transitioning from
a small industry contributor that can be ignored as “negative
load,” to a seriousplayer that cangreatly help in addressingour
growing energy security and climate change challenges within
the electricity sector.
REFERENCES

[1] REN21, Renewables 2011 Global Status Report 2011.
[2] I.MacGill, “Electricitymarket design for facilitating the
integrationof
wind energy: Experience and prospects with the Australian
National
Electricity Market,” Energy Policy, vol. 38, pp. 3180–3191,
2010.
[3] N. Enzensberger, M. Wietschel, and O. Rentz, “Policy
instruments
fostering wind energy projects—A multi-perspective evaluation
approach,” Energy Policy, vol. 30, pp. 793–801, 2002.
[4] European Renewable Energy Council, Renewable Energy
Policy Re-
view Ireland, Mar. 2009.
[5] F. Wen, D. Hua, Q. Wang, and S. N. Singh, “Wind power
generation
in China: Present status and future prospects,” Int. J. Energy
Technol.
Policy, vol. 6, pp. 254–276, 2008.
[6] The Support of Electricity From Renewable Energy Sources
Commi-
sion staff working document; SEC(2008) 57, Jan. 2007.
[7] K. Cory, T. Couture, and C. Kreycik, Feed-in Tariff Policy:
Design,
Implementation, and RPS Policy Interactions NREL, Tech. Rep.
NREL/TP-6A2-45549, Mar. 2009.
[8] Z.-Y. Zhao, J. Zuo, L.-L. Fan, and G. Zillante, “Impacts of
renewable
energy regulations on the structure of power generation in

China—A
critical analysis,” Renew. Energy, vol. 36, pp. 24–30, 2011.
[9] South Australia Meets 20% Renewable Energy Target,
Ministerial
Statement Jun. 2011.
[10] DSIRE, Database of State Incentives for Renewables &
Efficiency
[Online]. Available: http://www.dsireusa.org
[11] Order Approving Facilities and Associated Capacities,
Adopting
Requirements for M-RETS Participation, and Requiring Report
Iowa
Utilities Board Order, Docket AEP-07-1, Nov. 2007.
[12] Federal Policy, American Wind Energy Association
[Online]. Avail-
able: http://www.awea.org/issues/federal_policy/index.cfm
[13] C.-J. Yang, E. Williams, and J. Monast, Wind Power:
Barriers and
Policy
Solution
s Duke University, Nov. 2008.
[14] Current State of Intraday Markets in Europe ETSO, May
2007.
[15] Nord Pool Spot, The Power Market—How Does It Work

[Online].
Available: http://www.nordpoolspot.com/How-does-it-work
[16] NordPoolSpot
IntroducesNegativePricesonElbas,andReducesGate
ClosureonElbas inGermanyPieceofMarketNewsNo.11/2011,Feb.
2011.
[17] H. Holttinen, P. Meibom, A. Orths, F. Hulle, B. Lange, M.
O’Malley,
J. Pierik, B. Ummels, J. O. Tande, A. Estanqueiro, M. Matos, E.
Gómez, L. Söder, G. Strbac, A. Shakoor, J. Ricardo, J. C.
Smith, M.
Milligan, and E. Ela, Design and Operation of Power Systems
With
Large Amounts of Wind Power Final Report, IEA WIND Task
25,
Phase One 2006–2008, Aug. 2008.
[18] C.Klessmann,C.Nabe, andK.Burges, “Pros andconsof
exposing re-
newables to electricity market risks—A comparison of the
market in-
tegration approaches in Germany, Spain, and the UK,” Energy

Policy,
vol. 36, pp. 3646–3661, 2008.
[19] Act on Granting Priority to Renewable Energy Sources
(Renewable
Energy Sources Act—EEG), Version Applicable as at 1 January
2012.
[20] Association of Electricity Producers, Electricity Market
[Online].
Available: http://www.aepuk.com/about-electricity/electricity-
market
[21] SEMO, Single Electricity Market Operator [Online].
Available: http://
www.sem-o.com
[22] ERCOT,TexasNodalMarket
ImplementationArchive[Online].Avail-
able: http://nodal.ercot.com/index.html
[23] MISO, Market Information [Online]. Available:
https://www.mid-
westiso.org/MarketsOperations/MarketInformation/Pages/Marke
t-

Information.aspx
[24] CAISO,Market Processes [Online].Available:
http://www.caiso.com/
market/Pages/MarketProcesses.aspx
[25] J. Rivier Abbad, “Electricity market participation of wind
farms: The
success story of the Spanish pragmatism,” Energy Policy, vol.
38, pp.
3174–3179, 2010.
[26] F. V. Hulle, N. Fichaux, A.-F. Sinner, P. E. Morthorst, J.
Munksgaard,
S. Ray, C. Kjaer, J. Wilkes, P. Wilczek, G. Rodrigues, and A.
Ara-
pogianni, Powering Europe: Wind energy and the electricity
grid, A
Report by the European Wind Energy Association, Nov. 2010.

[27] C. Hiroux and M. Saguan, “Large-scale wind power in
European elec-
tricity markets: Time for revisiting support schemes and market
de-
signs?,” Energy Policy, vol. 38, pp. 3135–3145, 2010.
[28] A. Helander, H. Holttinen, and J. Paatero, “Impact of wind
power on
the power system imbalances in Finland,” in Proc. 7th Int.
Workshop
onLarge-ScaleIntegrationofWindPower
intoPowerSystemsMadrid,
España, 2008.
[29] NERC, Accommodating High Levels of Variable
Generation Apr.
2009, Special Report.
[30] T. Ackermann, G. Ancell, L. D. Borup, P. B. Eriksen, B.
Ernst, F.
Groome, M. Lange, C. Mohrlen, A. G. Orths, J. O’Sullivan, and
M.
de la Torre, “Where the wind blows,” IEEE Power Energy Mag.,
vol.
7, no. 6, pp. 65–75, Nov./Dec. 2009.

Néstor Aparicio (S’06–M’12) received the M.Sc.
degree fromtheUniversity JaumeIofCastelló (UJI),
Castelló de la Plana, Spain, in 2002, and the Ph.D.
degree from Universidad Politécnica de Valencia,
Spain, in 2011.
He is an Assistant Professor of the Electrical
Engineering Area at Universitat Jaume I, with re-
search interests in thegrid integrationofwind-power
generators. For 6 months, he visited the Institute
of Energy Technology of Aalborg, Denmark and
the Centre for Energy and Environmental Markets
(CEEM), Sydney, Australia in 2006 and 2008, respectively.
Iain MacGill (M’10) is an Associate Professor in
the School of Electrical Engineering and Telecom-
munications at the University of New South Wales,
Sydney, Australia, and Joint Director (Engineering)
for the University’s Centre for Energy and Environ-
mental Markets (CEEM). His teaching and research
interests at UNSW include electricity industry
restructuring and the Australian National Electricity
Market, sustainable energy technologies, renewable

energy integration into power systems, and energy
policy.
Juan Rivier Abbad (M’99) received the Electronic
Engineering degree from the Universidad Pontificia
Comillas,Madrid, in1992,andthePh.D.degreefrom
the same University in 1999.
He joined the Instituto de Investigación Tec-
nológica (IIT) in 1992 as a research fellow and the
Electrical Department of the Engineering School
(ICAI) in 1999 as an assistant professor, both at the
Universidad Pontificia Comillas. He was a Visiting
Research Fellow at the Centre for Energy and
Environmental Markets (CEEM) of the University
of New South Wales, Australia, during the academic year
2005/06. He is
currently the Energy Management Responsible at Iberdrola
Renovables. He
has experience in industry joint research projects in the field of
electric energy
systems in collaboration with international and Spanish utilities,
and with
energy regulatory commissions. His areas of interest are

regulation, operation,
and integrationof renewableenergysourcesgenerators,
andelectricitymarkets.
Hector Beltran received the M.Sc. degree in indus-
trialengineering, in2004, fromtheUniversitatJaume
I (UJI),Castellóde laPlana,Spain, and thePh.D.de-
gree inelectricalengineering, in2011, fromtheTech-
nicalUniversityofCatalonia (UPC),Terrassa,Spain.
During 2003, he worked at the European Centre
forNuclearResearch (CERN),Geneva,Switzerland.
From 2004 to 2006, he worked as a Researcher at
theElectronicandEnergyDepartmentsof theEnergy
Technologycal Institute (ITE),València,Spain.Since
2006, he is an Assistant Professor in the Electrical
Engineering Area at UJI. Meanwhile, he visited the Institute of
Energy Tech-
nology, Aalborg University, Denmark (for 6 months), and the
Renewable Ener-
gies Electric Systems Research Group at the Technical
University of Catalonia
(UPC),Spain(for9months).Hiscurrent researchinterests
includemassivepho-

tovoltaic integration into the grid, energy-storage systems, and
microgrids.
1
Effects of increasing wind power penetration
on the physical operation
of large electricity market systems
Bernd Klöckl, Member, IEEE, and Pierre Pinson
Abstract—This contribution describes indirect coupling effects
between wind power infeed and physical operation of power
market systems by means of qualitative hypotheses, backed by
suited exploratory data analyses. As an example, the case of a
central European TSO located in the vicinity of a control block
with high wind penetration is demonstrated. It shows, based
on established methods of computational statistics, considerable
nonlinear effects on cross border power flows and transmission
system flows of that control block, conditional on increasing
wind power penetration in an interconnected market system.
The
observed effects are theoretically explained through the

influence
of the wind infeed on the behaviour of market participants
and attributed to indirect coupling between wind power and
conventional generation in adjacent control blocks.
Index Terms—Wind power, energy market, cross border flows,
principal component analysis.
I. INTRODUCTION
T HE rapidly increasing share of wind generation in
theEuropean energy markets, together with further progress
in their liberalization has led to a number of direct and indirect
effects of wind power injection on flow patterns in the trans-
mission systems and has increased the operational challenges
for TSOs. The developments throughout the last years have
shown that increasing the share of wind generation beyond
certain levels has created the need for further investigations
on the following issues:

1) connection issues and grid codes (technical)
2) high concentrations of volatile infeeds in certain areas,
e.g. close to windy coast lines, leading to the need for
infrastructure reinforcements, especially in the transmis-
sion system (technical)
3) implications of day-ahead wind forecasts on the whole-
sale electricity prices in a given market area (economic)
4) coupling of these implications to adjacent markets and
TSOs (economic)
5) increasing wide-scale influence of wind generation on
the behaviour of conventional generation (economic)
6) thus increasing operational challenges for TSOs beyond

the effect of the wind infeed itself (techno-economic).
The first two points above are currently being extensively
treated by planning and operation staff of DSOs and TSOs,
and the last point is experienced more and more by the grid
B. Klöckl is with the Market Management Department of
Verbund Austrian
Power Grid, Vienna, Austria, EU (e-mail: [email protected])
P. Pinson is with the Technical University of Denmark, DTU
Informatics,
Kgs. Lyngby, Denmark, EU (e-mail: [email protected])
operation staff. In this paper, the authors elaborate on the
points 3 to 5, which is a simplified functional chain creating a
feedback loop between technical and economic issues of wind
generation.
The remainder of the paper is structured in the following

way: First, a general overview of the EU transmission system
with respect to the market operation and a glimpse on the
most obvious effects of wind infeed is given. Then, a suited
statistical analysis method for the problem set is derived. This
analysis is exemplified on the measurements taken from the
system of a central European TSO being indirectly affected
by large wind power penetrations in adjacent markets. The
conclusions attempt to relate the observed effects to future
investigation needs.
II. EFFECTS OF WIND POWER ON MARKET PRICES
A. The EU market and transmission system
Fig. 1. Overview of the ETSO member states and the related EU
market

zones. The geographical distance between northern Germany
and the APG
block as indicated is approximately 750 km.
The European transmission system operators are organized
in ETSO [1], which is an association that covers five different
synchronous zones. The largest zone of the EU transmission
network is the UCTE grid, that links all national transmission
systems of continental Europe except for the Baltic states
(Fig. 1). It is one of the largest synchronous systems in the
world with an installed generation capacity of approximately
650 GW [2]. For the time being, the different national trans-
mission networks represent market zones that are technically
© CIGRE2009 978-2-85873-080-3

2
connected by sets of tie lines and commercially linked by
allocation mechanisms for the cross-border transmission ca-
pacities.
B. Observed interaction of wind generation and spot prices
Fig. 2. Potential shift of the market settlement price on the
aggregated merit
order curve of a price zone in dependence on wind generation.
Quantiles of
the merit order curve (e.g. a 90% quantile) are indicated as an
illustration of
the uncertainty of the market settlement prices as a consequence
of market
imperfections and other factors.
Recently, the effect of wind power on market prices has
been discussed for the cases of market areas with high wind

penetration. Generally, there seems to be an agreement that
wind generation lowers the spot market prices, while the
mechanisms behind are not entirely clear. For a first general
treatment of the effect, see e.g. [3]. In [4], the reduction of
price in dependence on wind generation is described. This
does not come as a big surprise since wind generation in the
German system is prioritized in the dispatch and can thus
be regarded as a negative load. In [5], this is interpreted
as an effect on the activation of the merit order curves of
the GENCOs, meaning that in the presence of wind there
is increased probability that expensive power plants will not
settle the spot prices. This effect is then opposed to the costs
of the renewable energy support scheme1. See Fig. 2 for an

graphical illustration of the mechanism. Jóhnsson [6] provides
detailed modeling recommendations for the price reduction
effect applied to the Danish case and states that the decisive
variable for the reduction effect is clearly the wind power
prediction, rather than the actual production at the time of
delivery.
In addition to the findings cited above, it can be shown that
there is a second effect of high wind generation on the market
prices which has not yet been discussed in the literature. One
would assume that, especially for market zones with priority
wind dispatch, the variable to look at is simply a fictitious
expected system load,

Lf ic = L − P̂ w, (1)
1However, the reduction of energy price for the end consumers
must not
be confused with the calculation of the total socio-economic
benefit, which
has to take into account also the financial impact on the
GENCOs, the wind
turbine manufacturers etc.
where L is the system load (the consumption within the market
area), and P̂ w is the predicted wind generation. Fig. 3 shows
the German EEX spot prices for 2006-07 in a logarithmic scale
plotted against Lf ic. It can be observed that for equal Lf ic,
the mean prices at high wind generation are still slightly lower
and that price spikes virtually do not occur in periods of high
forecasted wind penetration
r̂ w = P̂ w/L. (2)
Fig. 3. EEX market prices for 2006 and 2007 in dependence on
different

levels of Lf ic (Sources: [2], [7], [8], [9], [10]), plotted for
events with a
wind penetration below 7.8% (red) and above 7.8% (green),
where 7.8% is
the mean of the wind penetration in the period. The logarithmic
scale for the
spot price is depicted on the l.h.s., while the plot on the r.h.s.
shows the mean
values in a linear scale. It can be observed that for the region in
which the
resulting loads to be covered by the market are equal, the price
is still always
lower for high r̂ w . This might be caused by price expectations
of the traders
or by the fact that not the entire volume of energy consumed in
Germany is
traded at the power exchange.
There is little doubt that this general price damping effect of
predicted wind infeed has an influence not only on the affected
market zone itself, but also on adjacent interconnected areas
with correlated market prices. In the following, we would like

to explain how this can be detected from observed data.
III. METHODOLOGY: CONDITIONAL MULTIVARIATE
DATA
ANALYSIS
Analysing the operation of a complex system may translate
to simultaneously studying the behaviour of a large number
of possibly redundant variables in a multivariate data analysis
framework. At a given time t the measured values for this
set of m variables are gathered in a single vector Xt. For
the example of the present study, these variables may be
cross-border flows or flows on transmission systems. Owing
to the instantaneous nature of electricity transmission, flow
data recorded at different points in the horizontal network
may often cloud the true underlying mechanisms by containing

redundant information and noise components that arise from
mechanisms other than the one that is subject to investigation.
A possibility to alleviate this problem is to employ Principal
Component Analysis (PCA) for dimension reduction. This will
3
be described in a first part below. One can then work in a
reduced basis defined by principal components, and develop
conditional parametric models for capturing the nonlinear
effects of the influential variables of interest, e.g. forecasted
wind power penetration, on the flows.
A. Dimension reduction with Principal Component Analysis
(PCA)

PCA is a classical method in multivariate data analysis,
which allows one to reduce the dimension of the problem at
hand, and to potentially work in a reduced orthonormal basis
defined by the set of (uncorrelated) principal components. For
a nice introduction to multivariate data analysis and PCA, the
reader is referred to [11], while more extensive mathematical
developments may be found in e.g. [12]. Consider a number
N of measured flow values Xt (being of dimension m), and
define X̃ t the centered and standardized version of Xt. This
simply means that for each dimension of X̃ t one has
X̃ t,j = τj (Xt,j ) =
Xt,j − X
̄ j
σj
, j = 1, . . . , m (3)

where X
̄ j and σj are the mean and standard deviation of
the jth flow variable. We will denote by τ the simultaneous
application of the τj transformations to all components of Xt.
Finding the principal components for the dataset considered
may be performed by diagonalizing the covariance matrix of
the data, given as
R =
1
N
N∑
i=1
X̃ tX̃
�
t (4)
By arranging the eigenvalues in decreasing order and identi-
fying the corresponding eigenvectors, one obtains the set of

principal components. Using the average eigenvalue method
[12, pp. 348], the set of retained principal components are
those for which the related eigenvalue is larger than the mean
eigenvalue of R. By writing Yi (i = 1, . . . , n, n < m) these
principal components, one then obtains an orthonormal basis in
which X̃ t can be written as linear combination of the principal
components,
X̃ t =
n∑
i=1
αiYi + �t, ∀ t (5)
plus �t which is a m-dimensional centered noise of finite
variance. By comparing the eigenvalues corresponding to the
retained principal components to the trace of R (i.e. the
sum of all eigenvalues), one can determine the degree of

explanation of the variance in the data, giving a hint on the
information content of the remaining noise. Note that the
principal components Yi can be seen as orthogonal modes
explaining the interrelated variations of the flows considered,
while
P = [Y1 Y2 . . . Yn] (6)
can be seen as the projection matrix allowing to project
standardized flow values X̃ t in the basis defined by the
principal components.
B. Conditional parametric models for local smoothing
The model in Eq. (5) permits one to express the flows as
a linear combination of the principal components Yi. Such
models can be extended in order to account for the potential
nonlinear effects of influential variables on the flows. Denote

by ut the values of these influential variables at time t. They
may include forecast wind power penetration, fictitious load or
the spot price on the EEX market for instance. The dimension
of ut should be kept low (say, lower than 3) owing to the so-
called curse of dimensionality. The model of Eq. (5) is then
extended to
X̃ t =
n∑
i=1
αi(ut)Yi + �t, ∀ t (7)
where the αi coefficients are not constant anymore, but instead
coefficient functions of the set of influential variables ut. �t
is still a m-dimensional centered noise of finite variance.
We do not describe here the detail of the method for
estimating the coefficient functions. In general, the challenge

is to define a fitting procedure for the coefficients αi(ut) that
allows to consistently eliminate �t in Eq. (7) (i.e. to regard
it as noise component). The basis for their estimation can be
found in e.g. [13]. No assumption is made about the shape of
the coefficient functions, except that they are continuous and
suffiently smooth for being locally approximated. The method
for their estimation consists of approximating them locally
at a number of fitting points with first order polynomials,
and of using weighted least squares for determining the
polynomial coefficients. Different variants of the method for
their estimation can be found in e.g. [14], [15].
C. Identifying trend surfaces
There may be different ways of using the estimated coef-

ficient functions in Eq. (7) for analysing the impact of the
defined influential variables on the flows. One can in a first
stage analyse the estimated α̂i functions themselves in order
to see how influential variables act on the contribution of the
various identified modes to the observed flows. Alternatively,
it may be easier to go back to the original flow variables and
to show what is the mean effect of the influential variables
in the variations of the various flows considered. It is this
alternative that will be preferred in the following. Indeed, by
simply discarding the noise term in Eq. (7), projecting the α̂i
functions back to the basis in which X̃ is defined, and using
the inverse τ transformation for getting back to the original
X variables, one obtains
X
̂ (u) = τ −1 (Pα̂i(u)) (8)
which defines the mean flows as a function of the influential

variables u. Variations in such mean flows as a function of u
can be seen as trends induced by u, which can be for instance
a trend induced by forecast wind power penetration. Examples
of such trend surfaces will be given and discussed below.
4
IV. METHODOLOGY APPLICATION EXAMPLE
A. Influential variables in the analysis
For this study, a closer look is taken at the control block
of Austria, operated by the TSO APG (Fig. 1). The installed
generation capacity in Austria is more than 19 GW (12 GW
in hydro power units), while the maximum load is less than
10 GW [2]. In spite of this excess of installed generation

capacity, the block has shifted its overall characteristics from
export to import throughout the last years, which can be caused
by a number of factors. The block is physically linked to six
different control blocks and a total of nine control zones.
The complexity of such structures is demanding a concise
top-down approach based on recorded data. Effects caused
by the following influential variables were subject to detailed
investigations:
1) fictitious load (Eq. (1)) in Germany, and
2) forecasted wind power penetration in Germany (Eq. (2)),
both together generating an indirect effect on generation and
trading within the APG block via market-related effects. The
trends of cross-border flows and APG system flows were

subsequently analysed for 2006-07 data according to the
methodology outlined in Sec. III above. The vector ut as
introduced in Eq. (7) then includes the measured values for
these 2 influential variables at time t. In parallel, the vector
Xt of flow values measured at time t may either relate to
the overall block balance (being one-dimensional in this case,
thus not needing the PCA step of the methodology introduced
above), or gathering the set of cross-border flows (m = 6), or
finally the flows on transmission lines (m = 23).
B. Flow patterns as a result of wind power
Fig. 4. Trend of the block balance of APG in dependence on the
wind power
penetration in Germany.
1) Control block balance: The control block of APG is
closely linked to the control block of Germany, due to high
transmission capacities and mutual benefits in the generation

mixes the two block are able to share. In particular, the pumped
hydro facilities within the APG area serve as buffer for low
Fig. 5. Trend of the block balance of APG in dependence on the
wind power
penetration and the fictitious load in Germany. Due to
computational reasons,
the three-dimensional fitting procedure cannot be applied with
suffcient
numerical reliability up to the full measured wind power
penetration of 40%.
price energy injected into the German block. The question
is if a general tendency for a relation between the wind
penetration in Germany and the export/import balance of APG
can be found. Since there are many different simultaneously
relevant influential variables, the method described in Sec. III
has been used for the identification of the influence of the

wind power penetration alone. The result is depicted in Fig. 4
and shows an interesting feature: The general trend indicates
that for zero wind penetration in Germany, there is even a
slight export to be expected from the APG block. For higher
penetrations, the sign turns to negative, that is, the consumers
and pumped hydro plants clearly start to import “green”
energy. Disintegrating this information to the influence of the
wind power penetration and the ficititious load in Germany
separately results in the trend surface plot shown in Fig. 5.
It shows the trends of the export/import balance of APG in
dependence on the wind power penetration and the fictitious
load in Germany at the same time.

2) Physical cross border flows: The cross-border flow
trends of the APG block have been identified following the
methodology outlined in Sec. III. The dimension reduction
part of the methodology has led to the identification of 2
modes explaining 60% of the variations in the data. The
resulting trend computations are shown in Figs. 6, 7 and 8.
The analysis indicates that for high wind penetration and low
load in Germany, the northern flowgates of APG (Germany and
Czech Republic) are importing considerably more energy than
in periods of low wind penetration and high load in Germany.
For the export flow at a southern flowgate to Switzerland (and
then further to Italy), the opposite is the case, which is a clear
indication of wind energy transit through the APG block.

3) Unintended cross border flows: The unintended cross-
border flow at a flowgate is defined as the difference between
the physical flow and the scheduled flow. It turns out that the
level of wind power penetration influences the probability of
unintended exchanges on APG’s borders. Again, the dimension
5
Fig. 6. Trend of the physical cross-border flow at the border
between APG
and Germany dependent on the wind power penetration and the
fictitious load
in Germany.
Fig. 7. Trend of the physical cross-border flow at the border
between
APG and Czech Republic dependent on the wind power
penetration and the
fictitious load in Germany.

2018 4th International Conference on Green Technology and Sust.docx

2018 4th International Conference on Green Technology and Sust.docx

Recommended

Recommended

More Related Content

Similar to 2018 4th International Conference on Green Technology and Sust.docx

Similar to 2018 4th International Conference on Green Technology and Sust.docx (20)

More from lorainedeserre

More from lorainedeserre (20)

Recently uploaded

Recently uploaded (20)

2018 4th International Conference on Green Technology and Sust.docx