Sustainable Energy Technologies and Assessments 26 (2018) 47–76

Contents lists available at ScienceDirect

Sustainable Energy Technologies and Assessments

journal homepage: www.elsevier.com/locate/seta

Original article

Frequency regulation capabilities in wind power plant T

⁎
Asma Aziz , Aman Than Oo, Alex Stojcevski
Deakin University, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: The design of frequency regulation services plays a vital role in automation and eventually reliable operation of
Synthetic inertia control power system at a satisfactory and stable level. Frequency response capability offered by wind plant is not same
Active power frequency control as the primary control capability of conventional plants, hence the integration of wind energy based generation
Automatic generation control at large scale has widespread impacts on power system stability and reliability. With the changing generation
Frequency control ancillary service
mix, modern electric power systems are facing a critical challenge in the real-time balancing of demand and
Forecasting
supply. This paper comprehensively reviews the various control functionalities available in wind energy systems
for supporting frequency regulation at different levels of frequency control services starting from inertial control
to the secondary control. An insight to new research challenges for better frequency control ancillary services in
wind integrated system is also provided. Though wind-based ancillary services are still in research and devel-
opment stage in most of the countries, future wind energy system participation is expected to contribute to
enhanced market efficiency, improved system reliability and macro economic benefits to all stakeholders.

Introduction
Present power system is undergoing several changes in its core
structure which are associated with the adoption of new power pro-
duction technologies and rapid integration of Renewable Energy
Sources (RES). There has been a tremendous increase in generated
megawatts by the RES all over the world including Australia. There
were 43 countries with Renewable Energy Target (RET) in 2005 which
rose to 164 countries in 2015 [1]. Table 1 provides a comparison of RET
for some of the countries. Proven and mature wind power technology
has profoundly penetrated energy matrix at global level, hence among
all available RES, maximum impact potential lies with wind energy at
largest scale. Wind is the most cost competitive renewable source of
electricity generation behind hydro. Share of wind and solar as primary
renewable generation in electricity production for some of the countries
in year 2016 is represented in Fig. 1. Energy is generated from wind in
79 countries around the world, and 24 of the countries have already
installed capacity of more than 1000 MW by 2013 [2]. Denmark leads
the world, followed by Sweden, Spain, and Germany [3] while Australia
ranks 11th in the world for wind generation per capita ahead of
countries like China and France. Among renewable energy sources,
wind power had the fastest growth in Australia; increasing on average
by 67% per year since 2000 [4]. There has been a steady increase in size
and output power of wind turbines. Australia’s first large-scale grid-
connected wind farm (at Crook well, New South Wales) in 1998 com-
prised eight 600 kW wind turbines each with a rotor diameter of 44 m
for a combined energy output of 4.8 MW [4]. Today most onshore wind
turbine generators have a capacity of 1.5–5 MW. The largest wind off-
shore turbines (IEC class S) at present is installed at Burbo Bank wind
farm, U.K. Each turbine has a capacity of 8 MW, rotor blades of 262 feet
length (80 m) and 195 m tower height [5] [6].
Design and operation of power system in presence of wind energy is
one of the major issues in wind power integration. Renewable energy
including wind power integration assessments are widely transformed
now since their starting stage in late 1970s and early 1980s [17]. Lit-
erature presents wide difference in the viable penetration level of the
intermittent generations in the power system. Technically, penetration
level of wind energy system is dependent upon existing generation
systems, their regulation capabilities, demand characteristic and cor-
relation with resources [17,18]. Electrical system modelling including
wind turbine technology and wind intermittency are the major factors
contributing to the system integration effects [19]. Low penetration
level and negative integration impacts are the strong reflection of
simplistic and conservative assumptions and data input for wind-speed
variations and spatial diversity in the wind integration studies [17].
Past studies raised the concerns that the integration of wind energy
systems at large penetration level will have widespread technical im-
pacts on power system stability and reliability. Studies showed that
wind energy integration effects on system frequency and power fluc-
tuation are nonzero and become more significant at higher sizes of
penetrations [19–22]. Type-I, Type II, and Type-III (without auxiliary
controls) wind turbines react weakly to frequency changes – hence,

⁎
Corresponding author.
E-mail address: asma.aziz@deakin.edu.au (A. Aziz).

http://dx.doi.org/10.1016/j.seta.2017.10.002
Received 2 August 2017; Received in revised form 9 September 2017; Accepted 9 October 2017
2213-1388/ © 2017 Elsevier Ltd. All rights reserved.
A. Aziz et al. Sustainable Energy Technologies and Assessments 26 (2018) 47–76

Nomenclature R speed regulation or droop of governor

R var variable droop
ACE Area Control Error Mact actual power margin available at a wind turbine
AGC Automatic Generation Control Mmax maximum power margin available among all wind tur-
AEMO Australian Energy Market Operator bines
EMS Energy Management System PAGC power output from automatic generation control
FCAS Frequency Control Ancillary Service Kpi proportional time constant of AGC controller
MPPT Maximum Power Point Tracking Ki integral time constant of AGC controller
ROCOF Rate of Change of Frequency NIA sum of the actual power flows on all tie lines or inter-
SCADA Supervisory Control and Data acquisition connectors
TSO Transmission System Operator NIS sum of the scheduled flows on all tie lines or inter-
VSWTG Variable Speed Wind Turbine System connectors
ROC Renewable Obligation Certificate B frequency bias setting
CPF Causer Pays Factor IME interchange (tie line) metering error
AEMC Australian Energy Market Commission T-ω curve torque-speed curve
AWEFS Australian Wind Forecasting System P-ω curve power-speed curve
NWP Numerical Weather Prediction Pref − del power set point for the de-loaded wind turbine
RTFDDA Four-Dimensional Data Assimilation ωdel deloaded wind turbine rotor speed
AEEGSI Autorità per l’energia elettrica, il gas e il sistema idrico ωmeas measured rotor speed for wind turbine
MSD Mercato Servizi di Dispacciamento ωopt optimum rotor speed
CPS Control Performance Standard Pmax maximum mechanical power from turbine
FLSR Frequency Limited Sensitive Response Pdel deloaded Power
FSR Frequency Sensitive Response K del deloading constant
WECC Western Electricity Coordinating Council K opt optimum mechanical power constant
FERC Federal Energy Regulatory Commission β0 pitch angle
RET Renewable Energy Target Δβ offset pitch angle
RES Renewable Energy Source Cp turbine power coefficient
ENTSO-E European Network of Transmission System Operators for Cp (λ opt,β ) coefficient of power at optimum value of tip speed ration
Electricity and pitch angle
UCTE Union for the Co-ordination of Transmission of Electricity λi tip speed ratio at ith time step
f frequency ρ air density
df
rate of change of frequency A turbine blades swept area
dt
Δf frequency deviation R blade length
HG inertia constant of conventional generating unit Vwind velocity of the incident wind
SG MVA rating of conventional generating unit KG system stiffness
W kinetic energy of rotating mass MG inertia of power supply
Hz/s Hertz per second Pavail available wind power
LP wind penetration level v wind speed
J moment of inertia of a rotating mass DR cmd derating command mode
M equivalent moment of inertia of all generators and the Pref reference power setpoint
motors connected to the grid Poperator transmission system operator commanded power set point
Pm generator mechanical power ωref reference angular rotor speed
Pe electrical load PAPC active power setpoint to turbine controller
Heq equivalent system inertia constant Pfinal − ord final power command provided to generator-converter
SB system MVA base model
SGi power rating of individual generating unit Lp wind penetration level
ΔPSG change in active power output of synchronous generators HWT frequency responsive wind plant inertia
in conventional generating plants Td total time delay associated with generating unit model
ωr angular speed p.u. per unit
Pprimary primary frequency control power Pinertia inertial power output from wind turbine

leads to smaller effective system inertia and degraded frequency re-
sponse [23–25]. Increased penetration of old technology based wind
generation in power system would, therefore necessitate a larger de-
pendence on regulation ancillary services to return to normal operating
conditions [26–28]. In contrast to technical impact studies, there are
several techno-economic and social-economic modelling based recent
research supporting 100% or near 100% renewable energy integration
to grid [28–31]. 100% renewable energy for 139 countries including
USA by 2050 is shown feasible by electrification of all applications
which will reduce demand by 42% [32,33]. Even though most of these
studies affirms that grid stability and security could be maintained
under 100% renewable energy scenario by stronger network inter-
connection [29] and coordination with flexible loads like electric car
[32,33], energy storage schemes [29,30] and efficient demand response
[33,34], the result needs to be supported with satisfactory analysis of
technical and economic implications. It is noticed from some of these
studies that wind integration simulation studies based on hourly wind
load data and production cost reflect higher penetration level while
absence of cost production models for short term operational concerns
like load frequency control gives lower penetration level.
Despite maximum elimination of interfacing concerns for wind en-
ergy and reducing technology costs, actual wind integration at this
large scale is still challenging globally. Collective evaluation of funda-
mental technical, economic and regulatory challenges in a consistent
framework is a requirement to ensure a safe, reliable, affordable, and
sustainable future energy system. Given this discussion, this study aims

48
A. Aziz et al.
to support wind energy system integration at large scale by evaluating
its technical functionalities and highlight latest challenges. This paper
reviews the concept of frequency regulation capabilities from wind
energy systems. The brief outline of the paper is as follows: Section 2
briefly introduces the idea of the frequency regulation services and
their nomenclature in various countries. A detailed review of various
frequency control functionalities available in wind power plants is
given in section 3. Final section 4 discusses some challenges which need
more research and implementation for better frequency regulation
services participation from wind power plants.
Introduction to frequency regulation in power system
Load/Demand is the primary independent control variable in a
power system. Rest of the dependent variables and controllable power
system elements respond to serve the continuously varying load. System
frequency is one of the dependent variable which indicates the system’s
energy balance and needs to be regulated for the stable operation. Low
frequency indicates the load increment while high frequency represents
the high generation. System frequency is controlled by injecting active
power into electric grid through power plants. This control is indis-
pensable for stable operation of network ensuring continuous adapta-
tion of generation to demand. Frequency below a certain prescribed
level will deteriorate power quality and may eventually result in system
outage. Any investigation of system frequency response requires fre-
quency regulation analysis at three different stages followed lastly by
an emergency contingency regulation service.
Fig. 2 represents a typical frequency regulation scenario under
contingency event which leads system frequency to cross governor dead
band and start deviating up to minimum limit defined as nadir point.
Normal power imbalance between generation and consumption is in-
stantaneously covered through system synchronous generator’s natural
inertial response at first stage followed by same synchronous gen-
erator’s governor response. The generator inertia is regarded as the first
level of frequency control reserve, and it has the characteristics of a
differential controller [36]. Since inertia is proportional to the rotating
mass of generators; greater the inertia, the lesser will be acceleration,
and smaller will be the frequency deviation. The conventional power
plants which remain connected to grid even in case of frequency de-
viations accomplish active power-frequency control through turbine
governor’s reaction to nominal synchronous speed variations and the
respective boilers-turbine frequency responsive controls. Under broad
range of ambient conditions, frequency sensitive power plants are ex-
pected to provide automatic power variation within a given time frame
and ramp limit to control the frequency drop. Frequency nadir point
achieved just after contingency event and the frequency settling time is
highly dependent upon the generator’s droop characteristic. A power
plant with higher droop and slower governor response will lead to
lower nadir point during a contingency. After governor based primary
Table 1
Renewable energy target comparison.
Country Renewable energy target (share of electricity in 2020 unless specified)
Australia [7] 23.5%
Brazil [7] 23% by 2030
Canada [8] Province-level targets (projected national share: 70%)
China [9] 15% final energy by 2015 Tech-specific targets; 200 GW wind
Denmark [10] 100% by 2035; 50% by wind
France [11] 27% by 2020 and 32% by 2030
Germany [1] 40–45% by 2025
India [12] 57% by 2027
S-Africa [13] 21% by 2030
U.K. [14] 30%
USA [15] 30% by 2030
EU [16] 20%
49
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
response, a comparatively slower; secondary controller generally
termed as Automatic Generation Control (AGC) is activated to bring
frequency to the set point. There are three different schemes for sec-
ondary AGC control as shown in Fig. 3. Some country like Australia has
centralized AGC where single transmission system operator (TSO)
controls all synchronous control areas. Pluralistic secondary control
implements decentralized AGC where multiple TSOs regulate multiple
control areas. Europe is an example where various countries are con-
trolled through their respective TSOs and follow same guidelines for
regulation. USA presents hierarchical secondary control where regional
transmission organization (RTO) controls multiple states and in-
dependent system operator (ISO) usually controls single state, and both
are governed and regulated by the Federal Energy Regulatory Com-
mission (FERC) [37].
Tertiary control is the centralized and manual control utilized by
TSOs in case the secondary reserves are saturated or exhausted before
stabilizing the system frequency. Tertiary control is applied to guar-
antee the timely provision of the adequate secondary control reserves
and economical distribution of the optimal secondary power control
set-points to various generators. Variation of generators set-points or
load participation is done automatically or manually through tertiary
control. In case of contingency event, tertiary control is also im-
plemented to restore the secondary control. Tertiary control may in-
volve power connection tripping, power redistribution, changing of
load control and interconnector power variation through changing
constraint equations. AGC and Tertiary control requires both increasing
or decreasing power. Load shedding, peak load generating units like gas
turbines or pumped storage units in turbine operative mode provide
tertiary control for power increment while pumped storage in pumping
mode is used for power decreasing stage. Tertiary control may extend
more than the range of secondary control restoration which is around
15–30 min in most grid codes while network optimization after a
contingency event may take longer duration [38]. Fig. 4 presents the
timing ranges of all three controls where the overlapping of primary
with secondary can be noticed and tertiary control extending ahead of
secondary control due to system optimization.
Regulation services nomenclature
Different countries have adopted different nomenclature for these
three frequency regulation services. Table 2 provides the terminology
adopted in various countries for generation demand regulation. For
example, European Network of Transmission System Operators for
Electricity (ENTSO-E) [39] has proposed more explicit terminology for
regulation services: Frequency containment reserve services, Frequency
restoration reserve services and Tertiary replacement reserve while
some countries like Australia [50] and U.K [42] have no explicit re-
presentation for tertiary control
Share of wind and solar in electricity production (2016)
7.91%
5.96%
5.6%
5.39%
> 40%
5.49%
17.99%
4.28%
2.08%
14.11%
7.06%
13.22%
A. Aziz et al.
Fig. 1. Percentage of wind and solar generation share in electricity production for some of the countries in year 2016, Data derived from [35].
Wind power plant participation in frequency regulation
Frequency regulation has become more challenging under the
conditions of low power-system inertia due to the high wind energy
penetration in the power system. Renewable energy sources like wind
having new production and nonlinear control technology imply new
energy management system (EMS) in modern power systems. This
system is indispensable for stable operation of grid ensuring continuous
adaptation of generation to demand. Under broad range of ambient
conditions, frequency sensitive power plants are expected to provide
automatic power variation with frequency drop within a given time
frame and ramp limit. The conventional power plants which remain
connected to grid even in case of frequency deviations accomplish ac-
tive power-frequency control through turbine governor’s reaction to
nominal synchronous speed deviations and the respective boilers-tur-
bine frequency responsive controls. For emulated synchronous gen-
erator like response, parameters like droop, dead-band, and headroom/
reserve power are needed to be considered in variable speed wind
turbine generator (VSWTG) modelling for frequency-active power
control. Wind turbine active power frequency regulation analysis is
50
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
receiving considerable research focus from academics and industry.
Wind plant capability to participate in frequency control ancillary
services has been analyzed based upon various technical and opera-
tional functionalities [52]. Among various functionalities, active power
control is the most important feature for frequency response from wind
power plant. This is followed by droop based frequency sensitivity and
power set-points processing and management.
Fig. 5 highlights related concepts available in scientific literatures
which have been applied for power-frequency regulation capability of
wind energy systems for different levels of frequency regulation. These
functionalities have been investigated either at individual wind turbine
(WT) level or wind farm level [53–58].
Following next sections discuss all relevant concepts related to the
frequency regulation capabilities of wind power plants.
System inertia in presence of wind penetration
Conventional power system operation is based around traditional
synchronous generators operating in synchronism, i.e. with the same
frequency (50/60 Hz), with a lot of kinetic energy stored in their
Fig. 2. Regulation services during frequency variation.
A. Aziz et al. Sustainable Energy Technologies and Assessments 26 (2018) 47–76

Fig. 3. Secondary control schemes.

rotating masses. Angular velocity of these synchronous generators is the mechanical power supplied by the generator (Pm) and the electric
proportional to system frequency. Kinetic energy which is stored in power demand (PL). Rate of change of frequency (ROCOF), denoted as
df
rotating masses provides a cushion for the momentary imbalance of , is thus directly related to system power imbalance and system in-
dt
power. This stored rotating energy in the synchronous generator is re- ertia. On the system MVA base value Ssys , equivalent system inertia
ferred as Inertia having the ability to oppose changes in frequency and constant (Heq ) based on conventional generation can be calculated as:
represented by inertia constant as:
SGi
HG =
Wi Heq = ∑ i = total generating units HGi *
Ssys (4)
SG (1)

1 where HGi and SGi are the inertia rating and power rating of in-
W= J (2πf )2
2 (2) dividual generating units.The system inertia constant depends on two
factors which includes the number of operating generators and the in-
where H is inertia constant in MWs/MVA denoting the time duration
ertia constant of each of these generators. A larger equivalent system
during which the generating machine can supply its rated power (SG ) in
inertia Heq is preferable as it signifies a higher amount of stored kinetic
MVA exclusively with its stored kinetic energy (W ) in MJ or MW s, J is
energy. Increasing ROCOF due to reduced system inertia have sig-
the moment of inertia of rotating mass in kg-m2 and f is system fre-
nificant effect on system frequency regulation. Inertia constant per unit
quency.
base MVA for different power plants and synchronous condensers is
dW df 2HSG df represented in Fig. 6. Single wind turbine has inertia constant value in
= J (2π )2*f * = = (Pm−PL)
dt dt f dt (3) the range of 4–6 s which is comparable to hydro and fossil steam based
plants in similar MVA ratings range. An inversely proportional trend
The change in kinetic energy leads to classical swing equation as between inertia constant and machine rating can also be observed for
shown in Eq. (3), which describes the inertial response of synchronous power plants.
generator as change in frequency following a power imbalance between

Fig. 4. Time range of various frequency control services.

51
A. Aziz et al. Sustainable Energy Technologies and Assessments 26 (2018) 47–76

Table 2
Regulation services nomenclature adopted by various countries. (See below-mentioned references for further information.)

Wind integrated system inertia calculations to system, thus decoupling generator speed with system frequency.
System inertia decreases with increasing wind power plant pene- Literature presents following three different equations for calculating
tration as these power electronics based units are not directly connected power system inertia in the presence of wind plant.

52
A. Aziz et al. Sustainable Energy Technologies and Assessments 26 (2018) 47–76

Fig. 5. Active power frequency control research study line in wind based power plants.

Fig. 6. Inertia constant values of various plants, data derived from

[59,60].

53
A. Aziz et al.
1 With Hi and SGi representing inertia rating and apparent power
rating of individual generating units, wind integrated system inertia
constant,Hsystem can be calculated as Eq. (5):
N responsive wind plant integration. Fig. 7 gives system frequency de-
∑i SG,i ∗Hi
Hsystem =
Ssystem (5)
N quency. It can be seen in Fig. 8 that frequency deviation reduces from
where Ssystem = SG,i + Sn,wind
∑i
2 Assuming a fraction of frequency responsive wind power plant
(z wind) contribution to system inertia by displacing a fraction of
conventional generation (zconv − gen) , new system inertia can be cal-
culated as Eq. (6) [61]:
Hnew = Hold (1 + zwind−z conv − gen ) (6)
3 Wind integrated system modified inertia constant,Heq using delay
theory of turbine governor [62,63] can be calculated as Eq. (7):
Heq = Hsystem (1−Lp) + HWT Lp (7)
where Hi and SiSi are the inertia rating and apparent power rating of
individual generating units. L p is wind penetrationlevel. Hsystem is cal-
N erators and type 3 and 4 WTGs. Fixed-speed WTGs have a lower cou-
culated as Eq. (5) with Ssystem = ∑i SG,i only. For frequency-responsive
wind plant, wind inertia contribution HWT to power system by pro-
viding ΔPe extra active power when the system is subjected to step load
disturbance ΔPL is calculated as Eqs. (8) and (9):
T D
d changes in grid frequency. It is however, possible to achieve short term
− lnX −2Heq (1−Lp)
HWT =
2Lp (8)
Td D
X=e
−
2Heq (1 − Lp) ⎛1− ΔPe Lp ⎞ + ΔPe Lp
⎜ ⎟
⎝ ΔPL ⎠ ΔPL (9)
Total time delay Td associated with hydro governor turbine model is
calculated on the basis of delay theory originally given in [63]. Sum-
mation of governor time constant, valve motion delay time and turbine
response delay time results in total time delay Td at which minimum
frequency deviation occurs after system disturbance. Value of Td is
calculated as 3.7814 s for hydro governor-turbine model with system
parameters given in [62].
Following third method of wind integrated system inertia calcula-
tion; under the 0.1 p. u. load disturbance and 0.05 p. u. temporary
active power support from VSWTG, a power system control area with
equivalent inertia of 6 s sees this frequency support from a machine
with 3.4 s inertia constant. This inertia constant will change to 7.17 s if
0.1 p. u. of extra temporary active power support from VSWTG is as-
sumed. Considering these inertia values, it will not be wrong to infer
that with increasing power support from wind turbines, wind turbines
can phase out conventional generation inertia. Figs. 7–9 are the simu-
lated frequency deviation examples for a control area having thermal
54
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
and wind power plant integrated to it. System inertia is taken as H = 5
and droop D = 4%. System inertia under normal wind plant integration
is calculated by Eq. (5) while Eqs. (7)–(9) are applied for frequency
viation when the normal wind plant is integrated to the area. Effect of
increased penetration can be seen with increasing deviations in fre-
1.5 Hz to around 0.4 Hz at 60% penetration of frequency responsive
wind plants. Frequency deviations are further reduced to 0.2 Hz if fre-
quency responsive wind plants also participate in secondary control.
Frequency responsive wind plant model and secondary frequency
control models are presented in next sections of this paper.
Synthetic inertial control capabilities in VSWTG
Background
Wind turbine generators (WTGs) are categorized into fixed/semi
fixed-speed WTG (Type 1 and 2), and variable speed WTG (Type 3 and
4) from inertial response perspective. Inertial response of type 1 and
type 2 WTGs is slower and lower in comparison to synchronous gen-
pling of induction generator rotational speed to system frequency, and
thus a marginally smaller inertia constant. Types 3 and 4 VSWTGs
provide negligible inertia as the fast control of power electronic con-
verters maintains a practically constant output power irrespective of
inertial response from variable speed wind turbines by modifying its
control loop. The inertial response from wind turbine has been in-
troduced with different phrases such as synthetic inertial response [64],
inertial response [65,66], inertia emulation [56], short-term frequency
regulation [67], and temporary frequency control [62,68]. A segment
of VSWTG’s kinetic energy is transformed into electrical power through
the inertial control and transferred to grid through converters.
Concept of inertial control
The basic idea behind most of these inertial response studies is to
add an additional signal as shown in Fig. 10 to the speed or torque
controller of variable speed wind turbine. Auxiliary signal sensitive to
frequency appears as the reference power set point in VSWTG model
thereby momentary increasing the wind turbine output power. Practi-
cally, 5–10% of additional active power is possible in a time frame of
5–10 s through inertial control under the constraint of minimal rotor
speed [52]. Wind turbine inertial response is essentially energy neutral,
meaning that the period of increased power is followed by a period of
decreased power. Temporary frequency response or inertial response
involves injection of active power for a short duration which is followed
by a power decay and lastly a power retrieval stage such that wind
turbine returns to initial pre-event condition. All these stages are
Fig. 7. Frequency deviation in a control area for different level of
normal wind plant penetrations under 0.01 p. u. load disturbance.
A. Aziz et al.
depicted in Fig. 11. The inertia constant H of a wind turbine is in the
order of 2–6 s [65] which is comparable to traditional generation with
2–10 s [69]. Wind turbines inertial response is highly dependent upon
turbine torque and pitch control. The kinetic energy in the rotor is high
enough to support the frequency but must be released using control
actions in the converter active power control to mimic inertia
In the inertial response, a part of the kinetic energy stored in the
turbine is released by temporarily increasing the output power. As the
energy is released from the rotor, its speed decreases. Slow turbine
speed tends to reduce the aerodynamic lift, thereby lowering the de-
livered mechanical shaft torque and increasing electrical torque which
could aggravate the speed decline and may lead to aerodynamic stalling
of blades. To avoid stalling, the overproduction period is followed by a
period of underproduction where the turbine speed is increased, and
the energy in the rotating mass is restored. The inertia based fast fre-
quency response controllers differ from synchronous inertia in the sense
that they are designed to respond to low frequency and large con-
tingency event. Further, their speed of response can be tuned as per
requirement of system stability. Synthetic inertia provision is relatively
controllable depending upon converter control response speed as well
as the wind turbine loading conditions. Type 3 WTG speed can drop as
low as 0.7 p. u, depending on the controls installed, allowing up to 5.25
times more kinetic energy to be obtained compared to conventional
synchronous machines of similar ratings [70]. The kinetic energy
available from type-4 wind turbine is generally more than type-3 wind
turbines, due to their speed variation from zero to full speed.
Examples of wind turbine inertial control models
Three common virtual inertia models are shown in Fig.12. Inertia
droop model [56] takes absolute variation in frequency as well as rate
55
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
Fig. 8. Frequency deviation in a control area for different level of
frequency responsive wind plant penetrations under 0.01 p. u. load
disturbance.
of change of frequency as input while second virtual inertia model [60]
takes absolute variation in frequency as input signal. Power reference
signal in inertia-droop based first model is calculated as:
df
P ∗f = K Δf + Kin
dt (10)
where Kin is the inertia constant weighting frequency deviation deri-
vative and is equal to 2H. Combination of the df and droop control loop
dt
maintains the active power support over an extended response time. A
high pass filter is included, so that permanent frequency deviation does
not affect control while low pass filter reduced frequency measurement
noise. In second inertia control model, an additional power signal Pinertia
is provided to turbine control loop which can be described as:
KTs
Pinertia = min [ max((f −fdb ), 0) Plm]
(1 + Tl s )(1 + Ts ) (11)
where K is the controller gain, and Tl and T are low pass filter and
washout filter time constants associated with the power shaping and
power conditioning block respectively, Δf represents the frequency
deviation from set-point, fdb is the frequency deadband and is the
maximum power increase that can be requested. The response to low-
frequency deviations is ensured through implementing max function in
the control.
Third inertia control model is based on continuous measurement of
frequency [71]. Wind turbine control provides a predefined power in-
crement when grid frequency drops below a threshold or trigger fre-
quency. Additional power (Pinertia) supplied by wind turbine control is
linearly dependent upon frequency deviation from nominal frequency
as:
Fig. 9. Frequency deviation in a control area for different level of
normal wind plant penetrations under 0.01 p. u. load disturbance.
Both of the thermal and wind plant participates in secondary control
as per the participation factor decided on the basis of system inertia.
A. Aziz et al. Sustainable Energy Technologies and Assessments 26 (2018) 47–76

Fig. 10. Virtual inertia model.

finertia,trigger −f (t )
Pinertia (t ) = Pinertia,set
finertia,trigger −finertia,min (12)

The maximum power increment up to the 10% of the wind turbine

nominal power is defined as Pinertia,set while the trigger frequency at
which power increase commanded is determined by finertia,min . The in-
ertial power increase is implemented till the time t < tinertia,max and in-
ertial control ceases if the frequency moves ahead threshold such that:
finertia,return > finertia,trigger + 0.05Hz (13)

A receding horizon controller (RHC) is proposed in [72] to provide

Fig. 11. Temporary inertial response stages.
frequency regulation by wind power. This controller is proposed to
takes future load and wind variations into account and therefore claims
to optimally exploit the kinetic energy reserves of the wind turbine over
a certain time horizon to damp over frequencies. Wind power fore-
casting involves conversion of atmospheric forecasts into turbine power
output forecasts. Inaccurate measurement and forecasting may highly
affect wind turbine output and turbine life. 20% error in wind speed
forecasting may introduce around 41% error in wind power output
[157]. Wind power calculation through manufacturer provided curve
applies an over simplified approach where wind power is modelled

Fig. 12. Inertia and droop control models for frequency

response from VSWTG.

56
A. Aziz et al.
primarily as the cube function of hub-height wind speed alone while
practically other factors like wind shear and turbulence are also in-
volved [73,74]. A Fast coordination between wind turbine control
loops, wind farm controller and swift interaction between all related
input data is necessary for wind turbine frequency response. Inclusion
of load and wind forecasting puts a question on the practical applica-
tion of this controller as the proposed work is based on simplistic input
data for load and wind. Anticipation of actual wind energy at time
horizon less than 1 min is hard as wind power forecasting involves lot of
uncertainty due to spatial and temporal variability of wind fields and
different forecasting tools. Accurate wind power forecasts, related un-
certainty and their corresponding effect on wind turbine controller is
computationally challenging and require a multiscale simulation ap-
proach [74]. Integration of wind forecasting will significantly increase
the processing time of the turbine controller. Efficient and high fidelity
forecasting models relating power to atmospheric conditions may off
course benefit wind turbine manufacturers to refine their power curve
which would be beneficial in improving turbine controls and assessing
components fatigue. An enhanced inertial control scheme coordinated
with primary frequency regulation and conventional generators was
Fig. 13. Wind turbine generator generic model for primary frequency control. CP refers to the coefficient of wind turbine. More details can be found in [60].
57
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
proposed in [75] so that sudden load changes are nor mask and con-
ventional generators response in not delayed. Fast Droop control is
offered as a viable alternative to synthetic inertia control in wind pe-
netrated system, but the optimal selection between these two strategy
was shown to be highly dependent upon power system composition of
other generating units [76].
Inertial control limitations
Provision of repetitive inertial response leads to heat generation and
stress on turbine components due to aero-mechanical limits electrical
limits and speed ratings limits. The inertial control must respect elec-
trical and mechanical loading limits of wind turbine generators. The
inertial control must provide margin above stall, and is consequently
limited when the initial rotor speed is low. This means that the power
response is limited whenever the wind speed is at or below rated. At
low wind turbine power levels, the available energy of the inertial re-
sponse starts to decline rapidly below about 50% rated power, dropping
to zero below about 20% [77]. Wind turbines specially type 3, present a
limitation of power curtailment down to a level of 15–25% of the rated
power to avoid rotor stall [25]. While the issue of improving turbine
A. Aziz et al. Sustainable Energy Technologies and Assessments 26 (2018) 47–76

life, is being addressed at industrial level, wind turbine inertial control
operation only during large under frequency event which represent
security and continuity-of-service risks to the grid is the preferred op-
tion in many grid codes [40]. Delivery of synthetic inertia lowers the
wind turbine's rotor speed. Re-accelerating to optimal speed thereafter
absorbs some of the wind power that the turbine can export to the grid.
This power reduction can be up to 60% for some wind turbines ne-
cessitating extra reserve activation. This energy recovery phase delays
the grid’s frequency recovery. Post-inertia recovery could cause a
“double-dip” in system frequency, increasing the risk of triggering
protective relays at substations and causing blackouts [78].
Primary frequency control capabilities from VSWTG
Background
Following the inertial frequency response of the power system,
which reduces the ROCOF, the primary frequency response stabilizes
the frequency at a new value by increasing (or decreasing) the power
generation in proportion to the frequency variations. Primary frequency
control is a proportional control having system frequency as control
signal and it is independent of rate of change of frequency. Primary
frequency control is represented as:
1
Pprimary = − Δf
R (14)

Related grid codes scenario
TSOs have been working on grid code requirements for im-
plementation of a generic inertial response from wind power plants
(WPPs) [79]. Hydro-Québec (HQ) was the first TSO to include man-
datory inertial response from the wind turbines in its grid codes. Ac-
cording to the HQ’s revised grid code 2005 [80], all new wind turbines
must present the capability of providing a power increment up to 6
percent of their nominal capacity for 9 s within 1.5 s during low-fre-
quency events. With the installation of first synthetic inertia enabled
wind turbine in 2011, currently two-third of Quebec’s wind capacity is
shared by inertia-compliant turbines. In view of double-dip implica-
tions of inertial response from wind turbines, HQ plans to limit power
reduction during recovery to no more than 20 percent of a wind tur-
bine’s capacity. Transmission operators like Independent Electric
System Operator (IESO) in Ontario [81,82] and National do Sistema
Elétrico (ONS) in Brazil [83] have formulated similar grid codes re-
quiring active power control from wind farms > 50 MW and inertial
response respectively. Inertial capability was demonstrated for Spanish
grid by the ACCIONA Energy managed wind farms on 28 th Feb. 2016
whereby a power increment of more than 150 megawatts (MW) was
provided in response to an instruction given by Red Eléctrica de España
(REE) [84]. European Requirements for all Generators (RfG) also ad-
dress the topic of frequency regulation, allowing TSOs to seek synthetic
inertia response from WPPs [85]. The Irish TSO EirGrid [45] and
Australia TSO AEMO [86] have also identified the need for fast fre-
quency response in the form of inertia control in their grid in con-
sideration of renewable energy target for 2020.
where R is speed regulation or droop of governor and Δf is frequency
deviation from set point value. Droop values which indicate the re-
quired change in power output in response to changes in frequency are
defined in the grid codes, and vary in different power systems. In a
typical large power system, the power plants allocated to accomplish
primary frequency response should be capable of increasing their
output real-power within about 10 s of the pre-defined frequency var-
iation and should be capable of maintaining this response for a further
30 s. For example, in Nordic power system, the frequency controlled
normal operation reserve must be at least 600 MW and fully activated
for frequency deviations of ± 0.1 Hz. The frequency controlled dis-
turbance reserve must be at least 1000 MW and fully activated at
49.5 Hz [87]. In Union for the Co-ordination of Transmission of Elec-
tricity (UCTE), 50% of the primary control reserves must be activated
within 15 s, and 100% within 30 s [88]. In Australian NEM, regulation
raise/lower Frequency Control Ancillary Services (FCAS) of 130/
120 MW must be available for bringing frequency in restoration range
(49.5–50.5) within 5 min [50]. Primary frequency response from wind
turbine is the autonomous frequency sensitive active power response
analogous to governor controls for synchronous generation. Working in
coordination with supervisory control, wind turbine primary frequency
controls respond to substantial grid frequency deviations during low or
high grid frequency events by raising or lowering active power output
respectively.
VSWTG control loop for power set-point processing
Power set-points are target for required electrical energy genera-
tion. These targets can be based on available wind, operational mode or

Table 3
Ramp rate comparison for different power plants.

Type of PLANT Reference Minimum load [%PNom ] Average ramp rate Type of Plant References Minimum load Average ramp rate [%
[%PNom /min] [%PNom ] PNom permin]

OCGT [89] 40–50% 8–12% Integrated Gasification [91] 50 3–4%

Combined-Cycle
* *
20–50% ( State of the art) 10–15% ( State of
the art)

CCGT [89] 40–50% 2–4% Natural gas combined cycle [91] 40–50% 2–6%
30–40% (20% for 4–8% (*State of the
Sequential combustion) art)
(*State of the art)

Hard coal-fired [89] 25–40% 1.5–4% Compressed Air Energy [92] 50 4–10%
Storage Plant (262 MW)
25–40% (*State of the art) 3–6% (*State of the
art)

Lignite-fired power [89] 50–60% 1–2% Battery Energy Storage [92] 0 20%
Plant
*
35–50% ( State of the art) 2–6%

Pumped-Storage [90] 0–40% in turbine mode > 40% Nuclear [92] 50–80% 1–5%
Hydropower Plant
70–100% in pump mode
*
Wind [70] > 30% for Type 3 0.05–0.25 p. u./s State of the art refers to the plant which are currently available for installations
0–100% (state of the art) 20–30% per second

58
A. Aziz et al.
required frequency response. Active power set-points characteristics
and the expected behavior of the different generating units under dif-
ferent set-points are defined in grid codes. An accuracy up to 2% of
power set-point is usually required from wind power plants. Wind
power plants can be operated as grid code compatible unit by mod-
ifying its control loop. Like inertial control, active power based primary
control capability in wind plant also requires a coordination with
torque control and speed control loop of wind turbine model. Fig. 13
shows a type 3 wind turbine generator model [60] with a modified
torque speed control loop for achieving grid code compatible active
power-frequency response. A simpler torque-speed control loop of
VSWTG is also highlighted as an inset in the same Figure. The VSWTG
model is included here to present the interaction and coordination of
different power set-points inside turbine control loop. Rotor speed error
is provided as input to both torque and pitch control loops. Pitch
compensation block receives frequency responsive active power set-
point PAPC generated from frequency responsive controller. Power order
from torque controller (Pinp ) is altered by passing frequency responsive
active power set-point PAPC through the power response rate limiting
block. Final power order provided to generator converters model is
generated by summing active power set-point PAPC to the difference
between active power limited set-point power and power order set-
point from speed controller as represented in Eq. (15):
sTw ⎞
Pfinal − ord = PAPC + ⎛
⎜ ⎟ (PAPC−Pinp)
⎝ 1 + sTw ⎠ (15)
Additional power set-points for inertial control or droop control
when turbine is participating in a cluster can also be added to the
control loop. One of the major requirement of sufficient and faster
frequency response is generators ramp rate limit which is defined as
percentage of the rated power output of the generating unit per unit of
time. Normally the ramp rate is specified in MW per minute, MW/min
or in percentage of nominal load per minute, % PNom/ min . Nominal load
or nameplate capacity (PNom) is the steady state total power output of a
power plant functioning under design conditions while Pmin (%PNom)
represents the lowest total power which is deliverable under design
operating conditions. The ramp rate in presented VSWTG model is
implemented by including a washout filter whose time constant (Tw ) is
detrimental to rate limit imposed on changes in power order set-point.
Table 3 gives the typical ramp rate values of different power plants. As
can be seen, wind plants have comparatively faster fast ramp rate for
active power changes. Under favorable wind conditions, wind power
plants can increase their active power from minimum level to rated
level within 6–10 s [52]. A simulated example of power deviation in p.
u. MW in a control area with frequency responsive wind plant and
thermal plant is shown in Fig. 14. Frequency responsive wind plants are
capable of providing power output change comparable to thermal
59
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
plants under favorable wind conditions.
VSWTG operation for primary frequency response
Wind power plants can participate in frequency control by ramping
their output up and down as per the set droop characteristic for a
certain time-period. Wind turbines operating on the basis of maximum
power capture curve to extract maximum wind energy requires re-
placement to deloading power curve to save the available power as
reserves for the long-term negative frequency control. De-loaded wind
turbine operation provides a power margin for ramping capability of
wind turbines during negative frequency deviations. Most of the past
researches focus on modifying individual turbine control algorithm for
analyzing primary frequency control capabilities from wind turbines.
As represented in Fig. 4, wind plant primary control functionalities are
based on either achieving de-loaded performance [62] [93,94] to en-
hance VSWTG’s frequency response or implementing droop control for
frequency stability [95–98]. De-loading through rotational over speed
control [99–102] and pitch control has been discussed in various stu-
dies [103–105]. Both methods are based on the idea of achieving a non-
optimal working point in the torque–rotor-speed curve of the turbine as
shown in Fig. 15.
Concept of deloading VSWTG
In maximum power point tracking (MPPT) operating mode, turbine
active power set-point is determined such that:
0.5πR5ρ
Popt = K opt ωr3, where K opt = Cp (λ opt, β0)
λ 3opt (16)
When turbine switches from MPPT mode to commanded reserve
power mode, MPPT power set-point (Popt ) changes to deloaded power
reference (Pref − del ) and power-speed characteristic of wind turbine is
varied consequently. Basic concept behind various deloading method is
to operate wind turbine at deloaded power set-points achieved through
deloaded/reduced power of coefficient (CP,deload ) which is a function of
pitch angle (β ) and tip speed ratio of (λ ) turbine. Whole concept can be
expressed as: Achievement of CP,deload = f (λ,β ) , subjected to rotor speed
(ω) limitation ωmin ⩽ ωdel ⩽ ωmax and pitch angle limitation:
−βmin ⩽ β ⩽ βmax .
In rotor speed control deloading method as shown in Fig. 16, the
wind turbine controller changes the reference rotor speed towards op-
timal speed value such that generator active power is increased. Re-
ference power set-point to rotor converter control of VSWTG model is
retrieved from measured rotor speed, de-loading margin and torque-
speed (T−ω) or power speed (P−ω) curves. Under the assumption that
maximum turbine power and rotor speed are known, power set point
for de-loaded wind turbine can be calculated as:
Fig. 14. Comparison of power deviation for thermal-based control
area integrated with frequency responsive wind plant when both
participates in secondary control. 0.4 participation factor (PF) for
thermal and 0.6 PF for the wind is considered here.
A. Aziz et al. Sustainable Energy Technologies and Assessments 26 (2018) 47–76

Fig. 15. Representation of optimum and de-loaded curve for VSWTG.

Fig. 16. Control loop for de-loading control of VSWTG.

can be calculated by increasing pitch angle as given in Eq. (22):

Cp (λ opt , β ) = (100−K del )% × Cp,opt (λ opt , βmin ) (22)

Fig. 18 gives an example of pitch angle offset variation with respect

to de-loaded power serve variation while Fig. 19 presents the zoomed
frequency nadir point obtained under 10% and 20% pitch de-loaded
VSWTG operation. An improvement in frequency nadir point with in-
creased deloading can be noticed in the figure. Besides the precise
setting of pitch angle offset values, fast servo time constants are also
important for the deep analysis of this method. Two-dimensional

Fig. 17. Pitch droop de-loading control.

(ω −ω )
Pref − del = ⎡ del meas ⎤ (Pmax −Pdel ) + Pdel
⎢ (ωdel−ωopt ) ⎥
⎣ ⎦ (17)

Pdel = K del Pmax (18)

%del
K del = 1−
100 (19)
3
Pmax = K optmax ωopt (20)

1 CPopt
K opt = ⎛ ⎞ ρ ⎜⎛ 3 ⎞⎟ πR5
⎝ 2 ⎠ ⎝ λopm ⎠ (21)

In pitch controller based deloading method, which was first pro-

posed in [103] and shown in Fig. 17, pitch angle is regulated as a
summation of pitch angle (β0) from pitch controller and an offset pitch
angle (Δβ ) such that βref = β0 + Δβ and Δβ = K1 Δf . Applied pitch offset
value (K1) can depend upon the droop characteristics, active power
Fig. 18. Pitch angle offset variation with de-loaded power reserve in VSWTG.
reference or de-loading power. De-loaded VSWTG by certain value K del

60
A. Aziz et al.
Fig. 20. Frequency Grid Code UK [111].
lookup table dependent rotor speed and pitch control based hybrid de-
loading control is presented in [106]. Rotor over speeding technique is
applied for lower rotor speed while for rotor speed near or above rated
value, pitch control is applied. This method has a limitation in the form
of complex lookup table processing for non-linear Cp calculations and
requires extremely fast coordination between speed controller and
comparatively slow acting pitch controller. A model reduction based
control architecture dependent upon integrated pitch angle and power
regulation is proposed in [107] for primary control during sub-syn-
chronous and super-synchronous operation of the wind turbines. Pro-
posed methodology has practical limitation due to the necessity of in-
cluding wind measurement and online model tuning.
Deloading methods limitations
With all type of deloading methods, wind farm participation in
primary frequency regulation is improved but this leads to reduction of
wind farm annual capacity factor. Deloading wind turbines by a fixed
percentage may require rotor over speeding and can have detrimental
effect on turbine life. Moreover, maintenance of rotor speed within
rated limit also requires a strong coordination with wind speed mea-
surement. Pitch control methods are preferable in medium and high
wind speed but lower in comparison to over-speeding method which is
preferable in low wind speed. Combination of pitch and over speed
control is preferable in medium wind speed. Including wind speed
measurement is highly questionable for determining power set-points
61
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
Fig. 19. System frequency variation with pitch droop de-loaded
VSWTG.
for wind turbine investigations as inaccurate measurement may highly
affect wind turbine output and turbine life. Owing to high dependence
on available power reserves upon maximum allowable rotor over-
speed, de-loading of wind turbine currently finds the limited practical
application. It is highly important to investigate and then adopt proper
methodology for deloading operation for improving turbine life and
capacity factor of wind farms.
Static and dynamic droop control for VSWTG primary frequency response
An important requisite for stability of VSWTG during primary fre-
quency response provision is that its primary frequency power con-
tribution should always be sufficiently lower than the available power
margin. Both static and variable droop control concept has been ex-
plored in literatures for primary frequency response of individual wind
turbine. Fixed power response is achieved by implementing static fixed
droop value as given by equation 11. Variable power response is
achievable through dynamic droop value which can be calculated based
on wind and power reserve conditions [108] or by ROCOF [109]. Under
variable droop setting as given by Eq. (23), optimum wind turbine
operation was proposed such that wind turbine having highest power
reserve is assigned lowest droop value so that it gets maximum parti-
cipation in primary frequency response.
M −Mmin ⎤
Rvar = Rmax −(Rmax −Rmin)*⎡ act
⎢ Mmax −Mmin ⎦
⎣ ⎥ (23)
Here M denotes power margin which is calculated as power margin:
M = PMPPT −Pout . So Mmax denotes maximum power margin available
among all wind turbines while Mact is the actual power margin available
at a wind turbine. Dynamic droop leads to achieving improved system
stability and avoiding reserve exhaustion. Another study presented a
control system for active power control of wind turbine by tracking a
power reference and provides a primary frequency response with con-
stant and dynamic droop in absence of any dead-band with diminishing
turbine structure loading [110]. Simulations were conducted on NREL
CART 3 turbine with a simple generator rather than a variable wind
turbine generator model. Communication link has an important role in
droop response as there can be a delay of 1s at wind turbine level while
there can be an additional delay up to 2s at wind farm level.
Few related grid codes
Grid codes like that of U.K. [41,42] requires a response from wind
farms under normal conditions and limited up/down response under
high-frequency conditions. The response to a deviation should begin
within two seconds, and the wind farm should be capable of achieving
up to 10% response at 10 s for a 0.5 Hz ramp over 10 s [111]. Fig. 20
A. Aziz et al.
shows the example of UK grid code which requires two types of fre-
quency response from wind farms:
1. Frequency Sensitive Response (FSR) Mode: Wind farm power
changes proportionally to both up and down frequency deviation
from reference frequency signal according to set droop character-
istic (3–5%). The frequency sensitive mode is the generic term for
describing primary control and/or secondary control for negative
frequency deviation. The wind plant should be capable of being in
FSM within 2minutes of receiving the instruction.
2. Frequency Limited Sensitive Response (FLSR) Mode: There should not
be any power variation corresponding to any frequency changes ex-
cept when the system frequency exceeds set an upper limit of 50.4 Hz.
This mode supports high frequency response during positive fre-
quency deviations such that the active power output is reduced by a
minimum of 2% of output for every 0.1 Hz rise above 50.4 Hz.
Fig. 21 shows the active power set-point for wind turbines as per
frequency sensitive and frequency limited sensitive response and the
corresponding frequency. Frequency sensitive response provides both
up and down power set-point as per the system frequency. Ireland grid
codes follow UK grid code for deloaded wind turbine operation and
ENTSO-E grid code have also indicated the need of deloaded operation
for primary control. Being a member of ENTSO-E, Germany receive
support for primary control reserves from highly-interconnected net-
work with an established integrated electricity market with 43 TSO’s
from 36 countries across Europe [112]. This allows trading across
borders and high synchronous generation from neighboring countries
help to regulate and support frequency deviations within 30 s in Ger-
many. Hence, country like Germany [40] have grid codes that do not
require deloaded operation but require power curtailment in case fre-
quency rises. According to Australia’s national electricity market
(NEM), it is compulsory for generators > =30 MW to participate in
FCAS [113] [114] though none of the wind farms participate in it. In
62
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
Fig. 21. Active power set-point for wind turbines as per UK grid code.
System frequency used to depict active power set-point for wind farms
is shown below.
view of changing requirements from transmission system operators due
to diverse generation portfolio, future wind farms will be required to
participate in primary response for islanded and interconnected grids.
Wind power plants participation in secondary frequency control/AGC
The AGC problem has been extensively studied during the last four
decades and it has been one of the most highlighted issues in the design
and operation of independent and interconnected power systems.
Preponderance on renewable sources like wind energy that demand
different management as compared to conventional system, has moved
the electricity grid in a transition phase to a new model. Automatic
Generation Control (AGC) is one of the major units of EMS in con-
temporary electrical grid dealing with new production and nonlinear
control technologies [115]. The analysis and design of AGC system in a
more deregulated complex environment plays a vital role in automation
and eventually reliable operation of power system at a satisfactory
level. The secondary frequency Control or AGC represents an important
tool for regulating FCAS dispatch and distributing the imbalance in
energy market between designated generating units. AGC which is
slower than the two-aforementioned inertial and primary responses,
regulates the frequency at its nominal value. It is the IEEE working
group which provided the standard definition for power system AGC
[116]. AGC as defined by IEEE is the regulation of the power output of
electric generators within a prescribed area in response to changes in
system frequency, tie-line loading, or the regulation of these to each
other, to maintain the scheduled system frequency and/or the estab-
lished interchange with other areas within predetermined limits.
Conventional AGC model
Load-Frequency Control (LFC) is used as part of AGC to maintain a
constant frequency and to regulate tie line flows. It has made the op-
eration of interconnected systems possible and today it is still the basis
of any advanced concept for guidance of large systems. Secondary
A. Aziz et al.
Fig. 22. Basic loop of AGC control.
reserve requires the capability of increasing the output real-power
within about 30 s and maintaining the response for a further 30 min.
LFC or AGC is implemented usually through Proportional-Integral (PI)
or Integral controller as shown in Fig. 22 and mathematically written as
Eq. (24) below.
1 in most of the wind farms.
ΔPAGC = −(Kpi + ) ΔACEi
sKi (24)
where ΔPAGC is the AGC controller correcting power set-point which is
applied to governing generators in the control area, ΔACEi is the area
control error for an area i. Kpi and Ki are gain of AGC controller. The
AGC controller gain settings follow a common guideline for pool op-
eration in the synchronous area. ACE is the real-time representation of
utility’s obligation to regulatory authority for maintaining net power
inflow/outflow along with a small bias obligation to maintain fre-
quency. AGC participating Generators are adjusted every 2 or 4 s such
that the ACE is close to zero. The ACE is calculated as per the following
Equation 25:
ACE = (NIA−NIS )−10B (FA−FS )−IME (25)
where, ACE is the Area Control Error in MW. Negative values denote a
condition of under generation and positive values denote over gen-
eration. NIA is the algebraic sum of the actual power flows on all tie
lines or interconnectors (MW). NIS is the algebraic sum of the scheduled
flows on all tie lines or interconnectors (MW). It is defined as the mu-
tually prearranged intended net MW flow on the area tie lines, in-
cluding dynamic schedules. NIA − NIS is basically tie-line interchange
error or flat tie-line control which compares the net tie-line actual in-
flow/outflow with power scheduled to be sold or bought. Flow into a
system is defined as negative while flow out as positive value. B is the
frequency bias setting (MW/0.1 Hz). Frequency bias reflects the ability
of any control area to respond to frequency disturbance under the
utility’s obligation to generate or absorb power to regulate the total
system frequency. FA is the actual frequency (Hz). FS is the scheduled
frequency (Hz). IME is Interchange (tie line) Metering Error which is
accounted from meters installed at end of tie-lines. Time error is the
accumulation of frequency deviation over a defined time-period. When
time error reaches ± 10 s, the Interconnection Time Monitor initiates a
time correction. More advance formula for ACE can be found from
63
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
WECC [117].
Wind AGC model
It has been indicated by recent investigations that large penetration
of wind energy has an impact on modern power system frequency
regulation along with AGC systems and other control operation issues
[22]. Hydro turbine governors require large transient droop compen-
sation because of opposite transient power change of the turbine due to
gate opening at penstock [69]. Because of this phenomena, during a
generation deficit situation, the decelerating power (energy) is higher
for a hydro turbine compared to that of a steam turbine with/without
the reheat. This results in long resetting time for stable speed operation
and give a relatively slow response. For a hydro plant dominated
system, frequency settling time may increase in case of large penetra-
tion of intermittent wind generation. However, it also provides a op-
portunity for fast active power support from frequency responsive wind
plants to arrest initial frequency decline. Steam and gas turbine output
is highly dependent upon fuel and load limit. Wind power fluctuation
during the conditions of high load and low fuel can have detrimental
effect on system frequency. North American Electric Reliability Cor-
poration (NERC) has enacted control performance standard (CPS) for
evaluating frequency control performance in a control area. Control
areas must not be less than 95% compliant with CPS1 and no less than
85% compliant with CPS2 [118]. CPS1 and CPS2 deteriorate with in-
creasing wind penetration and this effect is observed more for large
interconnected systems Application of several computational techni-
ques have been researched for the frequency regulation in conventional
power system; however, the large integration of wind energy system
has created a gap in theory and application. AGC simulation model
assumes a direct physical connection for wind power generators set
point change and secondary control signal provided by the transmission
system operators. This direct connection is still not practically available
Fig. 23 gives a generalized AGC simulation model for a control area
with n conventional generator units and wind power plants. The
parameters of Fig. 23 are defined as follows: Δf is the frequency de-
viation, ΔPM is the mechanical power output from turbine unit, ΔPC is
the supplementary regulation command set-point, ΔPL is the load dis-
turbance, Hsys is the equivalent inertia constant of the control area, Dsys
is the equivalent damping coefficient, B is the frequency bias, Ri is the
droop constant, ΔPP is the primary control power command, ∝i is the
LFC participation factors, ΔPWP is the wind power, and Mi (s ) is the
governor-turbine model. TSOs control the amount of wind power to be
captured by sending control signals to AGC capable wind plants. During
the frequency deviation, due to the load/generation disturbance, the
AGC regulation mechanism generates raise or lower power commands
for participating generating units as per their participating factor.
Practically the participation factor for generating unit is a complex
problem involving economic dispatch and unit commitment, it can be
simply expressed in simulation studies in terms of the defined partici-
pation factors. Wind farm participation in AGC is highly dependent
upon suitable wind conditions. This uncertainty in AGC participation is
compensated by conventional generation units. Assuming n generating
units in AGC system with n-1 being conventional unit and one wind
farm, participation factor distribution can be obtained as:
∝2 + ∝3 + ⋯.+∝n = 1−∝W (26)
where ∝i , i = 2, …, n are the participation factors associated with the
conventional units and ∝W is the wind farm's participation factor. Un-
certainty in wind farm participation can be represented as per pene-
tration factor Lp as:
∝W = Lp−γ (27)
where γ is any random parameter between zero and γ . The aim of the
AGC controller is to drive the variations in tie-lines power and fre-
quency to steady state value by matching generation with demand. In a
A. Aziz et al.
Fig. 23. Generalized AGC simulation model considering renewable power regulation [119].
high wind penetrated power system, with varying wind power and load
uncertainties, there is a spatial variation of frequency standards and
required regulation over the control area. So, for wind penetrated
control area, traditional lumped ACE changes to distributed ACE and
conventional AGC changes to spatially differentiated frequency reg-
ulation. The control area ACE reflect the wind farm total power var-
iations which is usually smooth in comparison to individual wind tur-
bine power. ACE signal representing generated wind power changes can
be given as Equation 28 [119]:
ACE = B Δf + (ΔPtie − con + ΔPtie,WTG )−IME
where,
ΔPtie,act = ∑ (Ptie − con,act −Ptie − con,sched )
ΔPtie,WTG = ∑ (Ptie − WTG,act−Ptie − WTG,estim)
Ptie − con,act is actual conventional generator power and Ptie − con,sched is
scheduled power from conventional generator. Ptie − WTG,act is actual wind
power and Ptie − WTG,estim is estimated maximum wind generation which is
included in central dispatch process. BΔf is frequency bias setting
multiplied with frequency changes between scheduled and actual
power. The forecasted wind energy is generally based upon the number
of available generators, wind measurement, and the current generation
level. Wind farm real time supervisory control and data acquisition
(SCADA) data and metrological SCADA data from the turbine nacelle
are provided for generating wind farm power forecasting into dispatch
system. Generation forecasts for individual unconstraint semi-sched-
uled wind generators dictates the generation capacity of a wind
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
generator that is available for dispatch (without consideration of net-
work limitations, price bids etc.). For an unconstraint wind farm, un-
constraint intermittent generation forecast (UIGF) is based upon actual
MW output received through SCADA system while in case of constraint
wind farm, generation forecast is based on meteorological data, number
of available wind turbine, wind speed and MW set point [120]. Due to
spatially distributed wind farms having variable wind resources, single
wind turbine generator output cannot be held representative of net
generation output of a wind farm, so estimated power has to be in-
cluded in dispatch rules. An inaccurate forecasted wind generation may
(28)
hamper the dispatch operation of energy market due to increased area
control error.
(29) Wind-AGC research
Lots of research is underway to review and analyze the wind energy
association with AGC performance criteria, capabilities, and technolo-
(30)
gies for ensuring proper system performance. The conventional LFC
designs based on classical control have less adaptability, are not very
efficient and are therefore not suitable for modern power systems with
large interconnections, varied structure and more penetration of re-
newable energy sources, specifically wind energy with varied uncertain
source and urgent demand for power quality. It is expected that ap-
plication of modern robust, automatic and intelligent control techni-
ques to the LFC schemes will be more adaptive and appealing approach
for wind based power system to overcome impositions set up by tech-
nical standards of energy market regulators. AGC implementation in
wind energy system is studied basically at two levels: at wind farm level
and wind turbine level. Fig. 24 presents the basic concept of AGC im-
plementation in wind-based power plant. To bring control area
64
A. Aziz et al.
Fig. 24. Basic schematic diagram for wind farm frequency control.
frequency to required set point level, TSO sends reference power set-
points to wind farm controller which in turn dispatches demanded
power set points to individual wind turbines. A more detailed re-
presentation can be seen in Fig. 25 where TSO can provide different
types of power set-point (balance, ramp or delta) as per system need.
Dead band and droop settings are provided by TSO. Basic idea behind
this model is to replace MPPT power set-point to wind turbine con-
troller with power set-point from wind farm controller while MPPT
power output is sent to controller as instantaneous available power. In
case of no power limitation from TSO, same MPPT power set-point is
provide to turbine controller. Active power set-point is primarily con-
strained by the available wind and different power commands re-
garding de-rating power or reserve power from TSO. Three de-rating
command mode were defined in [122] as:
DR cmdPrated, Pavail−(1−DR cmd) Prated and DR cmdPavail
Fig. 25. VSWTG based wind farm controller supporting AGC [121].
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
In another study, three different operating modes (de-rated, abso-
lute reserve and spinning reserve) [123] are defined such that operating
mode is de-rated or normal where:
Pref = Poperator, if Poperator ⩽ Pavail (29)
Pref = Pavail, if Pavail ⩽ Poperator (30)
A WPP running as a spinning reserve will produce less power at all
wind speeds; thereby always providing a power reserve. Frequency
controller will act as spinning reserve controller or de-rating controller
to vary the WPP grid power production as per transmission system
operator request. During wind farm operation, turbine set-point can be
multiplied by active power demand which is provided by farm con-
troller to generated frequency responsive power demand.
Even though wind farm operators strive for maximum power
output, wind-AGC especially downward regulation is highly dependent
upon grid code implementation. For example, Irish grid code implies all
grid integrated generating units to operate at a reduced output level.
Fig. 26 is an simulated example of various power set-point generated
through a wind turbine frequency response controller shown in Fig. 13,
at variable wind speed and fluctuating grid frequency. The blue color is
the power set-point required as per a specific grid code while green is
the available wind power. This available power is curtailed according to
desired command mode thereby providing a reserve power which can
be used to control frequency deviations. We can notice an increase in
generated electrical power from 1 per unit to 1.1 per unit when the
turbine is working in reserve mode. Fig. 27 presents the simulated re-
sults for a control area frequency response when wind plant is in-
tegrated to it with different frequency control capabilities. Highest
deviation and longest settling time are observed when wind plant is just
feeding power to control area and smallest deviation with lowest set-
tling time is observed for control area without wind plant. Grid code
compatible wind plant with automatic generation control (AGC) par-
ticipation presents the best performance with its frequency response
comparable to control area without wind plant with identical settling
time.
Wind AGC barriers
Modern wind turbines are available with power set-point processing
(28) methods but their frequency response capability is generally limited by
65
A. Aziz et al.
the technical constraint in the form of communication delay with wind
farm controller and individual wind turbines. All grid code compatible
wind farms integrated to the grid are required to ensure a fast pitch
control as well as other mechanical controller participation in the fre-
quency response services in acceptable time. Fast Coordination and
processing of power set-points at wind turbine and wind farm level is a
key to improve frequency support from wind power plants. Lack of
power set-point prioritization and providing synchronous connection to
different stakeholders is also not commonly available in wind farm
controllers. A reasonable range of communication latencies for present
technology commercial wind plants is on the order of 200–500 ms [52].
Due to the time delay limitation involved at farm controller level and
unavailability of a specific standard for connection, researchers are
focusing on possibility of implementing AGC control at wind turbine
level itself. Table 4 provides some of the relevant studies related with
wind-AGC investigations where modifications were done at turbine
level or wind farm controller level.
Frequency regulation challenges in wind integrated system
Renewable energy market schemes and its integration with FCAS market
Electricity markets around the world are regulated by the respective
governments, either directly or through nominally independent energy
regulators. Electricity market in general, is a complex system with
different capital intensities, very different fuel costs and risks associated
with generation technology. Different countries have different energy
market structures and their interaction with renewable generations
66
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
Fig. 26. WTG power set-points and corre-
sponding generated WTG power.
Fig. 27. Frequency response comparison for a control area under 0.1
per unit load disturbance and different wind integration scenario.
varies. In liberalized energy markets including USA, Europe, Australia
etc., electricity exchange is separated into two types of market: trading
market (Wholesale/Retail/Capacity/Spot) and the balancing/ancillary
service market. Electricity trading market is classified depending upon
dispatch and bidding cycle and varies from country to country. Fig. 28
represents the classification of electricity market in presence of re-
newable energy. A detail review of these market is out of scope of this
paper. More information about market structures can be obtained from
[141–143]. Wind energy generation presently receives jurisdiction
support through price and quantity regulation in energy spot market
besides voluntary and indirect support [143]. Table 5 presents various
market support schemes for wind energy. In fixed price-oriented
scheme, the wind energy producers are either paid an above-market
price for the energy generation and the required quantity is determined
by energy market. Example Germany, Denmark, Spain and France,
Canada, Ireland and Greece. In legally obligated quotas or quantity
based scheme, the share of the energy supply is guaranteed by gov-
ernment while the market determines the energy price. Examples, UK,
Sweden, Poland, Italy, Belgium, Japan and several states of US. The
capacity of installed wind energy systems is significantly different due
to the diverse economics of the two schemes. There are some jurisdic-
tions like Denmark and Spain, which operate both types of the schemes
and is known as Green marketing. Green marketing or voluntary system
Program is based on customer willingness to pay extra price for re-
newables. In all direct schemes, price for wind energy is directly or
indirectly related with trading market schemes and respective pricing
mechanism. Energy markets participants are also required to make
A. Aziz et al.
Table 4
Wind – AGC: A Review.
Refs. Methodology
Modelling frameworks for wind energy penetration effect on AGC performed by synchronous generators
[123] Analytically traceable AGC model is developed using non-Linear differential
algebraic equations (DAE) describing synchronous generator, wind plant and
electrical network. Third order model of conventional system and first order
model of wind system was used in this study with the assumption of small
forecast error
[124] Flexible Energy Scheduling Tool for Integration of Variable generation (FESTIV) Proposed high-resolution steady-state power system simulation model provides
is presented to mimic actual system operations at a high time resolution using
an amalgamation of security constraint unit commitment, security constraint
economic dispatch and AGC program based on four different modes
[125] A model is developed for power system imbalance control by program
responsible parties PRPs via minimization of their energy program deviations
and the assessment of AGC performance with high integration of wind power
[126] Distributed networked control (DNC) scheme and its stability analysis
framework for automatic generation control is presented for networked
interconnected power systems with participation of wind turbine
Active power frequency control strategies at wind turbine level and/or wind farm controller level
[127] A hierarchical architecture with a supervisory control system for active and
reactive power control at wind farm substation was presented. Supervisory
control system active power output is dependent upon TSO provided power set
point.
[110] [128] Active power controller was incorporated with AGC command tracking from
wind farm controller
[129] DFIG based wind farm hierarchical control strategy to regulate the wind farm
power production to the reference power ordered by the system operators
[130] An aggregated wind power plant model which is integrated into a generic power
system model
[131] Power/load optimization in presence of ambient wind, wake deficits, wake
meandering, the wind turbine system, the wind farm controller, and a network
operator
[132] Modelling of wind turbines and simple wind farm for power set-point tracking
under AGC operation is presented
[133] A control strategy to control active power of wind farms is proposed to
participate in the load frequency control of power systems
Wind plant coordination with conventional generation plants and/or energy storage for AGC
[134] Coordinated control strategy for the AGC between combined heat and power
plants (CHPs) and WPPs to enhance the security and the reliability of a power
system operation in the case of a large wind power penetration
[135] Wind farms incorporating energy storage systems for AGC participation
67
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
Special Findings
Electrical network modelling is included in analytical AGC model which is
usually suppressed in conventional AGC model. The electrical network also has
some smoothing effects on frequency response under wind power variability.
The dynamic behavior of the overall closed-loop system is affected under
electrical network constraints, which is not captured in traditional AGC models.
No wind participation in AGC is assumed for developed model. Even though no
direct results for power system frequency performance metrics under wind
penetration were provided, analytical AGC modelling framework was applied to
the simplified Western Electricity Coordinating Council (WECC) model
flexibility in configuring systems parameters and different operational and
market structures. Model allows for wind energy penetration impacts analysis in
terms of reliability and costs by varying dispatch interval, forecast errors and
AGC modes. The variability and uncertainty impacts on the PJM 5-bus system
under actual wind and load data is investigated using Absolute ACE in Energy
(AACEE), CPS2 violations, σACE, and costs as the metrics under different modes
of AGC. Under perfect forecast, it is observed that the only variability in dispatch
could occur because of different ramping capabilities of power plants.
Persistence wind and load forecast errors have significant effect on imbalances,
CPS2 violation and ACE increment
Wind power variability may lead to higher ACE, especially if insufficient
amounts of reserves are taken into account during the unit commitment and
economic dispatch calculations. The capabilities of the generation mix face
increasing challenges at higher wind power penetrations and additional reserves
are required for keeping area control error at the same level
Time varying network-induced delays in the states of the subsystems models and
latency in the interconnection links due to random packet dropouts is considered
in proposed model. A linear matrix inequality (LMI)-based method is proposed to
design the distributed AGC controller for better system performance in wind
integrated 4-area power system
Improved AGC performance without significant extra costs and better
integration of wind with grid is possible when supervisory system controls active
and reactive power while machine system controls turbine power set points.
Static and dynamic droop curves were implemented and Field tested for NREL’s
550 kW 3-bladed Controls Advanced Research Turbine (CART3) for frequency
regulation participation at all three levels. Fast Wind power AGC response
especially at high wind speed and AGC error is low with multiple turbines in
wind farms.
Implemented in DIgSILENT, the central wind farm control level controls the
power production of the whole farm by sending out reference power signals to
each wind turbine, while the local wind turbine control level ensures that the
reference power signal sent by the central control level is reached
Wind plants can deliver extra control support such as temporary frequency
response (TFR) and power oscillation damping (POD) to enhance the stability of
power systems with large share of wind power
Centralized wind farm model and control principles are used for control of a
wind power to increase energy quality and reduce the fatigue loads
Investigated the effect of individual wind turbines on wind farm performance
under the set-point operation. Pitch angle control and rotor speed control
strategies is investigated for power set point tracking and effect on frequency
response. Simplified wind farm is proposed based on effectiveness of set-point
operation. Wind turbines can track and produce output power for the condition
when reference set-point is less than maximum power point else maximum
available wind power is harvested
A coordinated deloading strategy is applied to wind turbines for active power
command tracking which is provided by system level controller. The proposed
coordination control strategy allocates participation factor to wind turbines and
conventional generation by monitoring online ACE, active power output of wind
farm and maximum generation, thereby calculating ramping capability of
generating plants
Proposed control strategy has shown that the WPPs can actively help the AGC,
and reduce the real-time power imbalance in the power system, by down
regulating their production when CHPs are unable to provide the required
response
Energy storage devices are used to neutralize the uncertainty of wind power
plant output and maximize the dispatch ability. With reduced uncertainty, wind
farms are shown to have better AGC performance
(continued on next page)
A. Aziz et al.
Table 4 (continued)
Refs. Methodology
Modelling frameworks for wind energy penetration effect on AGC performed by synchronous generators
[136,137] DFIG based wind farm integrated with thermal energy system was used in study
[138] Cooperation scheme for wind power and battery storage providing frequency
regulation
Intelligent AGC
[139] (PSO-Fuzzy PI) Fuzzy logic load frequency controller and Particle swarm
optimization technique were applied
[140] Intelligent agent-based control scheme, using Bayesian networks (BNs) based
AGC framework is proposed in a multiarea power system.
supplementary control contributions to TSO for the stable and reliable
power system. FCAS represents a guarantee that power will be con-
tinued to be delivered in the case of system contingency.
Implemented under different names in different countries, fre-
quency regulation services are generally divided into two categories:
Compulsory provision of ancillary services and market oriented provi-
sion of ancillary services [144]. For example, in North America, new
generators are required to provide compulsory primary frequency re-
sponse. In UK, all licensed generators are obliged under the Grid Code
to provide mandatory frequency response. In market-oriented ancillary
services, FCAS providers bid their services into the FCAS markets in a
Fig. 28. Classification of electricity markets in presence of renewables. Most of the liberalized markets globally recognize this classification though it varies in terms of time horizon for
dispatch and trading. Block in green show the market where effect of wind energy is highly crucial.
68
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
Special Findings
Operational impacts of the wind power fast fluctuations are largely absorbed by
the thermal unit’s large mechanical and thermal time constants as well as control
dead-bands and the system dynamics varies with the size of the AGC time delay.
It is estimated in this study 5% of rated plant power can be accepted as
perturbation power by thermal plants, without exceeding the 1%frequency
deviation margin
A real-time cooperation scheme is proposed to exploit the complementary
characteristics of battery storage and wind power and an optimal bidding
strategy is developed for participation in joint energy and regulation markets
Intelligent computational techniques were applied to improve AGC controller
performance in New England test system with wind penetration but no specific
wind farm frequency support was accounted
The BN provides a robust probabilistic method of reasoning under uncertainty
and undefined system dynamics. Multiagent structure application in the control
framework realizes parallel computation and a high degree of scalability. The
proposed system is tested for optimal performance of AGC system through
simulation and laboratory experiment
similar fashion like generating units bidding into the energy market and
receive financial incentives. To meet control area performance criteria,
AGC enabled FCAS participants must track and keep a record of in-
stantaneous fluctuations in demand load for the continuous duration so
as to keep ACE within limits. FCAS requirements are calculated by
contingency size, total inertia of power system and load demand.
Payments for ancillary services include payments for availability and
delivery of the services. Like electricity trading market, FCAS price also
depends very much on the institutional setting in which the regulation
service is delivered.
Wind turbines have been considered as passive components in the
A. Aziz et al. Sustainable Energy Technologies and Assessments 26 (2018) 47–76

Table 5
Market support schemes for wind energy, derived from [143,144].

Direct Indirect

Price-driven Quantity-driven (Legal Obligatory quotas)

Regulatory Investment focused
Generation based
• Investment subsidies
• Tax credits
• (Fixed) Feed-in tariffs
• Fixed Premium system
• Tendering system • Environmental taxes (CO2 taxes)
• Tendering system
• Tradable Green Certificate system
• Renewable Portfolio Standards

Voluntary Investment focused • Shareholder Programs Voluntary agreements

• Contribution Programs
Generation based • Green tariffs

past. They could cut off from grid in case of grid voltage or frequency
crossing permissible limit. With large penetration of wind energy, such
response may aggravate grid instability. Wind farms are now con-
sidered as power units having similar characteristics as that of con-
ventional generation units. So, FCAS availability holds a very sig-
nificant value for high wind energy penetrated power system with low
inertia. For the TSO, assignment of control duties gets constricted with
the presence of spatially distributed wind farms. TSO aims to minimize
the payments to participants for providing AGC ancillary service while
capital-intensive wind energy participants striving for maximum rev-
enue, voluntarily opt out of FCAS provision. Wind farms which pro-
duces electricity at low prices and have active power frequency support
capabilities have option of gaining financial revenues by participating
in the FCAS market for provision of the regulation services. The ancil-
lary services market for renewables is slowly growing in many countries
worldwide. Wind farms have been participating in FCAS in Quebec,
Ontario and some European Countries. As a recent example, the Italian
energy and gas regulator Autorità per l’energia elettrica, il gas e il sis-
tema idrico (AEEGSI) has issued directions authorizing pilot renewable
energy power generators and storage units to participate to the ancil-
lary services market known as Mercato Servizi di Dispacciamento
(MSD) operated by the country’s grid operator Terna [145]. Australia
electricity market operator (AEMO) is supporting first trial for FCAS
from 100 MW Hornsdale 2 wind farm [146]. Newzeland’s 64.4 MW Te
Uku Wind Power Plant has successfully completed commissioning test
for providing frequency regulation services [147]. Utilizing governor
control on full converter based wind turbines, the wind farm can
provide active power response by ramping from zero to 85% rated
output in 4 s. Table 6 lists various technical and operational function-
alities matched with respective frequency control services, their im-
plementation level and their availability status in current industrial
world [52]. It can be noticed that current wind turbines provide good
power frequency support technical capabilities but their provision in
ancillary market is limited due to regulatory policies and some opera-
tional challenges. New frequency responsive wind plants still face a
challenge for adaptation from present grid structure and seek reg-
ulatory certainty and stability from government for economic gains.
Challenges
Technically, wind generation can secure the quality, stability, and
reliability of the grid by delivering ancillary services. Proper manage-
ment of FCAS can be a key factor for better penetration of the wind in a
deregulated complex power system. But some challenges remain in
optimizing the control of the wind farms to maximize the economic
opportunities while ensuring that the power system operates securely.
In consideration to increasing wind energy penetration, following
challenges needs attention for proper management of FCAS in wind
penetrated large grid like Australian National Electricity Market.
Wind energy forecasting improvement. The balancing market and day
ahead spot market, shown in green color in Fig. 28 is very important for
better integration of wind energy in electrical grid. Main challenge for
wind integration lies in accurate and reliable forecasting. With
increasing wind energy penetration, uncertainty in wind power

Table 6
Wind energy system frequency support functionalities and their participation in frequency control ancillary services.

Type of functionality Functionality Frequency Regulation Services

Inertial/Fast Primary Secondary Tertiary Implementation Level

Frequency

Technical Active Power Control ✓ ✓ ✓ WT/WF

Frequency Sensitive/Droop
Control Mode
Frequency Sensing
Power Set Point Processing
Set Point Priority Management
Temporary Power Increment
WT/WF
✓ ✓ ✓ WT/WF
✓ ✓ ✓ ✓ WT/WF
• • • • WF
• WT

Operational Active Power Production WT

Calculation
Power Production Forecasting ✓ ✓ WF
Communication and Control ✓ ✓ ✓ ✓ WT/WF
Wind Power Plant Management WF
System
✓ Functionality required and today generally available. • Functionality rarely available/
Functionality required but not always available or optionally available. not implemented but
implementable/programmable

69
A. Aziz et al.
Table 7
Wind energy forecasting classification on the basis of time horizon, application and methods.
Wind forecasting horizon Time scale Application
Very Short Term Seconds to minutes Wind turbine control, frequency response
Short term Hours to days Economic dispatch Reserve requirement Day-ahead
electricity market
Medium term Days to week Unit commitment Maintenance scheduling
Long term Week to years Wind power planning Power system planning
forecasting has direct effect on wind turbine operation, reserve
requirements and grid stability. Wind forecasting has seen continuous
improvements since last twenty decades of operational use [148].
Table 7 provides wind power various forecasting techniques, their
horizon in terms of time scale and applications [149,150]. Persistence
forecasting is applied for very short term time horizon, generally up to
5 min and short term up to 48 h. Persistence forecasting is based on the
assumption that conditions at the time of forecasting will remain
constant. State-of-the-art forecasting techniques utilize a modelling
chain to predict the wind energy for look-ahead periods. Fig. 29
represents an example of the entities involved and respective data
flow in short term prediction of wind power output. Output of
Numerical Weather Prediction (NWP) model is coupled with wind
power prediction conversion model which also utilize the on-site
SCADA data and other data related with orography and wind farm
layout. Different countries apply different forecasting duration and
models (NWP and wind power prediction models) for wind energy
estimation. A detailed review of current state of the art in wind power
forecasting can be obtained from [149,151].
Diverse practical applications demand assorted forecasting under
different timescales. Wind farm operators are more interested in very
short term forecasting (< 5 mins.) local to wind farm and very specific
to the operating turbines. As discussed previously in section 3, wind
turbine controllers apply wind power calculation in their algorithm for
providing frequency response [60] through wind-power curve. Persis-
tent wind forecasting using real time SCADA data is applied for these
very short term range. Non-linear S shaped wind power relationship
always tends to amplify the uncertainty in the wind forecasting.
Talking about the balancing market, frequency regulation costs are
highly dependent upon generation technology, predictability of de-
mand, predictability of wind generation and the specific regulation
framework of the respective jurisdiction. Electrical grid with pre-
dictable balancing requirements can easily schedule least expensive
regulating units for frequency regulation ancillary service. Grid op-
erators are interested in short term wind forecasting for balancing as-
sessments while electricity traders and system planners require medium
70
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
Forecasting methods
Persistence forecasts (for below 5 mins) with real time SCADA
Physical, Statistical model + NWP model (Now-casting models for
1-3h)
NWP models + real-time Four-Dimensional Data Assimilation
(RTFDDA) for 3-6h)
NWP model + Three-Dimensional Variation Methods (3DVAR) for
12 h-2 weeks
Climatology (> 14 days)
to long term forecasting for generator scheduling and system long term
planning respectively. The dispatch levels for generators are calculated
by normally centralized forecasting system which provides wind fore-
cast for time frame from 12 to 36 h (Europe), 1 to 168 h (PJM) and
5 min-2 years (Australia). The accuracy of wind forecasts depends on
the location, time of year, SCADA quality from individual wind tur-
bines, wind farm layout, and weather conditions. Wind power varia-
tions can arise over numerous time horizon, ranging from sub-hourly
fluxes to diurnal, seasonal, and ever interannual [148,150]. Accurate
prediction of wind power output a few hours or few days in advance is
comparatively more difficult than annual generation prediction. Wind
power forecasts also have the highest uncertainties mainly due to the
temporal and spatial variability along with predictability of the wind
field. It is a general acceptance that power forecast errors are highly
affected by NWP model errors [152,153]. Most NWP models face a
difficulty in incorporating any physical terrain which is less than the
size of its smallest grid cell [150]. Hence NWP model accuracy is af-
fected with presence of structures like mountains. There has been an
increase in temporal resolution with most of the metrological systems
giving hourly data but many others are still handle three-hour data
[148]. There are further concerns regarding higher temporal resolution
due to storage and computing requirements. Day-ahead forecasting
errors for wind generation have declined in the previous years but they
are still much higher than errors when the forecast is made few hours
before real time [174]. Fig. 30 gives the relation between mean abso-
lute error/average production and prediction rate for wind generation
in Spain. Mean average error for one hour ahead forecast for a wind
turbine varies from 4 to 5% of rated capacity which rises to 8% of rated
capacity for day ahead error in year 2015. UK National Grid has an
error of approximately 5%, measured as mean absolute forecast error
against capacity [155].
In the case of wind farms inability to match the forecasted power
submitted to the dispatch market, there is extra burden on other gen-
eration technologies to raise or lower their power output for regulating
system, thus increasing total regulating cost. Similarly, if there is excess
power in market and wind farms are generating more then offered
Fig. 29. Short term prediction of wind power output.
A. Aziz et al.
forecast, other generators must adjust their output for maintaining
system balance. In both the case, the wind farms will be penalized and
get a lower price for his electricity production than the spot market
price. The regulation price and penalty varies from system to system.
There is a linear variation of regulating cost with wind penetration. The
spatial distribution and variability of wind resources affect power
system on time scale being followed by AGC and raise the cost of
electricity. Nevertheless, in comparison to wind forecast errors, varia-
bility has lesser impacts on electricity cost and system operation
[150,152]. System operating cost has rose by 1–4 €/MW h of wind
generation at 20% wind energy penetrations [143]. In terms of fore-
casting horizon, short term forecast errors have considerable impact on
ACE and balancing costs while ACE and system reliability is lesser ef-
fected by day-ahead forecasts. Sufficient frequency reserves can miti-
gate impacts of day ahead forecasting though at an increased cost. With
increased wind penetration, most the cost is anticipated to be borne
solely by wind farms leading to an escalation in regulation costs for
wind generators. There are many instances when erroneous forecasting
results from Australian Wind Forecasting System (AWEFS) have re-
sulted in scheduling error declaration by the transmission operator and
penalty imposed on wind farms [156].
Considering the significant impact of wind forecasting on system
balancing and stability as well as wind farm frequency support, it is of
utmost importance to advance the present forecasting techniques and
their operational issues over wide range and resolutions. Replacing
deterministic forecasting with probabilistic forecasting along with wind
power pooling can reduce the wind power variability and diminish the
uncertainty. A collaboration of various NWP models with real time field
data can provide a common benchmark for error and uncertainty
quantifications. In recognition of need for forecast advancement,
International Energy Agency (IEA) has set up a task force on wind en-
ergy forecasting [148]. Following are some of the important recognized
forecasting issues which need further research and investigation:
1. Common availability of frequent and rapid update cycle model.
2. Meteorological forecasting below 1 km horizontal resolution
3. Data assimilation (specially from offshore wind farms)
4. Improvement in NWP model physics (specially in terms of stability
and its daily pattern calculations and nowcast model for problematic
conditions like thunderstorms, small low pressure systems)
5. Effect of icing.
6. Modelling of wind farms interaction under the wake effect and wind
direction forecasting
71
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
Fig. 30. Example of wind forecasting error variation
with respect to prediction time horizon [154].
7. Wind ramp forecasting
Dispatchable wind farms. Wind farms are currently not dispatchables in
most of the countries but could provide flexibility to the system
operations. With higher wind energy penetration, deloaded wind
turbines operation or wind integration with energy storage for fast
regulation services will become more practical for making wind farm as
dispatchables generation technology. Dispatchability improvement
authorize time shifting of wind power dispatched to the electricity
grid, enabling wind generators to supply baseload power, exploit
energy arbitrage and provide ancillary services [138]. Various studies
have support the concept of wind farm dispatchability with storage
support as technically and financially feasible option [158–162]. The
policy of dispatching wind farms is being translated into many
countries. Countries like UK and Spain, requires participation of wind
power plants in electricity markets and they are self-responsible for
power forecasting and balancing services [163]. Dispatching wind
farms will require inter-collaboration of several small wind farms,
simultaneous control of different related devices and the application of
business and market dynamics to the control of the wind farms.
Coordination of the large number of wind turbines under changing
wind conditions, their associated controls under external power set-
points and power production is a complex task with severe
computational limitations. Latest wind farm controllers have
capability to control multiple wind turbines, but availability of TSO
provided active power set-points and prioritization of different power
set-points is still lacking in practice [52]. Dispatchables wind farm
controlled through intelligent wind farm controller are not entirely
available in practice though they are being discussed in literature
[164,165]. Absence of fast and direct communication link between TSO
and wind power plant is a factor affecting wind-based ancillary service.
Very few countries like Spanish grid code requires a dedicated
connection between wind farm controller and TSO [166].
Governor dead-band setting regulation. Dead-band and speed droop are
two significant parameters in conventional turbine governor action
during system event [39]. Long-term system frequency regulation
reliability is highly dependent upon implemented dead-band.
Governor dead band determines the minimum frequency variation
needed before the governor action is activated. In response to frequency
deviations, generating units depart from their dispatch power targets
according to the set droop characteristics. Droop response of generating
machine can be avoided when frequency lies within the upper and
A. Aziz et al.
Fig. 31. Frequency relaxation on the basis of interconnection size.
lower dead-band range. Frequency control is generally tighter in case of
large interconnection as probability of disturbances reduces. Network
interconnection can be divided into three zones from frequency control
perspective as shown in Fig. 31 [167]. In zone 1 where interconnection
is small, active participation of all participating generators and load is
desired to avoid undesired large frequency deviations. A tight
frequency control through low dead-band ensures interconnection
flow to the contract. So, there are possibilities for relaxing frequency
control in zone 2 from economic perspectives. With growth in
interconnection, frequency standard deviation is improved. However,
frequency control problem becomes uneconomically suboptimal if same
tight frequency control standard is followed. This relaxation of
frequency relaxation is a progression of steps with interconnection
growth. With very large interconnection, frequency regulation is more
than just flat frequency control and becomes interconnection
management with tie-line frequency control. Such type of control can
be present in zone 3.
Large wind penetration in a network system have significant effect
on zone 1 and zone 2 and respective interconnection flow and active
support from synchronous generating units is desired. Grid congestion
and intra interconnection management is a major problem with in-
creasing wind penetration. It is advocated that renewable energy
72
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
integration has direct relation to 80% of the identified bottlenecks in
the European grid [168]. Unbalancing at deregulated interconnections
under variable wind power and frequency relaxation thus, increase
balancing costs. A small change in governor dead-band may have sig-
nificant changes in timing and frequency responsive primary control
from conventional generating units. Large governor dead-band settings
disable primary control and successive smaller contingency events
compromise the grid security and stability and may lead to outage too.
Recent South Australia outage on 28 Sept 2016 highlighted the effect of
the changed or relaxed governor settings from 0.1 Hz to 0.3 Hz. With
governor control, relaxed; conventional units get slow in responding to
frequency disturbance. Even though frequency relaxation through
dead-band increment has benefits in the form of lower reserves under
AGC control and saving through avoiding expensive generator dispatch
[77,169]; disabling governor control in synchronous generation to gain
incentives may also bring significant undesirable impact on grid FCAS
capability and penalty on wind farms who opt out from providing
regulation services. Without tighter dead-band, response of synchro-
nous generators soon after an event is not reliable which will rather be
dependent upon market enablement. Wind integrated interconnected
network calls for a review of frequency operating bands under wide
dead-band for improving FCAS and better wind integration to the grid.
Clear technical grid code regulations and wind turbine wear and tear
implications. Wind turbines are subject to dynamic forces arising due to
dynamic variations in wind speed and direction. Various loadings on
wind turbine generators can be classified into deterministic and
stochastic loadings [170]. A classification of different loadings on
WTG is represented in Fig. 32. Wind turbine structure undergo
fatigue damage which originates from various deterministic or
stochastic loads variation. There may be more than 500 million load
cycles occurring over a wide range of load ratios (Minimum load in one
load cycle divided by the maximum load in one load cycle) for a typical
normal wind turbine [171]. Onshore WTG unit has a life span of
20 years which could typically include 10 million revolutions. The life-
time fatigue Damage Equivalent Loads (DEL) is used to calculate fatigue
damage and it accounts for the amount of time the turbine is expected
to operate at each wind speed condition. Critical locations and of
correlation of load with fatigue strength at those locations account for
failure probabilities. The bearings in gearbox undergoes irregular
repeat start, acceleration, deceleration and stop operations. Gearbox
which converts rotor speed of 5–22 rpm to 1000–1600 rpm normally
fails within 5 years requiring replacement [172]. Fig. 33 presents
general failure rate of different components in a wind turbine [173].
Fig. 32. Classification of WTG loadings.
A. Aziz et al.
Fig. 33. Major failures in a WTG system.
As can be noticed from Fig. 32; gearbox, generator and rotor blades are
three important components in terms of increased failure. All three
makeup an important configuration for achieving frequency response
from wind turbines. As given in Table 3, wind turbine can ramp up the
power in the range of 0.2 p.u./sec but the corresponding effect on
turbine components and total life time is still not clear when ramping is
provided on sustained basis for frequency regulation. Investigative
studies are required to assess the effect on these components when WTG
participates in successive frequency response services. Also, wind
generating plants providing frequency regulation undergo wear and
tear for which they must be compensated through Operation and
Maintenance costs (O & M).
Though there is no significant technological barrier in provision of
frequency regulation from wind turbine, precise requirements from
TSOs are a necessity now. Wind turbine manufacturers have expressed
lack of clear specifications and regulatory standards from TSOs for
activating frequency control services and analyzing their impact on
power system [52]. For example, the wind inertial control raise the
system inertia by substantially releasing the kinetic energy from the
rotating mass, but the load disturbances may also be masked by the
extra wind inertial power for a limited time. This masking effect could
lead to delay in response from the synchronous generators to the fre-
quency events. Fast frequency response from wind turbines which de-
livers temporary instantaneous power also brings a power supply sag
and can have significant impacts on system stability. Varied decen-
tralized and distributed control methodologies to implement inertial
and AGC response at turbine level and real scale power plant level with
high ROCOF ride-through capability is still very nascent and needs
more attention as research in this field appears to be sparse
Incentivize frequency response from wind farms. Although most of the
contemporary WTGs exhibit the capability of quickly reducing power
during high grid frequency, the challenge is in providing a rapid
increase in active power. Due to the entry barriers to regulation
market, wind farm owners without any financial incentives have
restrictions about this approach which can have significant economic
implications. Besides technology; system infrastructure and market
policies also play a vital role in satisfactory FCAS participation of
wind power plant. Frequency regulation costs are ambiguous, and
significantly dependent upon the regulatory and institutional
frameworks and may vary considerably between countries. Wind
farms revenue in quantity based energy markets gets reduced during
power curtailment. Currently, wind farms are balancing responsible
both financially and legally alike conventional generations in many of
the EU member countries. But except Germany, Spain and Denmark,
wind farm participation in frequency control is not supported by any
special financial incentives. Spain’s experience shows that making wind
generators subject to the same balancing and scheduling obligations as
conventional power plants does not jeopardize the development of this
technology [174]. Denmark has fixed an additional compensation for
73
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
balancing costs (€3/MW h) for 20 years for new onshore wind farms. In
UK, conventional generators are compensated by a margin of 25%
accounting for fuel costs during low frequency response while there is
no such mechanism for wind plants [175]. Wind plants or their trade
suppliers must incur revenue loss associated with Levy Exemption
Certificates (LECs) and Renewable Obligation Certificate (ROCs). In
fact, wind generators are required to pay for reduction in output under
the assumption of fuel saving cost. Some of the countries like Australia
takes a technology neutral approach to the provision of FCAS; any
technology that is able to demonstrate the ability to provide the
particular FCAS service to the specified standards is allowed to do so.
Australian energy market operator (AEMO) has the requirement for
FCAS to ensure the frequency to be contained within 49 to 51 Hz in the
event of step changes in supply that result from credible contingencies
known as “separation event.” An impact assessment of large-scale
renewable energy target done for Australian Energy Market
Commission (AEMC) reported an increase in regulation requirement
from ± 120 MW to ± 800 MW and increase in frequency regulation
cost from 10 million per annum to 200 million per annum in 2020 due
to projected increase in intermittent wind energy. Causer-pays
methodology is being implemented in Australia for settling regulation
costs, with generators or loads paying for any deviation from their
expected dispatch. AEMO determines Causer Pays Factor (CPF) to
measure the variation of dispatched generator output from dispatch
level. The difference between the unit’s generation and the linear
interpolation of the current and previous dispatch interval values are
used as the measure of expected dispatch [176]. Scaled deviations are
recorded every 4 s, averaged for every 5 min’ dispatch interval and
finally averaged for 28 days’ period. A negative CPF means that
deviating generator must bear the increasing share of FCAS
regulation cost. Frequency control in most of the countries like
Australia has traditionally only been provided from synchronous
thermal generation. There has been little incentive for wind farms to
contribute to frequency regulation so wind farms prefer not participate
in FCAS. Participation of wind farm in ancillary service market besides
spot market can provide a boost to their revenue. Consistent integration
of wind calls for improved policies development and encouragement
from governing authority for efficient diverse frequency regulation
ancillary services market, better energy economics and compliance to
control performance standards.
Conclusion
Design and operation of power system in presence of wind energy
are one of the major issues in wind power integration. Variable speed
wind turbine generators (VSWTG) having incommensurable dynamic
characteristics than conventional synchronous generators no longer
receive negative load treatment. In view of increasing renewable pe-
netration and compulsory grid codes for frequency support, there is
increasing necessity for wind farms to perform as dispatchables gen-
eration technology. Modern power electronics based VSWTGs are cap-
able of supporting frequency regulation services at all three levels. This
paper presented frequency response capabilities of wind energy system
and perspective challenges for improving frequency regulation in wind
penetrated system. Advancement of wind energy forecasting tech-
nology over very shorter horizon is one of the most important challenge
in promoting wind energy integration. Effective research and im-
plementation of these forecasting technologies will help wind energy
market participants to make better contributions to power generation
and frequency control ancillary services. Current wind plant receives
direct and indirect support from respective jurisdictions in electricity
trading market but the economic benefits are still absent in most of the
countries for their contribution to balance the system generation and
demand. Market operators need to realize the capabilities and current
state of the art of wind generation technologies. Development of ef-
fective policies incentivizing wind farm contributions in frequency
A. Aziz et al.
responsive services will pave the way for an overall more reliable and
stable wind integrated power system. Network regulators should work
toward the establishment of appropriate standards related to frequency
regulation with a strong focus on service provider speed and accuracy.
A detailed review of governor settings for all generation technology
must be undertaken to recognize and address the current settings of
power system. Improved wind turbine controller technologies, external
support of conventional resources, precise grid code technical specifi-
cations and supporting government policies are the key to achieve long-
term economic sustainability of wind energy.
References
[1] IRENA, Renewable Energy Target Setting; 2015. Available at www.irena.org.
[2] World Energy Council. World Energy Resources; 2013. Available at www.
worldenergy.org.
[3] ARENA. Australian Energy Resource Assessment, Chapter 9, Wind Energy, pp 239-
260; 2010. Available at www.arena.gov.au/about-renewable-energy/wind-
energy/.
[4] Australian Energy Council. Renewable Energy in Australia; 2016. Available at
www.energycouncil.com.au/.
[5] Wind power monthly. Ten of the biggest turbines; 2017. Available at: http://www.
windpowermonthly.com/10-biggest-turbines.
[6] Dong Energy. Successful installation of world’s largest offshore wind turbine; 2017.
Available at http://www.dongenergy.com.
[7] REN21. Renewables 2016: Global Status Report; 2016. Available at www.ren21.
net.
[8] National Energy Board. Canada Renewable Power Landscape, Energy Market
Analysis 2016; 2016. Available at www.publications.gc.ca.
[9] Ma Tianjei. China's Ambitious New Clean Energy Targets, The Diplomat; 2017.
Available at www.thediplomat.com.
[10] Laurie Guevara-Stone. A small country goes big with renewables: Denmark’s goal
to be fossil-free; 2016. Available at www.reneweconomy.com.au.
[11] France in the United Kingdom. French renewable energy production up over 23%
in 2015, Energy policy/renewable energy – Communiqué issued following the
Council of Ministers’ meeting; 2016. Available at www.uk.ambafrance.org.
[12] Tsvetomira Tsanova. India reached 22% of 175-GW renewable energy goal,
Renewables Now; 2015. Available at www.renewablesnow.com/news.
[13] Oxford Business Group. South Africa’s push for Renewables-Economic News; 2016.
Available at www.oxfordbusinessgroup.com.
[14] Commons Select Committee, Energy and Climate Change Committee. 2020 re-
newable heat and transport targets report; 2016. Available at www.parliament.uk/
cec.
[15] Office of Energy Efficiency & Renewable Energy. Federal Energy Management
Program, Achieving 30% Renewable Electricity Use by 2025; 2017. Available at
www.energy.gov.eere.
[16] Eurostat. Renewable energy in the EU, Share of renewables in energy consumption
in the EU still on the rise to almost 17% in 2015; 2017. Available at http://ec.
europa.eu/eurostat.
[17] Wan Y-H, Parsons BK. Factors relevant to utility integration of intermittent re-
newable technologies, NREL, Aug. 1993.
[18] Electric Power Research Institute, “Requirements Assessment of Wind Power
Plants in Electric Utility Systems,” vol. 2, EPRI ER-978. Final Report. January
1979.
[19] Luo C, Far HG, Banakar H, Keung PK, Ooi BT. Estimation of wind penetration as
limited by frequency deviation. IEEE Trans Energy Convers 2007;22(3):783–91.
[20] Simburger EJ, Cretcher CK. Load following impacts of a large wind farm on an
interconnected electric utility system. IEEE Trans Power Apparatus Syst 1983;PAS-
102(3):687–92.
[21] Curtice D, Patton J. Operations of Small Wind Turbines on a Distribution System,
Systems Control, Inc., August; 1980.
[22] Wang C, McCalley JD. Impact of wind power on control performance standards. Int
J Electr Power Energy Syst 2013;47:225–34.
[23] Grubb MJ. The economic value of wind energy at high power system penetrations:
an analysis of models. Sensitivities and Assumptions Wind Eng 1988;12(1):1–26.
[24] Muljadi E, Gevorgian V, Singh M, Santoso S. Understanding inertial and frequency
response of wind power plants. In: Proc. IEEE symp. power electronics and electric
machines in wind applications; 2012. p. 1–8.
[25] AEMO. Wind Turbine Plant Capability Report; 2013. Available at aemo.com.au.
[26] Riesz J, Milligan M. Designing electricity markets for high penetration of variable
renewables, Wiley Interdisciplinary Reviews, Energy and Environment; 2015.
Available at: arena.gov.au/assets/2017/06/Riesz_et_al-2015_FD_8-2.pdf.
[27] Riesz J. Frequency Control: The Future Power System Security Program; 2016.
Available at aemo.com.au.
[28] Aghahosseini A, Bogdanov D, Breyer C. A techno-economic study of an entirely
renewable energy-based power supply for North America for 2030 conditions.
Energies 2017;7(10):1171.
[29] Blakers Andrew, Bin Lu, Stocks Matthew. 100% renewable electricity in Australia.
Energy 2017;133.
[30] Bin Lu, Blakers Andrew, Stocks Matthew. 90–100% renewable electricity for the
South West Interconnected System of Western Australia. Energy 2017;122.
[31] Bogdanov Dmitrii, Breyer Christian. North-East Asian Super Grid for 100%
74
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
renewable energy supply: optimal mix of energy technologies for electricity, gas
and heat supply options. Energy Convers Manage 2016;112:176–90.
[32] Jacobson MZ, Delucchi MA, Cameron MA, Frew BA. Low-cost solution to the grid
reliability problem with 100% penetration of intermittent wind, water, and solar
for all purposes. Proc Natl Acad Sci USA 2015;112(49):15060–5.
[33] Cee.stanford.edu. Mark Jacobson: The future of energy impacts the future of our
cities | Civil and Environmental Engineering; 2017. [online] Available at: https://
cee.stanford.edu/news/mark-jacobson-future-energy-impacts-future-our-cities.
[34] MacDonald, Cappers P, Callaway D, Kiliccote S. Demand response providing an-
cillary services. In: Proc. Grid-Interop 2012. Grid Wise Architecture Council,
Irving, TX; 2012.
[35] Enerdata. Global energy statistical yearbook 2017 Available at: ∗∗∗yearbook.en-
erdata.net.
[36] Borsche T, Ulbig A, Andersson G. A new frequency control reserve framework
based on energy-constrained units. In: Power systems computation conference
(PSCC) 2014; 2014. p. 1–7, 18–22.
[37] Ferc.gov. FERC: Industries – RTO/ISO; 2017. [online] Available at: https://www.
ferc.gov/industries/electric/indus-act/rto.asp [Accessed 9 Sep. 2017].
[38] Load Frequency Control (2004), UCTE_Operation_Handbook_Appendix1, v1.9
draft.
[39] ENTSO-E. Operational reserve ad hoc team report, final version; 2012, Available at
www.entsoe.eu.
[40] Verband der Netzbetreiber - VDN - e.V. beim VDEW. Transmission code 2007.
Network and system rules of the German transmission system operators. Available
at www.vde.com.
[41] National Grid. Mandatory Frequency Response | National Grid; 2016. [online]
Available at: http://www2.nationalgrid.com.
[42] National Grid. The Grid Code, Issue 4, Revision 10, National Grid Electricity
Transmission plc, UK, January 2012.
[43] Christer Bäck. Nordic implementation of Frequency Restoration Reserve
Automatic (FRR-A), Svenska kraftnät (SVK); 2013. Available at www.svk.se.
[44] Ignacio de la Fluente. Ancillary services in Spain: Dealing with high penetration of
RES. Red Eléctrica de España. Available at www.reshaping-res-policy.eu/
downloads/topical%20events/de-la-Fuente_Ancillary-Services-in-Spain1.pdf.
[45] EirGrid. Eirgrid grid code version 4.0, Available at www.eirgrid.com.
[46] Hewicker C, Werner O, Zeigler H. Qualitative analysis of cross-border exchange of
balancing energy and operational reserves between Netherlands and Belgium;
2013. Available at: http://www.tennet.eu.
[47] TenneT, The Netherlands, Samenvatting uit “Summary of the Current Operating
Principles of the UCPTE” met betrekking tot de primaire en secundaire regeling,
Dec. 2003.
[48] PJM. USA, Manual 12: Balancing Operations, Revision: 36; 2017. [Online].
Available at http://pjm.com/-/media/documents/manuals/m12-redline.ashx?
la=en.
[49] PJM. USA, Manual 10: Pre-Scheduling Operations, Revision: 354, Jan. 26; 2017.
Available at http://pjm.com/-/media/documents/manuals/m10-redline.ashx?
la=en.
[50] AEMO. Frequency Control Ancillary Services; 2012. Available at aemo.com.au.
[51] Dan Woodfin. Methodology for Determining Minimum Ancillary Service
Requirements; 2015. Available at www.ercot.com.
[52] Faiella M, Hennig T, Cutululis NA, Van Hulle F. Capabilities and costs for ancillary
services provision by wind power plants. Deliverable D 3.1. Fraunhofer IWES;
2013.
[53] de Almeida R, Lopes JP. Participation of doubly fed induction wind generators in
system frequency regulation. IEEE Trans Power Syst 2007;22:944–50.
[54] Juankorena X, Esandi I, Lopez J, Marroyo L. Method to enable variable speed wind
turbine primary regulation. In: International Conference on Power Engineering,
Energy and Electrical Drives; 2009. p. 495–500.
[55] Liserre M, Ćardenas R, Molinas M, Rodriguez J. Overview of Multi-MW Wind
Turbines and Wind Parks. IEEE Trans Indus Electron 2011;58(4):1081–95.
[56] Morren J, de Haan S, Kling W, Ferreira J. Wind turbines emulating inertia and
supporting primary frequency control. IEEE Trans Power Syst 2006;21(1):433–4.
[57] Gowaid, El-Zawawi A, El-Gammal M. Improved inertia and frequency support
from grid-connected DFIG wind farms. In: Power Systems Conference and
Exposition (PSCE), 2011 IEEE/PES; 2011. p. 1–9.
[58] Ma H, Chowdhury B. Working towards frequency regulation with wind plants:
combined control approaches. Renewable Power Gen IET, 2010; 4(4): 308–316.
[59] Anderson P, Fouad A. Power system control and stability. Piscataway, US: Wiley –
IEEE Press; 2002.
[60] Clark K, Miller NW, Sanchez-Gasca JJ. Modelling of GE wind turbine-generators
for grid studies, Version 4.5. General Electric International Inc; 2010.
[61] Nguyen N, Mitra J. An analysis of the effects and dependency of wind power pe-
netration on system frequency regulation. IEEE Trans Sustain Energy
2016;7(1):354–63.
[62] Ullah NR, Thiringer T, Karlsson D. Temporary primary frequency control support
by variable speed wind turbines— potential and applications. IEEE Trans Power
Syst 2008;23(2):601–12.
[63] Chan ML, Dunlop RD, Schweppe F. Dynamic equivalents for average system fre-
quency behavior following major disturbances. IEEE Trans Power Appl Syst. 1972;
PAS-91(4): 1637–1642.
[64] National Grid. Requirements for System Inertia, National Grid Code Frequency
Response Working Group, UK; 2010.
[65] Morren J, Pierik J, Dehaan S. Inertial response of variable speed wind turbines.
Electric Power Syst Res 2006;76(11):980–7.
[66] Mullane A, O'Malley M. The inertial response of induction-machine-based wind
turbines. IEEE Trans. Power Syst 2005;20(3):1496–503.
A. Aziz et al.
[67] Anaya-Lara O, Hughes F, Jenkins N, Strbac G. Contribution of DFIG based wind
farms to power system short-term frequency regulation. IEE Proc Gen Transm
Distrib 2006;153(2):164–70.
[68] Tarnowski GC, Kjar PC, Sorensen PE, Ostergaard J. Variable speed wind turbines
capability for temporary over-production. In: IEEE Power & Energy Society
General Meeting, Calgary, AB; 2009. p. 1–7.
[69] Kundur P, Balu N, Lauby M. Power system stability and control. New York:
McGraw-Hill; 1994.
[70] AEMO. Wind Turbine Capability Report; 2013.
[71] Engelken S, Mendonca A, Fischer M. Inertial response with improved variable
recovery behaviour provided by type 4 WTs. in IET Renewable Power Gen 2017;
11(3): 195–201, 2 22.
[72] Tielens P, Van Hertem D. Receding horizon control of wind power to provide
frequency regulation. IEEE Trans Power Syst 2017;32(4):2663–72.
[73] Wayne Miller, Arnie Heller. Breath of Air”, in Meteorological Technology
International Journal, pp. 22, April 2015; 2015. U.K. Available at: http://www.
mbw.ch/wp-content/uploads/2015/05.
[74] Str.llnl.gov. Predicting Wind Power with Greater Accuracy; 2017. [online]
Available at: https://str.llnl.gov/april-2014/miller.
[75] Mauricio JM, Marano A, Gomez-Exposito A, Martinez Ramos JL. Frequency reg-
ulation contribution through variable-speed wind energy conversion systems. IEEE
Trans Power Syst 2009;24(1):173–80.
[76] Van de Vyver J, De Kooning JDM, Meersman B, Vandevelde L, Vandoorn TL.
Droop control as an alternative inertial response strategy for the synthetic inertia
on wind turbines. IEEE Trans Power Syst 2016;31(2):1129–38.
[77] AEMO. Technology Capabilities for Fast Frequency Response. GE-FFR-Advisory-
Report; 2017. Available at: https://www.aemo.com.au//media/Files/Electricity/
NEM/Security_and_Reliability/Reports/2017-03-10-GE-FFR-Advisory-Report-
Final—2017-3-9.pdf.
[78] Peter Fairley. Can Synthetic Inertia from Wind Power Stabilize Grids?”, IEEE
Spectrum; 2016. Available at http://spectrum.ieee.org/energywise/energy/
renewables/can-synthetic-inertia-stabilize-power-grids.
[79] ENTSO-E. Draft Network Code for Requirements for Grid Connection applicable to
all Generators, European Network for Transmission System Operators for
Electricity; 2012.
[80] Transmission Provider Technical Requirements for the Connection of Power Plants
to the Hydro Québec Transmission System, Revision March 2006. Available at
www.hydroquebec.com.
[81] Market Rules for the Ontario Electricity Market, June 4 2014. Available at www.
ieso.ca.
[82] LRP I RFP Backgrounder – Connection, March 10, 2015’. Available at www.ieso.ca.
[83] Anexo XI – Requisitos Técnicos Mínimos para a Conexão de Centrais Geradoras
Eólicas, November 2015. Available at http://www.aneel.gov.br.
[84] Acciona-energia.com. ACCIONA Energía, pioneer in providing electric power
system adjustment only using wind power; 2017. Available at www.acciona-
energia.com.
[85] DS3 System Services. Technical Definitions, Decision Paper, SEM-13-098; 2013.
Available at http://www.allislandproject.org/.
[86] AEMO. Future Power System Security Program; 2016. Available at aemo.com.au.
[87] Nordel. Agreement regarding operation of the interconnected Nordic power
system (System Operation Agreement). Technical report; 2006. Available at www.
entsoe.eu.
[88] UCTE. P1 – Policy 1: Load-Frequency Control and Performance [C]. Technical
report, 2009. Available at www.eeh.ee.ethz.ch.
[89] Agora-energiewende. Flexibility in thermal power plants: With a focus on existing
coal-fired power plants; 2017. Avalable at: https://www.agora-energiewende.de/
fileadmin/Projekte/2017/Flexibility_in_thermal_plants/115_flexibility-report-
WEB.pdf.
[90] Euroelectric. Hydro power: supporting a power system in transition; 2015.
Available at: http://www.eurelectric.org/media/180752/hydropower-final-lr-
2015-2120-0003-01-e.pdf.
[91] IEA Greenhouse Gas R & D Programme (IEAGHG). Operating flexibility of power
plants with CCS; 2012. Available at http://hub.globalccsinstitute.com/sites/
default/files/publications/104631/operating-flexibility-power-plants-ccs.pdf.
[92] Black & Veatch Corporation. Cost and Performance Data for Power Generation
Technologies. Report for National Renewable Energy Laboratory (NREL); 2012.
Available at https://www.bv.com/docs/reports-studies/nrel-cost-report.pdf.
[93] Hansen AD. Evaluation of power control with different electrical and control
concepts of wind farms, Tech. rep., Project UpWind, Roskilde, Denmark; 2010.
[94] Ramtharan G, Ekanayake JB, Jenkins N. Frequency support from doubly fed in-
duction generator wind turbines. IET Renew Power Gen 2007;1:3–9.
[95] Attya AB, Hartkopf T. Wind farms dispatching to manage the activation of fre-
quency support algorithms embedded in connected wind turbines. Int J Electr
Power Energy Syst 2013;53:923–36.
[96] Hughes FM, Anaya-Lara O, Jenkins N, Strbac G. Control of DFIG-based wind
generation for power network support. IEEE Trans Power Syst 2005;20:1958–66.
[97] Mauricio JM, Marano A, Gomez-Exposito A, Ramos JLM. Frequency regulation
contribution through variable speed wind energy conversion systems. IEEE Trans
Power Syst 2009;24:173–80.
[98] Gautam D, Goel L, Ayyanar R, Vittal V, Harbour T. Control strategy to mitigate the
impact of reduced inertia due to doubly fed induction generators on large power
systems. IEEE Trans Power Syst 2011;26:214–24.
[99] de Almeida R, Castronuovo ED, Lopes JAP. Optimum generation control in wind
parks when carrying out system operator requests. IEEE Trans Power Syst
2006;21(2):718–25.
[100] Janssens NA, Lambin G, Bragard N. Active power control strategies of DFIG wind
75
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
turbines. In: 2007 IEEE power tech Lausanne; 2007. p. 516–21.
[101] Kanellos FD, Hatziargyriou ND. Control of variable speed wind turbines in is-
landed mode of operation. IEEE Trans Energy Convers 2008;23(2):535–43.
[102] Courtecuisse V, Robyns B, Francois B, Petit M, Deuse J. Variable speed wind
generators participation in primary frequency control. Wind Eng
2008;32(3):299–318.
[103] Holdsworth L, Ekanayake JB, Jenkins N. Power system frequency response from
fixed speed and doubly fed induction generator-based wind turbines. Wind Energy
2004;7(1):21–35.
[104] Conroy JF, Watson R. Frequency response capability of full converter wind turbine
generators in comparison to conventional generation. IEEE Trans Power Syst
2008;23(2):649–56.
[105] Moutis P, Loukarakis E, Papathanasiou S, Hatziargyriou ND. Primary load fre-
quency control from pitch-controlled wind turbines. In: 2009 IEEE power tech
Bucharest; 2009. p. 1–7.
[106] Moutis P, Papathanassiou SA, Hatziargyriou ND. Improved load-frequency control
contribution of variable speed variable pitch wind generators,“. Renewable Energy
2012;48:514–23.
[107] Ghosh S, Kamalasadan S, Senroy N, Enslin J. Doubly fed induction generator
(DFIG)-based wind farm control framework for primary frequency and inertial
response application. In: IEEE Transactions on Power Systems, 2016; 31(3):
1861–1871.
[108] Vidyanandan KV, Senroy N. Primary frequency regulation by deloaded wind
turbines using variable droop. IEEE Trans Power Syst 2013; 28(2): 837–846.
[109] Hwang M, Muljadi E, Park JW, Sørensen P, Kang TC. Dynamic droop-based inertial
control of a doubly-fed induction generator. IEEE Trans Sustainable Energy 2016;
7(3): 924–933.
[110] Buckspan A, Aho J, Fleming P, Jeong Y, Pao L. Combining droop curve concepts
with control systems for wind turbine active power control. In: 2012 IEEE Power
Electronics and Machines in Wind Applications, Denver, CO; 2012. p. 1–8.
[111] National Grid. Guidance notes-Power park module; 2012. Available at: www.2.
nationalgrid.com/uk.
[112] European Network of Transmission System Operators for Electricity, “Who is
ENTSO-E,” [Online]. Available: https://www.entsoe.eu/about-entso-e/Pages/
default.aspx.
[113] AEMC. Feasibility of Fast Frequency Response Obligations of New Generators;
2017. Available at http://www.aemc.gov.au/getattachment/661d5402-3ce5-
4775-bb8a-9965f6d93a94/AECOM-report-Feasibility-of-FFR-obligations-of-new.
aspx.
[114] Thorncraft SR, Outhred HR. Experience with Market-based ancillary services in
the Australian national electricity market. In: Proceedings of the IEEE PES General
Meeting, Tampa Florida, USA, June 24–28; 2007.
[115] Stanton NK, Giri JC, Bose A. Energy management. In: Grigsby LL, editor. Power
system stability and control. Boca Raton, FL: CRC Press; 2007.
[116] IEEE. Standard definitions of terms for automatic generation control on electric
power systems, IEEE committee report. IEEE Trans Power Appl Syst 1970; PAS-89:
1356–64.
[117] WECC. Area Control Error (ACE) Calculation- Description; 2006, website: www.
wecc.biz/library.
[118] NERC. Violation Severity Level Matrix-Reliability Standard; 2014. Available at
www.nerc.com.
[119] Bevrani H, Daneshfar F, Daneshmand RP. Intelligent Power System Frequency
Regulations Concerning the Integration of Wind Power Units, Chapter in Wind
Power Systems: Applications of Computational Intelligence, Springer Book Series
on Green Energy and Technology, Springer-Verlag, Heidelberg; 2010. p. 407–37.
[120] AEMO. Australian Wind Energy Forecasting System (AWEFS); 2016.
[121] Poul Sorensen, Anca d. Hansen, Florinn Iov, Frede Blaabjerg and Martin H.
Donovan. Wind farm models and control Strategies, Riso-R Report, Denmark;
2005. Available at orvit.dtu.dk.
[122] Tarnowski GC, Kjær PC, Dalsgaard S, Nyborg. A. Regulation and frequency re-
sponse service capability of modern wind power plants. In: Proc. power and en-
ergy society general meeting, 2010 IEEE, 25–29 July, Minneapolis, USA; 2010.
[123] Garcia AD. Models for impact assessment of wind-based power generation on
frequency control. Chakraborty A, Ilić M, editors. Control and optimization theory
for electric smart grids, vol. 3. Berlin: Springer-Verlag; 2012.
[124] Ela E, Milligan M, O'Malley M. A flexible power system operations simulation
model for assessing wind integration. In: 2011 IEEE power and energy society
general meeting. IEEE; 2011. p. 1–8.
[125] Bart CU, Madeleine G, Wil LK, Gerardus C. Performance of automatic generation
control mechanisms with large-scale wind power. In: Nordic Wind Power
Conference, Nwpc’2007, Roskilde, Denmark; 2007.
[126] Bijami E, Farsangi MM. Networked distributed automatic generation control of
power system with dynamic participation of wind turbines through uncertain
delayed communication network. IET Renewable Power Gen 2017; 11(8):
1254–69, 6 28 2017.
[127] Amenedo LR, Arnalte S, Burgos JC. Automatic generation control of a wind farm
with variable speed wind turbines. IEEE Trans Energy Convers
2002;17(2):279–84.
[128] Aho J, Buckspan A, Laks J, Fleming P, Jeong Y, Dunne F, Churchfield M, Pao L,
Johnson K. A tutorial of wind turbine control for supporting grid frequency
through active power control. In: Montreal: Proceedings of the 2012 American
Control Conference; 2012.
[129] Hansen Anca D, Sørensen Poul, Iov Florin, Blaabjerg Frede. Centralised power
control of wind farm with doubly fed induction generators, , 31(7). Renewable
Energy 2006;31(7):935–51. ISSN 0960-1481.
[130] Hansen AD, Altin M, Iov F. Provision of enhanced ancillary services from wind
A. Aziz et al.
power plants – examples and challenges. Renewable Energy 2016;97:8–18.
[131] Grunnet JD, Soltani M, Knudsen T, Kragelund M, Bak T. Aeolus toolbox for dy-
namic wind farm model simulation and control. In: Proc. Eur. wind energy conf.;
2010. p. 3119–29.
[132] Chang-Chien LR, Sun CC, Yeh YJ. Modeling of wind farm participation in AGC.
IEEE Trans Power Syst 2014; 29(3): 1204–11.
[133] Zhang W, Fang K. Controlling active power of wind farms to participate in load
frequency control of power systems. IET Gen Transm Distrib 2017; 11(9):
2194–2203.
[134] Basit, Abdul, et al. Wind power integration into the automatic generation control
of power systems with large-scale wind power. J Eng; 2014.
[135] Michael Antonishen, Han HY, Brekken TKA, von Jouanne A, Alex Yokochi,
Halamay DA, Song J, Naviaux DB, Davidson JD, Alex Bistrika. A Methodology to
enable wind farm participation in automatic generation control using energy
storage devices. In: IEEE power and energy society general meeting, IEEE, 2012. p.
1–7, 22–26.
[136] Luo C, Ooi BT. Frequency deviation of thermal power plants due to wind farms”.
IEEE Trans Energy Convers 2006;21(3):708–16.
[137] Luo C, Far HG, Banakar H, Keung PK, Ooi BT. Estimation of wind penetration as
limited by frequency deviation”. IEEE Trans Energy Convers 2007;22(3):783–91.
[138] He G, Chen Q, Kang C, Xia Q, Poolla K. Cooperation of wind power and battery
storage to provide frequency regulation in power markets. IEEE Trans Power Syst
2017;32(5):3559–68.
[139] Hassan Bevrani, Pourya Ranjbar Daneshmand. Fuzzy Logic-based load-frequency
control concerning high penetration of wind turbines. EEE Syst J 2012; 6(1).
[140] Bevrani H, Daneshfar F, Hiyama T. A new intelligent agent-based agc design with
real-time application. IEEE Trans Syst Man Cybernet Part C (Appl Rev)
2012;42(6):994–1002.
[141] Eric Schubert, Sam Zhou, Tony Grasso, Grace Niu. A Primer on Wholesale Market
Design, Market Oversight Division Staff Report, Public Utility Commission of
Texas; 2002. Available at: https://sites.hks.harvard.edu/hepg/Papers/TXPUC_
wholesale.market.primer_11-1-02.pdf.
[142] EWEA. The Economics of Wind Energy; 2009. Available at http://www.ewea.org.
[143] EWEASupport Schemes for Renewable Energy, RE-Xpansion project; 2002.
Available at http://www.ewea.org.
[144] Andrade, Juan, Dong, Yingzhang, and Baldick, Ross. Impact of renewable gen-
eration on operational reserves requirements: When more could be less. White
Paper UTEI/2016-10-1, 2016, Available at energy.utexas.edu/the-full-cost-of-
electricity-fce/.
[145] Emiliano Bellini. Italy opens ancillary services market to pilot renewable energy
and storage projects [on;ine]; 2017. Avaliable at www.pv-magazine.com/2017.
[146] AEMO. AEMO and ARENA looking to evolve traditional electricity market; 2017.
Available at: www.aemo.com.au.
[147] Ray W Brown, Hayden N Scott-Dye. Te Uku Wind Farm – Planning and Operation
of a Deeply Embedded Power Plant with Advanced Ancillary Services. Cigre
conference; 2012. Available at: www.cigre.org/content/download/17087/
680725/version/1/file/C1_108_2012.pdf.
[148] Giebel G, Cline J, Frank H, Shaw W, Pinson P, Hodge B-M, Kariniotakis G, Madsen
J, Möhrlen C. Wind power forecasting: IEA Wind Task 36 & future research issues.
Journal of Physics: Conference Series 753, IOP Publishing; 2016. Available at:
www.nrel.gov/docs/fy17osti/67542.pdf.
[149] Zhang Yao, Wang Jianxue, Wang Xifan. Review on probabilistic forecasting of
wind power generation. Renew Sustain Energy Rev 2014;32:255–70.
[150] Sue Ellen Haupt, William P. Mahoney, Keith Parks. Wind Power Forecasting, Book
Chapter for: Weather Matters for Energy, edited by Alberto Troccoli, Laurent
Dubus, Sue Ellen Haupt. Springer, New York; 2014.
[151] Foley Aoife M, Leahy Paul G, Marvuglia Antonino, McKeogh Eamon J. Current
methods and advances in forecasting of wind power generation. Renewable
Energy 2012;37(1):1–8.
[152] González-Aparicio I, Zucker A. Impact of wind power uncertainty forecasting on
the market integration of wind energy in Spain. Appl Energy 2015;159:334–49.
[153] Kariniotakis G, Martí I, Casas D, Pinson P, Nielsen TS, Madsen H, Giebel G, Usaola
J, Sanchez I, Palomares AM, Brownsword R, Tambke J, Focken U, Lange M, Louka
P, Kallos G, Lac C, Sideratos G, Descombes G. What performance can be expected
by short-term wind power prediction models depending on site characteristics? In:
Proceedings of the 2004 European Wind Energy Association Conference EWEC’04,
London, UK; 2004, [Online]. Available at: www.2004ewec.info/files/23_1400_
76
Sustainable Energy Technologies and Assessments 26 (2018) 47–76
georgekariniotakis_01.pdf.
[154] Ana Rodriguez. Red Electrica – Key note speech. Wind generation forecasting at
REE. IEA wind task workshop: Experiences in using Wind Power Predictions and
Gaps in Forecasting Research. Barcelona; 2016. Available at: http://www.
ieawindforecasting.dk/news/2016/05/workshop-9-june-2016-in-barcelona?id=
6270327c-6add-4128-ac04-823bf6031c01.
[155] Royal Academy of Engineering. Wind energy implications of large-scale deploy-
ment on the GB electricity system. UK; 2014.
[156] Infigen. Submission to FPSS program report; 2016. Available at: www.aemo.com.
au/-/media/Files/Electricity/NEM/Security_and_Reliability/Reports/
Submissions/Infigen-submission-to-FPSS-Progress-Report-Aug-2016.pdf.
[157] Sharman H. Why wind power works for Denmark. Proc Inst Civ Eng – Civ Eng
2015;158(2):66–72.
[158] Tarca Silvio, Roughan Matthew, Ertugrul Nesimi, Bean Nigel. Dispatchability of
wind power with battery energy storage in South Australia. Energy Proc
2017;110:223–8.
[159] Baone C, DeMarco C. Saturation-bandwidth tradeoffs in grid frequency regulation
for wind generation with energy storage. In: Proceedings of the 2011 IEEE PES
innovative smart grid technologies (ISGT), Hilton Anaheim, CA, USA; 2011.
p. 1–7.
[160] Mendis N, Muttaqi K, Sayeef S, Perera S. Control coordination of a wind turbine
generator and a battery storage unit in a remote area power aupply system. In:
Proceedings of the 2010 IEEE power and energy society general meeting,
Minneapolis, MN, USA, 25–29 July 2010. p. 1–7.
[161] Banham-Hall DD, Taylor GA, Smith CA. Irving MR. Frequency control using va-
nadium redox flow batteries on wind farms. In: Proceedings of the 2011 IEEE
Power and Energy Society General Meeting, San Diego, CA, USA; 2011. p. 1–8.
[162] Han H, Brekken TKA, von Jouanne A, Bistrika A, Yokochi A. In-lab research grid
for optimization and control of wind and energy storage systems. In: Proceedings
of the 2010 49th IEEE conference on decision and control (CDC), Atlanta, GA,
USA; 2010. p. 200–5.
[163] Klessmann C, Nabe C, Burges K. Pros and cons of exposing renewables to elec-
tricity market risks a comparison of the market integration approaches in
Germany, Spain, and the UK. Energy Policy 2008;36(10):3646–61.
[164] Spudíc V, Baotíc M. Fast coordinated model predictive control of large-scale dis-
tributed systems with single coupling constraint. In: Proceedings of the 2013
European control conference, Zürich, Switzerland; 2013. p. 2783–2788.19.
[165] Madjidian D, Mirkin L. Distributed control with low-rank coordination. IEEE Trans
Control Network Syst 2014;1(1):53–63.
[166] REE. P.O 9.0 - Información intercambiada por el operador del sistema. Abgerufen
am 01. 03 2013 von REE - Red Electrica de España; 2012: http://www.ree.es/
operacion.
[167] Graeme A. Chown. Economic analysis of relaxing frequency control. (Thesis);
2007. Available at: http://146.141.12.21/bitstream/handle/10539/5326/
G%20Chown%20PhD%20Thesis%20final%20.pdf?sequence=1&isAllowed=y.
[168] ENTSO-E. Ten-year network development plan, Tech. rep., European Network of
Transmission System Operators for Electricity; 2012.
[169] Kate Summers. Fast Frequency Service – Treating the symptom not the cause?
2017. Available at: http://www.pacifichydro.com.au/english/information-centre/
industry-and-policy/pacific-hydro-exposes-a-major-threat-to-power-system-
security/.
[170] Burton Tony, Jenkins Nick, Sharpe David, Bossanyi Ervin. Wind energy handbook.
John Wiley & Sons; 2011.
[171] Freebury G, Musial W. Determining Equivalent Damage Loading for Full-Scale
Wind Turbine Blade Fatigue Tests, NREL/CP-500-27510, Tech. Rep; 2000.
[172] Ragheb A, Ragheb M. Wind turbine gearbox technologies. In: 1st international
nuclear & renewable energy conference (INREC), Amman; 2010. p. 1–8.
[173] Nivedh BS. Major failure in wind turbine components nad importance of periodic
inspection. DEWI (UL International GmbH); 2014. Available at: www.dewi.de/
dewi_res/fileadmin/pdf/papers/WindInsider_November_2014.pdf.
[174] Euroelectric. Integrating intermittent renewables sources into the EU electricity
system by 2020: challenges and solutions; 2010. Available at: www.eurelectric.
org/media/45254/res_integration_paper_final-2010-030-0492-01-e.pdf.
[175] National Grid. Frequency Response Energy Payment; 2013.
[176] Harley Mackenzie, Jonathon Dyson. NEM FCAS causer pays procedure, HARD
Software & Greenview Strategic Consulting; 2017. Available at www.aemo.
com.au.