3140 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 34, NO.

4, JULY 2019

A Review of Power System Flexibility With High

Penetration of Renewables
Baraa Mohandes , Member, IEEE, Mohamed Shawky El Moursi , Senior Member, IEEE,
Nikos Hatziargyriou , Fellow, IEEE, and Sameh El Khatib, Member, IEEE

Abstract—The notion of secure operation of power systems, with pk The probability of occurrence of scenario k.
its present semantic, dates back to the last decade. Since then, r Vector of reserve §r is a system parameter in
tremendous research effort has investigated secure operation of the
power system. Nevertheless, operators are still faced with security the case of implicit security only.
issues, with even larger uncertainty, however, caused by the inte- R Vector of governor-droop for generation units.
gration of renewable energy sources (RES). The system’s ability to Δmax {g, d} Vector of maximum ramping (up/down) capa-
cope with the volatility of RES and load becomes a critical issue in bility of generation or demand, respectively.
power system operation. This paper surveys the literature on the v Value of lost load (VOLL).
concepts of power system flexibility, indices of flexibility, and im-
plementation of the concept of flexibility in power system security. Decision Variables:
The paper proceeds to review the origin of the reserve problem,
the meaning of reserve, its technical classification and related eco- d0 , dkp,s,t Vector of demand size in pre-contingency, or
nomical aspects. The paper highlights the effect of renewables on in each interval of post-contingency scenario
these aspects, and suggests new research directions. (k) operation.
g 0 , gp,s,t
k
Vector of generation size in pre-contingency, or
Index Terms—Reserve, Flexibility, Security, Renewable En-
ergy System, Regulation, Load-Following, Contingency, Spinning, in each interval of post-contingency scenario
Survey, Review. (k) operation.
Lk Vector of involuntary load shedding (decision
NOMENCLATURE variable).
u Vector binary variables denoting whether a ma-
Set Indices:
chine/unit is committed (1) or not (0).
i Generation units: i ∈ I = {1, . . . , I}.
System Variables (Byproducts of the Solution):
k Contingency scenario. k ∈ K = {0, 1, . . . , K}.
δk Vector of bus angles in scenario (k).
System Parameters:
fk Vector of transmission line flow limits in sce-
C(·) Cost function. nario (k).
Bk The matrix of network susceptance in scenario ΔF Post-contingency frequency deviation.
(k). B changes only when a contingency in- μk The dual variable for the constraint of nodal
volves the failure of a transmission line. power balance for scenario k.
Hk Network flow sensitivity matrix in scenario (k) Subscripts:
(Similar to J Jacobian matrix).
(·){p,s,t} Denotes a vector or variable correlated with
Manuscript received July 22, 2018; revised November 16, 2018; accepted primary, secondary or tertiary reserve.
January 28, 2019. Date of publication February 5, 2019; date of current version (·){g ,d,r } Denotes a vector or variable correlated with
June 18, 2019. This work was supported by the CIRA Project-2018-37, Khalifa
University, Abu Dhabi, UAE. Paper no. TPWRS-01124-2018. (Corresponding
generation demand, or reserve.
author: Mohamed Shawky El Moursi.) min, max
{ }
(·) up, dn
B. Mohandes is with the Electrical and Computer Engineering Department,
Khalifa University of Science and Technology, Abu Dhabi 54224, UAE (e-mail:, Denotes a vector or variable correlated with
baraa.mohandes@ku.ac.ae; mohamed.elmoursi@ku.ac.ae). max capacity, min capacity, up reserve or down
M. S. El Moursi is with the Electrical and Computer Engineering Department, reserve.
Khalifa University of Science and Technology, Abu Dhabi 54224, UAE, on leave
from the Faculty of Engineering, Mansoura University, Mansoura 35516, Egypt
(e-mail:,mohamed.elmoursi@ku.ac.ae). I. INTRODUCTION
N. Hatziargyriou is with the Power Division of the Electrical and Com-
HE electricity sector is responsible for the largest share
puter Engineering Department, National Technical University of Athens, Athens
15773, Greece (e-mail:,nh@power.ece.ntua.gr).
S. El Khatib is with the Engineering Systems and Management, Mas-
T of global carbon emissions [1], [2]. Renewable energy
sources (RES) are being integrated in the power system at a fast
dar Institute of Science and Technology, Abu Dhabi 54224, UAE (e-mail:,
selkhatib@masdar.ac.ae). rate, pushed by political pressure to subdue carbon emissions.
Color versions of one or more of the figures in this paper are available online RES differ largely from conventional generation units. Thus, the
at http://ieeexplore.ieee.org. growing penetration of RES in the power system presses drastic
Digital Object Identifier 10.1109/TPWRS.2019.2897727

0885-8950 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
MOHANDES et al.: REVIEW OF POWER SYSTEM FLEXIBILITY WITH HIGH PENETRATION OF RENEWABLES
changes on power system operation. Three attributes compose
the main difference between RES and conventional generation
[3]. They are:
1) RES production is strongly dependent on weather condi-
tions, therefore, it is highly unpredictable, and volatile [4].
At high RES penetration levels, variations in RES output
might have serious ramifications [5]–[11]. One extreme
example of this problem is voltage collapse in a system
with high penetration of Photovoltaic (PV) generation,
due to partial clouding [12].
2) RES has very low generation cost, and government sub-
sidy programs may also render this cost negative to private
RES owners. Thus, RES will probably be dispatched at
their maximum output.
3) Besides short-term fluctuations of RES output, RES ex-
hibits medium-term (daily, weekly) and long-term (sea-
sonal) variations.
Electronically interfaced RES reduce the system’s inertia,
which exacerbates the RES volatility problem [13], [14]. Work
in [15]–[18] highlights the need for sufficient resources char-
acterized by flexibility, i.e., fast response and spare generation
capacity, to tackle RES volatility. Flexibility in power systems
denotes the ability of the system components to adjust their
operating point, in timely and harmonized manner, to accom-
modate expected, as well as unexpected, changes in system
operating conditions [3]. According to this definition, flexibility
is not exclusively needed for load and RES variation, but also
equipment contingencies. In general, power system flexibility is
a basic prerequisite for allowing higher RES penetration [14],
[19]. Moreover, contemporary power systems become more de-
centralized and digitized adopting the smart-grid concept [20],
[21]. These factors call for revisiting the way these systems are
operated.
The literature is rich with studies on maintaining security
in systems with high penetration of RES. Several methods have
been published to determine the optimum reserve size and price,
as enumerated in this paper. However, the literature lacks a com-
prehensive review and assessment of these methods. Moreover,
the literature also lacks a framework that reconciles the concepts
of flexibility and reserve. This paper aims to fill these voids. It
also highlights the direct effects of RES on the secure operation
requirements, in terms of flexibility and reserves.
Power system security is not limited to the performance of
the system in short-term, i.e., events taking place within the
next few minutes, hours or days, but also concerns the system’s
aptitude in the medium and long term. Meeting demand in the
future means that the seasonal variation of load and load growth
are satisfied by expanding the system’s generation capacity. It
should be noted, that the increase of RES penetration at the
expense of more flexible thermal units does not guarantee suffi-
cient ramping capability in the system. This review focuses on
short-term security.
This paper is organized as follows. Section II familiarizes
the reader with power system flexibility indices, sources, and
optimization. Section III discusses the effects of RES on power
system security and flexibility. Section IV views power system
flexibility from the perspective of the electricity market, and
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
3141
addresses the economic challenges pertaining to flexibility. A
smart RES management strategy is proposed and explained in
Section V.
II. POWER SYSTEM FLEXIBILITY
System operation is continuously faced with the challenge
of maintaining balance between generation and demand. This
balance is perturbed by three types of events of different time
frames: rapid random fluctuations, slow periodical fluctuations
and rare abrupt changes [22]. The variable production of RES
combined with fluctuations from the demand side results in a
net-load profile of high volatility. Flexibility is defined in [3],
[23]–[26] as the system’s ability to accommodate variation in
net-load, by modulating the feed-in or feed-out of power across
the grid over time. Flexibility requirements have historically
been fulfilled by central generation units [25], where the main
sources of variation and uncertainty were demand and equip-
ment failures.
A. Sources of Flexibility
Besides central generation, several potential sources of flex-
ibility can be recognized at the distribution level [27]–[29]:
Demand response, energy storage, and distributed energy gener-
ators providing ancillary services. Further sources of flexibility
can be provided by grid interconnections and combined-cycle
units.
1) Demand Response (DR): Demand response can be
achieved by suitable tariffs aiming to motivate customers to
change their consumption patterns [30], [31]. This possibility
involves both generation and demand in mitigating demand-
supply mismatch is the sole responsibility of the generation side
only. DR may involve few large industrial loads, or numerous
small consumers.
Different designs of DR programs have been proposed in the
literature, characterized by different objectives, load involve-
ment levels, and decision variables. A comprehensive review of
DR programs is provided in [30]. Regarding system security and
flexibility, small loads can respond faster than thermal conven-
tional units [28]. Industrial loads provide the greatest potential
for DR program applications, since they comprise the largest
consumers in the system. Thus, they are capable of providing
sizable curtailments. In addition, a large industrial complex may
commission its own generation units on-site, while also being
connected to the grid. DR Programs are likely to reduce the price
of electricity through reduction of transmission losses, and the
price of security [32].
It should be noted that the application of DR programs as a
flexibility source faces various challenges. Conventional meters
and non-variable billing policies are not adequate for imple-
menting time-varying tariffs. Smart meters, secure and reliable
communication infrastructure, and responsive loads must be
available before embarking on this enterprise [9]. Attractive tar-
iffs must be in place to compensate possible interruptions in
the production process of industrial entities. Practical obstacles
of implementation of demand response are presented by [33].
3142
More challenges facing the wide implementation of DR pro-
grams are discussed in Section IV-B.
2) Energy Storage Systems (ESS): ESS add valuable flex-
ibility to the power system. They allow momentary mismatch
between supply and demand, such that this mismatch is
resolved over time [28]. Pumped hydro energy storage (PHES)
still represents the highest amount of total storage capacity
worldwide [34], however, smaller ESS are becoming more
popular, too. These technologies include different types of
batteries storage (BESS), ultracapacitors, superconducting
inductors (SMES), flywheels and compressed air (CAES). A
comparison of ESS technologies in terms of nominal capacity
and discharge time is provided in [28], [34]. ESS that discharge
their full capacity rapidly (i.e.,: within few seconds) contribute
to flexibility in the system’s primary frequency response
only.
Among the above technologies, significant effort has been
placed in improving capacity and capital costs of BESS. BESS
have zero or very-low minimum power output limit, short
start up and shutdown times, and no minimum up/down times
[35]. However, they suffer from limitations in the number of
charge/discharge cycles.
3) Dispatchable Distributed Generators (DG): Distributed
generation connected at the distribution level can have fast or
slow response to power mismatches, and thus, provide local
flexibility. The contribution of these units is comparable to the
contribution of DR programs, and can be incentivized in the
same way.
4) Grid Interconnection (GI): Interconnection with neigh-
boring grids enlarges the number of flexibility sources available
to all subsystems [7], [36]. It also opens the door for electricity
trade. Successful implementation requires investments in trans-
mission infrastructure and policy-making. A grid extending over
multiple time-zones, however, means that the peak-load period
for each time-zone occurs at different times. Solar irradiance is
also strongly connected to the time of the day. Interconnection
reserve is the term used in [37] to describe the application of
neighboring reserves to face wind volatility. The problem of op-
timum grid interconnection is approached from a security and
robustness perspective in [38].
5) Multi-Mode Operation of Combined Cycle Units: Ex-
haust heat from thermal units can be recovered and concentrated
to drive a steam-turbine to generate more electricity. Complex
units are designed for this process, featuring much higher effi-
ciency than classical thermal units. The heat recovery process
applies extreme physical stress on the equipment in the down-
stream cycles. Equipment that withstand such harsh conditions
can not be exposed to fast changes in physical stress [39]. There-
fore, these units have large start-up time, shut-down time, and
very poor ramping capability [39]. Forfeiting the exhaust-heat
and disabling the downstream process allow very short start-up
time and high ramping capability of the upstream cycle. How-
ever, these combined-cycle units must be designed apriori to
support this function. Although they incur higher maintenance
cost and reduced lifetime, these disadvantages are weighed
against potential benefits of providing flexibility to optimum
operation in [40], [41].
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 34, NO. 4, JULY 2019
B. Flexibility in Technical Terms
Power system flexibility expressed by its components has
been defined in the pioneering work of Makarov et al. [11] This
abstract, yet effective, representation opens the door for quan-
titative evaluation of flexibility in power systems. Accordingly,
flexibility is characterized by three indices: ramping limit (ρ),
power capacity (π) and energy capacity (). Each index is the
time derivative of the previous index, as described in Eq. (1).
The indices are discussed next:
  
ρ dt dt = π dt =  (1)
r Ramping Limit (ρ) is the maximum change a unit can
effect on its operating point in a certain time. This index
is a manifestation of the unit’s internal dynamics. For ex-
ample, a coal unit dispatched at its minimum power output
Pm in , cannot adjust its output to the maximum limit within
a short period of time. Similarly, large industrial loads par-
ticipating in Demand-Response have a maximum ramping
limit, dictated not only by device dynamics, but also by
production limitations.
r Power Capacity (π) refers to the minimum and maximum
power outputs of any generation source. This index is rep-
resented by inequality constraints in the OPF problem.
r Energy Capacity () concerns the fuel or energy supply of
a power source. In Fig. 3a, supply of power is interrupted
after a period of time, due to depletion of energy supply.
The fuel supply for central generation units is usually se-
cured for weeks in advance, and can be considered virtually
endless. Finite energy capacity, however, constrains ESS
and dispatchable DGs, such as local diesel engines.
Some studies, such as [24], include ramp-duration as the
third index of flexibility, instead of energy capacity. Ramp-
duration denotes the time period that a unit can continue to
change its output. In practice, this index is a realization of the
power capacity limits and ramp rate. Therefore, the original
classification [11] is considered sufficient. Flexibility provided
by different sources is characterized by these three indices. For
example, large thermal generation units have slow ramping rate
compared to gas turbines. In contrast, battery bank systems
can change their output quickly, but are limited in their energy
and power capacities. Figs. 1a and 1b demonstrate the typical
response for a number of sources of flexibility. The limits of each
device in light of the three indices of flexibility can be inferred
from their response. It is important to note that in the case of DR,
the actual response time and power magnitude depend greatly
on the type of device involved.
C. A New Flexibility Index: Response-Time (θ)
Central generation units are equipped with control systems
such that they automatically respond to system contingencies, if
they are online. In this case, the three flexibility indices describe
sufficiently well the available flexibility in the power system.
Considering more decentralized sources of flexibility, e.g., DR,
different ramp-rates, and response times, can be noted, as shown
in Fig. 1c. The response-time is a critical factor in planning
MOHANDES et al.: REVIEW OF POWER SYSTEM FLEXIBILITY WITH HIGH PENETRATION OF RENEWABLES
Fig. 1. Flexibility indices for different sources of flexibility.
the system’s contingency response. Regulating reserve is also
sensitive to this attribute, however, to a lesser degree.
The response of a single small load is more random. A profile
of numerous small loads combining the response characteristics
of all its small contributors, can be represented by a probability
distribution function (PDF) which may exhibit different char-
acteristics and shape. For example, the response curve of an
aggregated load may be non-linear or even non-monotonically
increasing. In the case of residential loads, the response time of
intelligent devices depends greatly on the processing and com-
munication delays of the devices. In contrast, load curtailment
that requires human response has a response time which greatly
depends on human behavior. For example, commands to delay
the charging of an EV might be delayed, or even declined by the
EV owner. Understanding and tackling this phenomenon is very
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
3143
Fig. 2. Cycling load profile, typical of VSDs.
important for the success of any load aggregator at providing
flexibility to the system.
The DR of large industrial loads has less uncertainty, however,
this also depends on the particular type of the industrial process.
However, an uncertainty component never seizes to exist in
some cases. For example, industrial load #1 in Fig. 1c responds
almost immediately to a request-for-curtailment, however, the
rate of reduction is small. This is typical for processes like
chilling compressors, electrolysis, mixing and assembly lines
[43]. In contrast, industrial load #2 is able to make large abrupt
reductions but only after a certain period of time. This is typical
of industrial processes such as crushing, grinding and milling
[43]. Processes based on variable speed drives (VSD) exhibit a
cyclic consumption profile as depicted in Fig. 2. In such case,
the delay in response to DR commands can not be determined
exactly, however, it is restricted within a range θ ∈ [0, τ ].
In light of this discussion, we propose a fourth flexibility
index, the Response Time, denoted as (θ). This index is applica-
ble mainly to distributed units. The literature on flexibility and
contingency response, such as [9], [43], incorporate response
time in their classification of load types. However, no work has
systematically investigated its particular effect on the system
behavior, and its economical value at sufficient depth.
Calculating a monetary value for response-delay (θ) would
help Virtual Power Plants (VPP) and load aggregators, who
provide flexibility services, to manage better their profiles, and
implement more effective contracts. Consequently, VPPs and
load aggregators would receive clear and direct incentives to
upgrade their response profiles in the most effective manner.
This will be the subject of future publications by the authors.
Studies such as [30], [44], compare different DR programs,
and investigate the relevant suitability of each load category
{industrial, residential, commercial} for different DR designs.
It is interesting to investigate how the response-time/delay char-
acteristic refines the suitability of a certain type of load for a
certain DR program. This question has been addressed on a lim-
ited scale. For example, thermostatically-controlled loads are
considered the best candidates for direct load control DR pro-
grams [29], [30], since their short interruption does not cause
any discomfort to the consumers. Other types of loads are, yet,
to be analyzed in terms of their response-time.
An effective technique for extending DR controllability is
Conservation Voltage Reduction (CVR), where the supply volt-
age is reduced, within limits, to reduce the power consumption
of loads [45]. Generators excitation control, and FACTs devices
3144
Fig. 3. Graphical representation of the aspects of flexibility.
are fast enough to implement CVR in a time-frame of ms. Power
reductions from CVR are in the range of 1%-2% of total load.
CVR also depends on the types of loads. For instance, fixed
power loads adjust their impedance to restore their original con-
sumption level. Thus, CVR, alone, is not a sufficient source for
flexibility.
D. Visualization of Flexibility
The original three indices of flexibility can be represented
graphically [11], [23], as shown in Fig. 3. These indices reflect
the physical constraints of each flexibility source in the system.
Applications of this depiction are presented in several studies,
such as [23], [24]. Flexibility sources can change their operating
points within a finite range, to support the system in a contin-
gency. This range is bounded by the power capacity limits, and
the ramping rates. Fig. 3a demonstrates the response of a unit
in a hypothetical situation. The change in the unit’s output is
limited by its ramping rate (ρ± m ax ) until it reaches its capacity
limits (π ± ).
Fig. 3b depicts the bounds of the operating region of a hypo-
thetical unit operating in steady state at (P0 ), represented by the
origin. Moving from (P0 ) to point (a) requires a small increase
in power output at a high rate. An example of this is the start
of a small load. Alternatively, moving from point (P0 ) to (b)
requires a large but slow change in power output. An example
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 34, NO. 4, JULY 2019
of this is the daily cyclic load variations. When the unit moves to
a new operating point, the flexibility region shifts such that the
new operating point is the new origin (0,0). Hence, the dimen-
sions of the flexibility polytope are dependent on the present
operating point. New operating points can emerge in the first
and third quadrants only. For example, a new operating point
in the second quadrant requires reduction in the power output,
with a positive rate of change, which is meaningless.
The third index of flexibility, energy capacity () can be added
as a new dimension. In this graphical space, the flexibility poly-
tope will occupy only two of the eight octants of 3D space.
For complex systems consisting of different units, flexibility
regions can be consolidated to form the system’s overall flexi-
bility region [23]. A contingency event can be described by the
flexibility terms (π ∗ , ρ∗ , ∗ ) required to mitigate it. The power
system can safely handle a contingency scenario if the system’s
flexibility polytope encloses the resulting operating point.
Dvorkin et al. [24] develop the mathematical model for op-
timizing flexibility requirements. Maintaining a large flexibil-
ity polytope requires substantial infrastructure investments and
high operation cost. The objective of optimization is, hence,
minimization of the volume of the flexibility polytope, which
encloses at least a certain number of points (contingency sce-
narios). One of the shortcomings of the model proposed by [24]
is the assumption that the flexibility polytope takes the shape of
a right rectangular prism. Although this assumption simplifies
the calculation of the region’s volume, the polytope of flexibility
for a system with numerous different components may not even
be a convex shape.
The graphical representation of system flexibility inspires the
development of metrics to quantify this critical trait. Such met-
rics would facilitate incorporating flexibility in operation and
optimization. Zhao et al. [26] classify metrics into deterministic
and stochastic. Deterministic metrics report crisp features of the
system such as: the amount of reserve available, whether or not
the system is secure against a predetermined set of plausible
contingencies, and well-being assessment [25]. Stochastic met-
rics of flexibility report system’s flexibility in terms of statistical
probability of failure. Furthermore, Zhao et al. [26] and Lan-
noye et al. [47] distinguish between flexibility metrics used for
long-term planning of system expansion, and flexibility metrics
used for short-term (daily operations) planning.
Evaluating a system by the worst contingency it can handle,
is neither simple nor accurate. The largest values of each of the
three indices (i.e.,: π, ρ, ε) may belong to different contingency
scenarios. One event is not representative of the system’s ability
to accommodate other plausible events. In contrast, different
metrics for system flexibility can be derived from the concept of
the polytope of flexibility, such as the volume of the polytope, the
vertices or the extreme points of the polytope, and the number
of contingency points the polytope encloses. In fact, the number
of enclosed points was used in [24]. The polytope of flexibility
maps the system’s flexibility and security status to mathematical
probability. This indirect mapping is valuable for comparing
different power systems, with different designs and structures.
Network design and transmission congestions will always be
a major constraint over contingency mitigation plans [48]–[50].
MOHANDES et al.: REVIEW OF POWER SYSTEM FLEXIBILITY WITH HIGH PENETRATION OF RENEWABLES
At the same time, some units may have some power and ramping
capacity, however, this capacity is locked for accommodating
anticipated load changes in the next time period [51]. Because
of this, any metrics developed to evaluate specific flexibility
aspects should only be used to compare between different sit-
uations for the same power system. For example, comparing
the same system at different time periods, with different loads
and dispatch schedules. Similarly, such metrics are useful, even
necessary, for the purpose of investigating different flexibility
solutions.
The generation adequacy metric evaluates the total generation
capacity in the system, with disregard towards the deliverabil-
ity of this capacity [47]. One of the earliest metrics to quantify
deliverable flexibility, for long-term planning, is the insufficient
ramping-resource expectation (IRRE) [25]. Lannoye et al. [50]
reported the realizable flexibility (RF) of a system as the maxi-
mum factor by which the existing loads around the grid can be
scaled up while also maintaining the system’s security against
contingencies. A case-study by the national renewable energy
laboratory (NREL) [27] evaluated the impact of flexibility on
the system, via the economic carrying capacity index (ECC).
With the flexibility sources in the system fixed, ECC is the RES
penetration level where it is not economically feasible to add
any more RES to the grid. At this point, the levelized cost of
energy from RES is equal to the present cost of electricity in
the grid. The cost of additional necessary flexibility would out-
weigh the benefit from cheap RES energy. This index accounts
for transmission limitations by allowing RES to be integrated
anywhere in the grid.
Alternatively, the authors in [26] propose a framework where
flexibility is measured in terms of four aspects: cost, uncertainty
level, time-horizon for acting, and the set of available actions.
III. POWER SYSTEM RESERVE
System flexibility depends on reserves, made available
through ancillary services (AS). AS are defined as the services
which support the grid and facilitate delivery of more energy
[22]. Therefore, AS are essential for power system security.
This section reviews the definitions and classifications of re-
serves and outlines the inherent effect of RES penetration on
flexibility.
The concept of reserve is analyzed in [52] from the physical
and economic perspectives. Based on the conclusions of this
work, several studies have been performed aiming to investigate
further the concept of reserve. It is unfortunate that the nomen-
clatures have not been standardized, and many terms are used in
the literature interchangeably [53], [54]. A common definition
is the accumulated capacity of resources that enable the imple-
mentation of preventive and corrective security control actions.
This definition provides a false impression that reserve is used
exclusively for contingency events. This is caused by the na-
ture of reserve services and the fact that they are provided by
the same units. We believe that distinction should be established
between services employed in normal-conditions (regulating re-
serve), and services employed in abnormal-conditions (contin-
gency reserve). Hour-Ahead (HA) adjustment, load-following
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
3145
(LF) and regulation are part of the daily challenges in operating
the power system in normal conditions. Operating-reserve is re-
quired to accommodate deviation from the Unit-Commitment
(UC) schedule to cover load forecasting inaccuracies. Plans to
deploy them are set shortly before their actual deployment. On
the other hand, contingency-reserve is made available as a pre-
emptive security measure, in order to ensure power system se-
curity.
An anticipated event, such as the solar eclipse of 2017 in
USA, allows ample time for planning mitigative actions and
availing the necessary flexibility [55]. NREL reports that this
event was accommodated, and did not pose any threat to the
grid. Alternatively, in the event of an unexpected contingency, if
the flexibility is not availed for the system beforehand, the secu-
rity of the power system will be in jeopardy. Without sufficient
flexibility resources, maintaining system’s security might im-
pose curtailment of RES output [27], such that the conventional
units can remain online (i.e.,: P > Pm in ) to provide the neces-
sary flexibility. Operating RES below their maximum rating is
also known as RES curtailment [13]. For example, 1,100GWh
of wind energy, worth between 110M$ and 240M$, was spilled
in Spain in 2005 [53]. Another possible scenario is selling the
energy surplus, from conventional units and RES, at a negative
price [3]. Concerns over system security and flexibility moti-
vated setting strict operating limits for wind farms in [56], [57].
The following taxonomy for the different types of reserve is
proposed by [53]. The characteristics of both types of reserve
are loosely connected to the three indices of flexibility. This is
discussed in more detail, next.
Normal-Conditions: Rapid random and slow periodical
fluctuations are common, and do not take the system out of its
secure state. Generation capacity that is fast enough to accom-
modate the deviation of load from the most recent forecast (5-15
minutes ahead), in real-time, is termed ”Regulation”.
This type of reserve handles the first type of fluctuation. Reg-
ulation is necessary to maintain grid frequency. Regulation is
usually provided by small and fast online units. The ability of
these units to respond quickly to demand fluctuations is ensured
by their controllers’ design, known as automatic generation con-
trol (AGC). The deviation between the dispatch schedule and
the 5-15 minutes ahead forecast requires generation capacity of
larger magnitude, but a bigger time-frame for response, com-
pared to regulation reserve. This is known as ”Load following”
reserve, and handles the second type of fluctuations.
Load-following services, too, are provided by online but
slower units. This definition aligns with the taxonomy pro-
posed in this paper. The two types of reserve required in normal
conditions are better understood in light of the timeframes of
operations planning, demonstrated in Fig. 4.
Abnormal-Conditions: Rare abrupt perturbations of the
power balance jeopardize the grid’s security. Although these
contingencies are rare, they might have serious economic im-
plications on the system. For example, in U.K., they amounted
to 50% of total cost of ancillary services in 2013–2014 [58].
Containment of these events requires generation reserves with
fast response and large capacity for an extended period of time.
These characteristics cannot be provided by one type of device.
3146
Fig. 4. Demonstration of timeframes of operation.
Fig. 5. Reserve classification by timeframe and different ISOs.
Therefore, navigating the system through a major contingency
is a multi-stage process. As soon as slow units become avail-
able, part of the generation duty is shifted gradually from the
fast units to the slow units. Therefore, the time-intervals defin-
ing each type of reserve in [22], [28], [59] are overlapping. This
can also be observed in Fig. 5 around the 5-minutes checkpoint.
Regardless of this overlap, the key aspect in this classification is
the starting time of a unit’s response, which is the time-period
before a unit can start contributing to the system’s collective
efforts to contain a major contingency. The response-time crite-
rion is not equal to the ramping rate limit.
Response time divides contingency reserve into classes: spin-
ning, and non-spinning reserve. Spinning reserve breaks further
down to primary and secondary reserve, while non-spinning re-
serve breaks down to tertiary reserve and slow reserve. Different
classifications of contingency reserve are presented in [28], [53].
Different time milestones are used by different ISOs to define
reserve requirements, illustrated in Fig. 5. The data in the figure
details the taxonomies of [9], [28], [60], [61]. The classification
by Ma et al. [9] is for reserve in a DR program.
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 34, NO. 4, JULY 2019
The proposed classification of reserve is illustrated in Fig. 6.
Distinction is held between services deployed during normal
conditions and services deployed during contingencies.
A. Power System Operation & Planning
The security of the power system can be taken into account
during operational planning in two ways [62]. The mathematical
formulation of both methods is presented in Table I.
r Explicitly: In the first approach, each contingency is ex-
plicitly represented in the optimization model, known as
the security-constrained optimal power flow (SCOPF).
This approach guarantees the availability and deliverability
of reserve during all time periods of the plan. The repre-
sentation of contingency scenarios can be implemented in
a deterministic or stochastic way. In the deterministic rep-
resentation, the system must remain stable for all plausible
contingencies without load shedding, therefore, determin-
istic methods are also referred to as robust methods. De-
terministic methods provide conservative solutions, leav-
ing little space for optimization and incurring high energy
costs [63], or renders the problem infeasible. On the other
hand, stochastic optimization weighs the cost of each con-
tingency scenario by the probability of its occurrence.
Stochastic models tolerate the possibility of load shed-
ding below a certain level [15], and replace the constraint
on maximum load-shedding with penalties incorporated in
the objective function [63]. This also improves the conver-
gence speed of optimization algorithms. However, tuning
the penalty function is a challenging task, since improper
choice of the penalty weight may result in unnecessary
load shedding [63]. It is fair to assume that the cost of
load shedding is higher than the cost of scheduling more
reserve, within certain limits.
r Implicitly: In the second approach, the operation schedule
for each time period is planned for one scenario only rep-
resenting normal-conditions or pre-contingency scenario.
Security against contingencies is incorporated by adding a
reserve constraint for each time period [25], commonly re-
ferred to as the reserve requirement. This constraint is satis-
fied by the sum of reserves provided by each unit. Reserve
requirements are decided ex-ante, beforehand, and for each
time-period separately [51], [64]. This secure planning ap-
proach is widely used due to its lower complexity, and
computational efficiency.
This approach ignores transmission bottlenecks in the provi-
sion of reserves. [37], [65] demonstrate scenarios where trans-
mission system congestion limit large wind installations. The
choice of reserve requirement size has been a subject of numer-
ous studies, and will be discussed next.
B. Reserve Requirement
The high cost of reserves [15] impose the need for their op-
timization. In the past, power system operators defined reserve
requirements heuristically [53], [66]. e.g., the available reserve
should be sufficient to replace the largest online generation unit
[63] whose outage is considered the worst (N-1) contingency
MOHANDES et al.: REVIEW OF POWER SYSTEM FLEXIBILITY WITH HIGH PENETRATION OF RENEWABLES 3147

Fig. 6. Map of short-term reserve & flexibility.

TABLE I
FORMULATION OF THE SECURE SYSTEM OPERATION PROBLEM

Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
3148
Fig. 7. Optimum reserve size as a function of actual demand.
scenario. Another criterion was to set reserve as a fixed per-
centage of demand [67]. An optimization formulation by [35]
constrained the spinning reserve to comprise more than half
of total reserve. The regulation reserve is also constrained to
a minimum of 2% of total load size. Reserve requirements of
different ISOs are presented in [60].
The amount of reserve available in the system determines
the probability of load shedding, also known as loss of load
probability (LOLP). LOLP is used to calculate the statisti-
cal expected amount of load shedding, denoted as expected
energy not served (EENS). The optimum reserve require-
ment can be determined by weighing its cost against the
cost of load shedding [53], [66]. The unit-cost of load shed-
ding depends on its socioeconomic consequences, as obtained
from ISO’s surveys [17]. This is known as the value of lost
load (VOLL), [68], or the interrupted energy assessment rates
(IEAR) [51].
The same concepts of LOLP, EENS, and VOLL are used in the
stochastic SCOPF formulation of [15], [63]. In [49], the relation
between the optimal reserve requirement and transmission line
congestions, wind fluctuations, load forecast error, and VOLL
are analyzed. In fact, reserve requirements depend on system
characteristics [69].
Fig. 7 illustrates the optimum reserve requirement as a func-
tion of demand, assuming a fixed VOLL. In segment (a), online
generation units increase their actual output to meet the rising
demand. The remaining generation capacity (margin) in these
units decreases. It is not economically justified to bring new
units online to provide reserve, therefore, the optimum amount
of reserve reduces, and LOLP increases until EENS justifies
the dispatch of a new unit. This appears as a step increase in
optimum reserve size, depicted in segment (b). Finally, when
the available amount of reserve is not limited by the capacity
limits, but by the ramping constraints, the optimum amount of
reserve remains flat while demand size rises. This is depicted in
segment (c).
1) RES Penetration and Contingency-Reserve Requirements:
Large RES fluctuations that qualify as contingencies are rare
events. Therefore, RES do not increase the probability of con-
tingency events, but require more regulating reserve. This is
discussed in the next section.
Power system inertia depends on the online capacity of rotat-
ing governor-based generators in the system. Integrating more
inverter-based RES, and consequently de-commiting thermal
generators, reduces the system’s inertia and increases the risks
of system insecurity. Such contingencies are not caused by RES
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 34, NO. 4, JULY 2019
themselves, but the system’s ability to handle contingencies is
reduced when the available amount of contingency reserve is
reduced as well.
2) RES Penetration and Regulating-Reserve Requirements:
The penetration of RES reduces the net-load seen by the system
and increases the volatility by combining load and intermittent
generation.. RES outputs have the greatest toll on regulating-
reserve requirements. Spectral analysis of wind speed variations
was carried out in [16]. A power spectral density (PSD) analysis
of wind speed profiles by [70] discloses that the output of a wind
power plant can drop by 0.05pu of the plant’s rating within 1
minute, by 0.25pu of the plant’s rating within 10 minutes, and/or
0.3pu of the plant’s rating within 30 minutes. However, these
extreme events occurred 2% of the time in an 11-months period.
The reported fluctuations are more probable when a wind power
plant is operated near its rating. Geographically concentrated
wind power plants exhibit more fluctuations, too. A diverse
portfolio of RES technologies features lower overall volatility,
too [71], [72].
The effect of wind generation on regulating-reserve require-
ments directly depends on its forecasting accuracy [17]. High
forecasting accuracy is realizable only for short-term fore-
casts (30minutes - 5minutes ahead) [4]. Therefore, response
decisions are made within a very small window of opportu-
nity, bounded by the available regulation and load-following
services.
When RES are integrated in the system, uncertainties are
consolidated either by convolving the probability distribution
functions of different sources, or finding the total standard de-
viation of the combination of sources [73]. Therefore, setting
reserve requirements, with RES in the picture, becomes more
complex. One approach is to assign a fixed reserve require-
ment for all scenarios of RES output. In this case, conservative
RES forecasts will be adopted as the benchmark for the reserve
requirement calculation. Alternatively, the reserve requirement
can be defined dynamically, as a function of the system’s op-
erating condition and RES output. This approach offers more
economical and efficient use of system resources; however, this
approach is computationally intensive.
A number of heuristic rules are listed in [8], [17]. e.g., the
reserve must be large enough to cover 99.7% of the probabil-
ity distribution function (PDF) of wind variations [17]. In [74],
[75], decision trees (DTs) are applied for on-line setting of re-
serves for ensuring dynamic security of islanded power systems
with high wind penetration. In [8], a comprehensive reserve
management tool is presented. The effect of reserve on the PDF
which combines RES output, load variations and equipment
failures, is used to determine the reserve requirements, given a
predetermined level of risk tolerance.
In [36], it is pointed out that the control and physical char-
acteristics of some RES technologies support faster response in
comparison with conventional thermal units. This implies that
in this case, spinning reserve provided by RES can meet the
flexibility requirements. This scenario is only realizable at very
high penetration levels of RES, such that surplus in generation
exists. A number of studies, summarized in Table II, analyze the
effect of RES on regulating reserve requirements.
MOHANDES et al.: REVIEW OF POWER SYSTEM FLEXIBILITY WITH HIGH PENETRATION OF RENEWABLES
TABLE II
SURVEY OF IMPACT OF RES ON REGULATING RESERVE REQUIREMENTS
IV. FLEXIBILITY ECONOMICS
A. Economic Value of Reserve
The exchange of products and payments in the electricity
market takes one of three forms. The first two forms are reviewed
thoroughly in [76]:
r Auctions: Each generator submits a number of offers, each
offer is for the supply of a fixed amount of power at a cer-
tain price. The stack of offers which meet the demand at the
lowest cost is chosen. Generators with winning offers dis-
patch to supply the amounts of electricity offered, and are
remunerated at the price of the last winning offer, called
the marginal cost. This is known as ”uniform pricing”
[76]. Reserve requirements are not calculated by solving
the SCOPF problem with explicit models of plausible con-
tingencies, but are estimated to facilitate trading reserve
and flexibility in auctions. This model can not incorporate
transmission constraints.
r Integrated Model: Each generator submits a complex
bid, stating the generator’s marginal cost of production,
start-up & shutdown costs, and its minimum & maximum
generation capacity, etc. The system operators run a com-
plex optimization algorithm to find the cheapest generation
schedule that fulfills the demand requirements. The algo-
rithm can be augmented with the physical network model,
to incorporate transmission limits.
r Bilateral agreements: Generators negotiate directly with
large load entities to arrive at a mutual agreement on prices
and quantities, without involvement of the ISO.
In the integrated-model, the cost of delivering electricity
to each node separately is calculated, thus, allowing targeted
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
3149
Fig. 8. Transmission line congestion & reserve.
pricing. Difference in the electricity price within one grid arise
mainly due to network congestions. In the OPF solution, the
marginal price of energy (λ) at a certain node (n), is the La-
grange multiplier, or the dual variable of the power balance
constraint. As discussed in Section III-A, the explicit approach
for secure-operation planning adds several constraints to the
OPF model, to formulate the SCOPF model. The difference of
the marginal cost of electricity between the SCOPF solution,
and the OPF solution is always positive, and can be perceived
as the security cost.
An infinitesimal increase in the demand at node (n) in the
SCOPF problem perturbs the power-balance constraint of each
scenario. The effect of this demand increase is the sum of the
Lagrange multipliers of all the power-balance constraints for
node (n). This is perceived as the cost of the secure energy
supply at this node [52], [59]. Perturbing the size of either up
or down reserve at one bus changes the available amount of
the other reserve type (down/up). In other words, the available
amounts of different reserves are interlinked, and the Lagrange
multipliers for up reserve, and down reserve can not be perceived
independently. Therefore, charging separate prices for up and
down reserve at the same node has little meaning [52], [77].
For example, consider the system in Fig. 8. Area 1 exports an
3150 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 34, NO. 4, JULY 2019

amount of power (PA 1→A 2 ) to area 2 over two transmission of the system is the responsibility of ISOs, who enact hard-
lines. The capacity of each transmission line is: rules to avail the needed flexibility [54], [79]. The increased
PT L (i) < PA 1→A 2 (2) penetration of Distributed Energy Resources (DER) connected
at the distribution networks have two direct effects on system
I
 flexibility. Primarily, DER, mostly based on RES, displace the
PT L (i) ≥ PA 1→A 2 (3) traditional sources of flexibility, i.e., large thermal units con-
nected at transmission level and secondly, increase the flexi-
The generation capacity in area 2 is sufficient to supply the
bility requirements. The system operators need to exploit all
loads in area 2, but generators in area 1 are cheaper. Failure
possible forms of flexibility in including the one provided by
of one of the transmission lines forces the generators in area
DER in the form of distributed generation, storage and demand
1 to reduce their output, because the transmission capacity to
response. The way the market is organized to achieve DER con-
Area 1 is reduced. At the same time, the generators in area
tribution to flexibility varies among various countries and is still
2 must raise their output to support the load in their area. This
an open discussion between TSOs and DSOs. It is clear that
scenario affects significantly the cost of operation, and the prices
the employment of very large number of flexibility sources in
of electricity. This scenario also features a large gap in the
the central market model faces many practical difficulties and
electricity price between the two areas. Such a price gap should
besides might cause constraint violations and security issues on
alert the system operators towards an impending security crisis,
the distribution networks. Distribution system operators (DSO)
and reserve shortage [50]. Private sources of reserve lose an
running local markets to satisfy TSO requirements is another
opportunity to sell their service, and make profit. Such scenarios
model which provides promising results [80]. The participation
discourage potential investors in Area 1. At the same time,
of DSO in flexibility management and provision is discussed in
generation units in Area 2 are faced with less competition, and
[54]. In any case, close collaboration between TSOs and DSOs
grow their influence over the market.
to exploit flexibility efficiently is indispensable for the secure
Using the economic interpretation of Lagrange multipliers, a
modern power systems.
price for uncertainty is derived as the cost of immunizing the
In the implicit approach, we can distinguish two forms of
system against uncertainty at node (n) [78]. These sources of
trading [59], [76].
higher uncertainty should be motivated to upgrade their assets. r Sequential/Unbundled: Reserve services of different
Analysis of the day ahead unit commitment model discloses
types and quality are offered after clearing the energy-
that the available ramping capability of some units exceeds the
only market. This type of trading adopts the auctions trade
ramping requirements for meeting the hourly load variation.
model. A review of the US electricity market [76] points
The value of this ramping capability is greatly affected by its
out that this approach exhibits inefficiencies and problems.
deliverability, which is determined by the transmission network.
One of the drawbacks of this topology is price-reversal
In the implicit approach of Section III-A, the constraint for re-
[76], where low quality ancillary services are priced higher
serve requirement is added in a single scenario. An infinitesimal
than high-quality services.
increase in the size of reserve-requirement causes the value of r Bundled: In this approach, energy and ancillary services
the objective function to change by an amount equal to the La-
are traded simultaneously. Adopting the auctions model
grange multiplier of the reserve constraint. This can be perceived
requires prior knowledge of the reserve-requirement. This
as the marginal cost of reserve requirement. If the reserve con-
model is widely adopted in the US electricity sector [81].
straint is binding, an infiniteseimal change in the direction of the
In the explicit approach, the integrated trading model requires
binding reserve (demand increase vs. binding up-reserve con-
co-optimizing energy and ancillary services, including flexibil-
straint, demand decrease vs. binding down-reserve constraint)
ity. This approach is deemed the most efficient and robust [3],
renders the problem infeasible. This situation has to be faced by
[76]. Its main drawback is the model’s complexity and long
the start-up of a new generation unit. Bringing a new generation
convergence time.
unit online changes the whole dispatch schedule, if this unit
It should be noted that the role of market design is key in
has a non-zero minimum power output Pmin , which reduces the
making the flexibility potential accessible, and easy to use [3],
price of reserve at the expense of raising the price of energy.
[27], [54], [61], [76]. The design of a market for AS is not a
An explicit representation of the flexibility indices can pro-
trivial task [66], made more complex by the existence of an
vide operators with an explicit price for the binding flexibility
energy market. For example, although mitigation mechanisms
index. Units capable of providing this particular flexibility crite-
of RES variability are already established in various countries,
rion should be remunerated accordingly. For example, a demand
economic incentives for flexibility remain an open issue [3].
response offer to provide large but slow curtailments, has little
These following studies recommend specific modifications in
value to the grid when ramping rates are the binding flexibility
market design, and operators practices:
index on this node. Such directed pricing and remuneration is
1) More frequent dispatching of the power system units
very effective in motivating investors, and achieving the objec-
(shorter operating periods) [3], [61]: Generators are of-
tives of preserving system security.
ten paid based on their average output over the operating
period (1 hour). This practice does not reward units pro-
B. Market Design & Flexibility viding regulating reserve for a highly volatile net-load.
Section III-B demonstrates how reserve requirements are set. Shorter operating periods would also provide the opera-
Procurement of this reserve is still a subject of debate. Security tors with better RES forecasts. Furthermore, King et al.
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
MOHANDES et al.: REVIEW OF POWER SYSTEM FLEXIBILITY WITH HIGH PENETRATION OF RENEWABLES
[61] report that frequent dispatching reduces the ramping
requirement at the first place.
2) Upgrading the transmission network [27]: This eliminates
transmission congestions, reduces the cost of flexibility
and may even release some depressed flexibility potential.
European operators [54] indicated that customers might
be interested in providing funding for network expansion
projects, rather than enduring their own energy curtail-
ment due to network congestions.
3) Lowering the minimum power rating Pmin of combined-
cycle units [27]: This allows combined-cycle units to stay
online, and reduce the number of start-up and shutdown
events.
4) Administrative price setting [3]: When the actual RES
output during system operation is very far from the fore-
casted output, the system operator might need to raise the
price of flexibility to an artificially high value to motivate
providers to cover the possible flexibility deficit. Since
the size and price of flexibility is obtained as an output of
an optimization process, this practice is only justified in
cases of unexpected high RES variability.
5) Make-whole payment [3]: Compensate generators com-
mitted (online) for the sole purpose of providing flexibil-
ity. The cost of operating at this point might surpass the
electricity price, therefore, incurring financial losses. A
compensation scheme is discussed in [64].
6) Identifying distinct ancillary services, and trading them
as independent products [3], [79]. For example, trading
primary frequency response, and ramping capability. A
framework is proposed by [51] to evaluate and set the
price for ramping capability as a product in the AS market.
A market design specific for primary frequency response
is also presented in [81].
7) Profit-guarantee [3]: RES variability causes larger differ-
ence between the generation schedules planned one day
ahead or one hour ahead, and the actual generation plan.
The actual revenues might be lower than the anticipated
revenue, because of their commitment to providing down-
reserve for example.
C. Implementation Obstacles
Concerns exist about the way flexibility is incorporated as a
market product, i.e., which actor is responsible for management
of DER, what policies should regulate and govern this interac-
tion, the efficiency and reliability of communication operated
independently or with distribution system operators (DSO) act-
ing as mediators [79], [82]. Tools and policies to regulate a
swarm of DER have already been proposed, albeit their effi-
cacy and practicality need to be proven in practice. The smart
grid is inherently dependent on the information and commu-
nication infrastructure, which raises serious concerns over the
cybersecurity and the security of the whole grid operation.
Further concerns are addressed in [79], for grid-
interconnection as a potential provider of flexibility. These
concerns include: special versus standard metering equipment,
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
3151
strategies to attract investors, rewards for entry and for long-term
commitment, penalties for non-compliance and premature ter-
mination of agreements, diversity versus homogeneity of the
devices providing flexibility, exclusion of certain parties or de-
vices from participation. Furthermore, flexibility providers are
likely to have a finite window of availability, and a limited num-
ber of activations within a given period of time (i.e.,: one week).
V. RENEWABLE ENERGY MANAGEMENT SYSTEM
Section III-B2 demonstrated that operating RES at their max-
imum output (determined solely by weather conditions) exposes
the system to high RES volatility. As RES penetration increases,
power system operators will be faced with a difficult choice be-
tween power system’s security & reliability on one side, and
RES integration targets on the other side. Fast thermal units
have poor fuel efficiency, and high operating cost. Reliance on
these units to provide flexibility corresponds to high GHG emis-
sions, which annihilates the goal of RES integration at the first
place. The rise of operation cost and the consequent hike in en-
ergy prices will question the RES operation philosophy. Several
factors will determine the optimum RES operation point (below
maximum). Some of these factors can be: volatility of a particu-
lar RES type, location of RES unit with respect to the grid, grid
structure, and RES response speed.
The curtailed RES output is not necessarily wasted, but can
be treated as reserve. The works of [48], [83]–[86] suggest oper-
ating wind stations below their maximum power point, such that
this spare-margin serves as virtual inertia [85], or as regulation
reserve against large forecast error. The curtailment margin can
be set to a fixed value (in MW) or as a percentage of the avail-
able wind energy [87]. A fixed margin, whose size depends on
the total reserve requirements of the system, is spared from RES
output in [84]. The study shows significant cost savings and price
reduction from a 5% curtailment in wind energy. Alternatively,
a curtailment factor is optimized online in [86], and compared
with scenarios which were optimized offline. These operation
philosophies aim to provide regulation reserve. In contrast, pri-
mary response contingency reserve is obtained from wind power
plants in [48], [87]. Due to wind volatility, the actual margin size
over a short time period is slightly lower than the planned mar-
gin size. The authors in [48] treat only 90% of the margin width,
as guaranteed reserve. The authors in [87] declare the minimum
size of this margin, over a short period, as firm reserve. The
analysis is based on ex-ante data, thus, incorporating volatility
but not uncertainty. The results of the study show that planning a
fixed margin size is more efficient. Different types of RES in an
islanded grid are operated such that they provide down reserve,
in [88].
The output range of a RES plant can be divided into three
intervals, according to the risk correlated with operating at this
interval. These intervals are: high risk, medium risk, and low
risk. In principle, as RES operate further below their maximum
output, the risk level is lower and the RES profile’s output can
be perceived as dispatchable. At medium risk levels, this power
output can not be taken for granted, however, it can be used
3152
Fig. 9. Smart RES profile concept.
to provide ancillary services such as flexibility. Furthermore,
inverter-based RES can provide other types of ancillary services,
such as dynamic VAR support, voltage regulation and power-
factor correction. This risk interval can be divided further based
on the risk level tolerable by each type of ancillary service.
Finally, operating RES near their maximum possible output
is considered high risk, and unreliable. The boundaries of the
three main categories of risk (unreliable, non-dispatchable and
virtually-dispatchble) depend on factors such as VOLL. A spare
margin of RES output for reserve is an addition to the sources
of flexibility in Section II-A.
As mentioned in Section III-B2, a diverse RES portfolio ex-
hibits lower volatility in comparison with the volatility of each
individual RES technology. At the same time, DR and ESS can
augment such a diverse RES portfolio to create a smart RES
profile with lower risk levels, as illustrated in Fig. 9. In fact, the
groundwork for this concept is laid in [71], where the optimum
mix of different RES types is found, and supplemented with
optimally sized BESS. In fact, a BESS whose energy capacity is
5% of the total RES mix power rating is sufficient to reduce fluc-
tuations enough to create a virtually dispatchable source able to
participate in the electricity market. The statistical traits of RES
output are analyzed, as well, to create an envelope of RES fore-
casts with 90% confidence interval rather than point estimates.
Such statistical analysis would yield less conservative and more
efficient RES curtailment. This concept can be extended further
in a VPP which spans a large geographic area such that scattered
RES plants are installed, and a VPP which features responsive
loads (i.e., DR) as well.
VI. CONCLUSION
This paper reviews the concepts of power system reserve and
flexibility. The power system needs different types of reserves,
to accommodate the variety of challenges in its planning and
operation. These types of reserve come from various sources
characterized by their power capacity, ramping capacity and en-
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 34, NO. 4, JULY 2019
ergy capacity. Recent work has been done on optimizing system
flexibility. Little work has been done on trading ramping capabil-
ity independently. An additional index of flexibility is proposed
in this paper, for DR units, denoted as the response-time (θ).
Two questions are posed on this regard: What is the price of
response time in a certain scenario, and which DR design and
load type are best suited for contingency reserve.
The paper enumerates the different classifications, methods
and forms of flexibility. In fact, secure operation of the power
system can either explicitly incorporate the plausible contin-
gency scenarios, or require the availability of a lumpsum of re-
serve in the system without guarantee towards deliverability. Ex-
plicit representation automatically determines the sources of this
reserve, its deliverability, and its economic value. The explicit
representation model can either be solved deterministically, or
stochastically. Alternatively, if a fixed amount of reserve is re-
quired, this reserve is obtained from the market traded jointly or
independently from energy. Consequently, the electricity market
witnesses the exchange of different competing products: energy
and ancillary services. Trading these products in order of their
importance have been shown to be inefficient and erroneous.
This approach gives rise to an accurate price for energy and
security, as biproducts of the SCOPF problem.
Focus this paper is in the impact of RES on power system
flexibility, attributed mainly to RES volatility. RES devices
replace large thermal generators and reduce the flexibility
available at the transmission level. This also reduces the sys-
tem inertia. The main impact of RES is higher requirement of
regulating reserve, specifically, load-following services. RES
increase the total generation capacity in the system, however,
induces a ramping-capability crisis.
While various methods have been proposed to optimize re-
serve requirements, the most efficient use of flexibility resources
connected at distribution level remains an open issue. This
imposes several questions on the feasibility, practicality and
business-sustainability of such solutions. Determination of the
MOHANDES et al.: REVIEW OF POWER SYSTEM FLEXIBILITY WITH HIGH PENETRATION OF RENEWABLES
actual cost of flexibility from transmission side, distribution
side, and the optimal mix is an active area of research. Modifica-
tion of market mechanisms and ISO codes & standards appears
necessary to exploit the full potential for flexibility provision.
REFERENCES
[1] T. Barker et al., “Technical Summary. In: Climate change 2007: Mitiga-
tion,” Cambridge Univ. Press, Cambridge, U.K., Tech. Rep. Contribution
of Working Group III to the Fourth Assessment Report of the Intergov-
ernmental Panel on Climate Change, 2007.
[2] R. P. Core Writing Team and L.M. (Eds.), “Climate change: 2014 synthesis
report,” IPCC, Geneva, Switzerland, Tech. Rep. Contribution of Working
Groups I, II and III to the Fifth Assessment Report, 2014.
[3] E. Ela, M. Milligan, A. Bloom, A. Botterud, A. Townsend, and T. Levin,
“Evolution of wholesale electricity market design with increasing levels of
renewable generation,” National Renewable Energy Laboratory, Golden,
CO, USA, Tech. Rep. NREL/TP-5D00-61765, Sep. 2014.
[4] E. B. Ssekulima, M. B. Anwar, A. A. Hinai, and M. S. E. Moursi, “Wind
speed and solar irradiance forecasting techniques for enhanced renewable
energy integration with the grid: A review,” IET Renewable Power Gener.,
vol. 10, no. 7, pp. 885–898, Jul. 2016.
[5] S. Persaud, B. Fox, and D. Flynn, “Modelling the impact of wind power
fluctuations on the load following capability of an isolated thermal power
system,” Wind Eng., vol. 24, pp. 399–415, Nov. 2000.
[6] S. Persaud, B. Fox, and D. Flynn, “Effects of large scale wind power on
total system variability and operation: Case study of northern ireland,”
Wind Eng., vol. 27, pp. 3–20, Nov. 2003.
[7] M. Huber, D. Dimkova, and T. Hamacher, “Integration of wind and
solar power in Europe: Assessment,” Energy, vol. 69, pp. 236–246,
2014.
[8] M. A. Matos and R. J. Bessa, “Setting the operating reserve using prob-
abilistic wind power forecasts,” IEEE Trans. Power Syst., vol. 26, no. 2,
pp. 594–603, May 2011.
[9] O. Ma et al., “Demand response for ancillary services,” IEEE Trans.
Smart Grid, vol. 4, no. 4, pp. 1988–1995, Dec. 2013.
[10] H. Holttinen, “Impact of hourly wind power variations on the system
operation in the nordic countries,” Wind Energy, vol. 8, pp. 197–218,
2005.
[11] Y. V. Makarov, C. Loutan, and J. Ma, “Operational impacts of wind
generation on california power systems,” IEEE Trans. Power Syst., vol. 24,
no. 2, pp. 1039–1050, May 2009.
[12] M. Fazeli, J. B. Ekanayake, P. M. Holland, and P. Igic, “Exploiting PV
inverters to support local voltage—A small signal stability model,” IEEE
Trans. Energy Convers., vol. 29, no. 2, pp. 453–462, Jun. 2014.
[13] M. B. Anwar, M. S. E. Moursi, and W. Xiao, “Dispatching and frequency
control strategies for marine current turbines based on doubly fed induction
generator,” IEEE Trans. Sustain. Energy, vol. 7, no. 1, pp. 262–270, Jan.
2016.
[14] Y. Wang, V. Silva, and M. Lopez-Botet-Zulueta, “Impact of high pen-
etration of variable renewable generation on frequency dynamics in the
continental Europe interconnected system,” IET Renewable Power Gener.,
vol. 10, no. 1, pp. 10–16, Jan. 2016.
[15] F. Bouffard and F. D. Galiana, “Stochastic security for operations planning
with significant wind power generation,” IEEE Trans. Power Syst., vol. 23,
no. 2, pp. 306–316, May 2008.
[16] F. Bouffard and M. Ortega-Vazquez, “The value of operational flexibility
in power systems with significant wind power generation,” in Proc. IEEE
Power Energy Soc. Gen. Meeting, Jul. 2011, pp. 1–5.
[17] M. A. Ortega-Vazquez and D. S. Kirschen, “Estimating the spinning
reserve requirements in systems with significant wind power genera-
tion penetration,” IEEE Trans. Power Syst., vol. 24, no. 1, pp. 114–124,
Feb. 2009.
[18] “Contribution to bulk system control and stability by distributed energy
resources connected at distribution network,” Power System Dynamic
Performance Committee, Portland, OR, USA, Tech. Rep. PES TR22,
Jan. 2017.
[19] E. Ela and M. O’Malley, “Studying the variability and uncertainty impacts
of variable generation at multiple timescales,” IEEE Trans. Power Syst.,
vol. 27, no. 3, pp. 1324–1333, Aug. 2012.
[20] A. Ulbig and G. andersson, “Analyzing operational flexibility of electric
power systems,” Int. J. Elect. Power Energy Syst., vol. 72, pp. 155–164,
Nov. 2015.
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
3153
[21] S. Acharya, M. S. E. Moursi, and A. Al-Hinai, “Coordinated frequency
control strategy for an islanded microgrid with demand side management
capability,” IEEE Trans. Energy Convers., vol. 33, no. 2, pp. 639–651,
Jun. 2018.
[22] D. Kirschen and G. Strback, Fundamentals of Power System Economics.
Chichester, U.K.: Wiley, 2004.
[23] A. Ulbig, “Operational flexibility in electric power systems,” Ph.D. dis-
sertation, EEH—Power Syst. Lab., Zurich, Switzerland, 2014.
[24] Y. Dvorkin, D. S. Kirschen, and M. A. Ortega-Vazquez, “Assessing flex-
ibility requirements in power systems,” IET Gener., Transmiss. Distrib.,
vol. 8, no. 11, pp. 1820–1830, Nov. 2014.
[25] E. Lannoye, D. Flynn, and M. O’Malley, “Evaluation of power system
flexibility,” IEEE Trans. Power Syst., vol. 27, no. 2, pp. 922–931, May
2012.
[26] J. Zhao, T. Zheng, and E. Litvinov, “A unified framework for defining and
measuring flexibility in power system,” IEEE Trans. Power Syst., vol. 31,
no. 1, pp. 339–347, Jan. 2016.
[27] P. Denholm, J. Novacheck, J. Jorgenson, and M. O’Connell, “Impact of
flexibility options on grid economic carrying capacity of solar and wind:
Three case studies,” National Renewable Energy Laboratory, Golden, CO,
USA, Tech. Rep. NREL/TP-6A20-66854, 2016.
[28] P. D. Lund, J. Lindgren, J. Mikkola, and J. Salpakari, “Review of energy
system flexibility measures to enable high levels of variable renewable
electricity,” Renewable Sustain. Energy Rev., vol. 45, no. 1, pp. 785–807,
2015.
[29] S. Acharya, M. S. El Moursi, A. Al-Hinai, A. S. Al-Sumaiti, and H.
Zeineldin, “A control strategy for voltage unbalance mitigation in an is-
landed microgrid considering demand side management capability,” IEEE
Trans. Smart Grid, vol. 10, no. 3, pp. 2558–2568, May 2019.
[30] J. S. Vardakas, N. Zorba, and C. V. Verikoukis, “A survey on demand
response programs in smart grids: Pricing methods and optimization al-
gorithms,” IEEE Commun. Surv. Tut., vol. 17, no. 1, pp. 152–178, First
Quater, 2015.
[31] “Benefits of demand response in electricity markets and recommendations
for achieving them,” US Department of Energy, Washington, DC, USA,
Tech. Rep., A report to the united states congress pursuant to section 1252
of the energy policy act of 2005, Feb. 2005.
[32] J. Wang, N. E. Redondo, and F. D. Galiana, “Demand-side reserve offers in
joint energy-reserve electricity markets,” IEEE Trans. Power Syst., vol. 18,
no. 4, pp. 1300–1306, Nov. 2003.
[33] W. Li et al., “Demand response management for residential smart grid:
From theory to practice,” IEEE Access, vol. 3, pp. 2431–2440, 2015, doi:
10.1109/ACCESS.2015.2503379.
[34] “Electricity energy storage technology options: A white paper primer on
applications, costs, and benefits,” Electric Power Research Institute, Palo
Alto, CA, USA, Dec. 2010.
[35] N. Li, C. Uçkun, E. M. Constantinescu, J. R. Birge, K. W. Hedman, and
A. Botterud, “Flexible operation of batteries in power system schedul-
ing with renewable energy,” IEEE Trans. Sustain. Energy, vol. 7, no. 2,
pp. 685–696, Apr. 2016.
[36] J. Kabouris and F. D. Kanellos, “Impacts of large-scale wind penetration on
designing and operation of electric power systems,” IEEE Trans. Sustain.
Energy, vol. 1, no. 2, pp. 107–114, Jul. 2010.
[37] K. R. W. Bell, D. P. Nedic, and L. A. S. S. Martin, “The need for intercon-
nection reserve in a system with wind generation,” IEEE Trans. Sustain.
Energy, vol. 3, no. 4, pp. 703–712, Oct. 2012.
[38] Y. Guo, S. Bose, and L. Tong, “On robust tie-line scheduling in multi-area
power systems,” IEEE Trans. Power Syst., vol. 33, no. 4, pp. 4144–4154,
Jul. 2018.
[39] N. Troy, D. Flynn, and M. OMalley, “Multi-mode operation of combined-
cycle gas turbines with increasing wind penetration,” IEEE Trans. Power
Syst., vol. 27, no. 1, pp. 484–492, Feb. 2012.
[40] M. Hermans, K. Bruninx, and E. Delarue, “Impact of CCGT start-up
flexibility and cycling costs toward renewables integration,” IEEE Trans.
Sustain. Energy, vol. 9, no. 3, pp. 1468–1476, Jul. 2018.
[41] C. Shen, C. Lin, L. Chang-Chien, C. Wu, and C. Lu, “Assessment
of cycling costs of combined cycle units for frequency regulation,”
IET Gener., Transmiss. Distrib., vol. 12, no. 14, pp. 3536–3541, Aug.
2018.
[42] “Challenges of electricity storage technologies,” American Physical So-
ciety, College Park, MD, USA, Tech. Rep. 2007.
[43] M. Starke, N. Alkadi, and O. Ma, “Assessment of industrial load for de-
mand response across U.S. regions of the western interconnect,’ Oak Ridge
National Lab., Oak Ridge, TN, USA, Tech. Rep. ORNL/TM-2013/407,
2013.
3154
[44] B. Jiang, A. M. Farid, and K. Youcef-Toumi, “Demand side manage-
ment in a day-ahead wholesale market: A comparison of industrial
& social welfare approaches,” Appl. Energy, vol. 156, pp. 642–654,
2015.
[45] Z. Wang and J. Wang, “Review on implementation and assessment of
conservation voltage reduction,” IEEE Trans. Power Syst., vol. 29, no. 3,
pp. 1306–1315, May 2014.
[46] A. Ulbig, and Göran andersson, “Chapter 18: Role of power system
flexibility,” in Renewable Energy Integration: Practical Management of
Variability and Uncertainty and Flexibility in Power Grids, L. E. Jones,
Ed. San Francisco, CA, USA: Academic, 2014, pp. 227–238.
[47] E. Lannoye, D. Flynn, and M. O’Malley, “Power system flexibility as-
sessment — State of the art,” in Proc. IEEE Power Energy Society Gen.
Meeting, Jul. 2012, pp. 1–6.
[48] X. Tang, B. Fox, and K. Li, “Reserve from wind power potential in system
economic loading,” IET Renewable Power Gener., vol. 8, no. 5, pp. 558–
568, Jul. 2014.
[49] G. Liu and K. Tomsovic, “Quantifying spinning reserve in systems with
significant wind power penetration,” IEEE Trans. Power Syst., vol. 27,
no. 4, pp. 2385–2393, Nov. 2012.
[50] E. Lannoye, D. Flynn, and M. O’Malley, “Transmission, variable gen-
eration, and power system flexibility,” IEEE Trans. Power Syst., vol. 30,
no. 1, pp. 57–66, Jan. 2015.
[51] H. Ye and Z. Li, “Deliverable robust ramping products in real-time mar-
kets,” IEEE Trans. Power Syst., vol. 33, no. 1, pp. 5–18, Jan. 2018.
[52] J. M. Arroyo and F. D. Galiana, “Energy and reserve pricing in security
and network-constrained electricity markets,” IEEE Trans. Power Syst.,
vol. 20, no. 2, pp. 634–643, May 2005.
[53] H. Holttinen et al., “Methodologies to determine operating reserves due
to increased wind power,” IEEE Trans. Sustain. Energy, vol. 3, no. 4,
pp. 713–723, Oct. 2012.
[54] D. S. W. Group, “Flexibility use at distribution level, a CEER conclusions
paper,” Council Eur. Energy Regulators (CEERS), Brussels, Belgium,
Tech. Rep. C18-DS-42-04, Jul. 2018.
[55] S. Veda et al., “Evaluating the impact of the 2017 solar eclipse on U.S.
western interconnection operations,” 2018.
[56] F. Qiu, Z. Li, and J. Wang, “A data-driven approach to improve wind
dispatchability,” IEEE Trans. Power Syst., vol. 32, no. 1, pp. 421–429,
Jan. 2017.
[57] J. Zhao, T. Zheng, and E. Litvinov, “Variable resource dispatch through
do-not-exceed limit,” IEEE Trans. Power Syst., vol. 30, no. 2, pp. 820–828,
Mar. 2015.
[58] D. Howes and E. Lewis, “Electricity balancing services: Briefing for the
house of commons energy and climate change select committee,” National
Audit Office Study Team, Victoria, U.K, Tech. Rep. DP Ref: 10434-001,
May 2014.
[59] F. D. Galiana, F. Bouffard, J. M. Arroyo, and J. F. Restrepo, “Schedul-
ing and pricing of coupled energy and primary secondary, and ter-
tiary reserves,” Proc. IEEE, vol. 93, no. 11, pp. 1970–1983, Nov.
2005.
[60] Y. G. Rebours, D. S. Kirschen, M. Trotignon, and S. Rossignol, “A sur-
vey of frequency and voltage control ancillary services-Part I: Technical
features,” IEEE Trans. Power Syst., vol. 22, no. 1, pp. 350–357, Feb.
2007.
[61] J. King, B. Kirby, M. Milligan, and S. Beuning, “Flexibility reserve
reductions from an energy imbalance market with high levels of wind
energy in the western interconnection,” National Renewable Energy Lab-
oratory, Golden, CO, USA, Tech. Rep. NREL/TP-5500-52330, Oct.
2011.
[62] R. D. Zimmerman, C. E. Murillo-Sanchez, and R. J. Thomas, ‘MAT-
POWER: Steady-state operations: Planning, and analysis tools for power
systems research and education,” IEEE Trans. Power Syst., vol. 26, no. 1,
pp. 12–19, Feb. 2011.
[63] F. Bouffard, F. D. Galiana, and A. J. Conejo, “Market-clearing with
stochastic security—Part I: formulation,” IEEE Trans. Power Syst., vol. 20,
no. 4, pp. 1818–1826, Nov. 2005.
[64] M. Ortega-Vazquez and D. Kirschen, “Optimizing the spinning reserve
requirements using a cost/benefit analysis,” IEEE Trans. Power Syst.,
vol. 22, no. 1, pp. 24–33, Feb. 2007.
[65] R. Billinton and W. Wangdee, “Reliability-based transmission reinforce-
ment planning associated with large-scale wind farms,” IEEE Trans. Power
Syst., vol. 22, no. 1, pp. 34–41, Feb. 2007.
[66] J. F. Prada, “The value of reliability in power systems: Pricing operating
reserves,” Ph.D. dissertation, Massachusetts Inst. Technol. Energy Lab.,
Cambridge, MA, USA, Jul. 1999.
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 34, NO. 4, JULY 2019
[67] F. Bavafa, T. Niknam, R. Azizipanah-Abarghooee, and V. Terzija, “A new
biobjective probabilistic risk-based wind-thermal unit commitment using
heuristic techniques,” IEEE Trans. Ind. Inform., vol. 13, no. 1, pp. 115–
124, Feb. 2017.
[68] H. B. Gooi, D. P. Mendes, K. R. W. Bell, and D. S. Kirschen, “Optimal
scheduling of spinning reserve,” IEEE Trans. Power Syst., vol. 14, no. 4,
pp. 1485–1492, Nov. 1999.
[69] N. Hatziargyriou, E. Karapidakis, and D. Hatzifotis, “Dynamic behaviour
of power systems in large islands with high wind power penetration,” in
Proc. Bulk Power Syst. Dyn. Control Symp., IV Restructuring, Santorini,
Greece, Aug. 1998, pp. 23–28.
[70] P. Sørensen et al., “Power fluctuations from large wind farms,” IEEE
Trans. Power Syst., vol. 22, no. 3, pp. 958–965, Aug. 2007.
[71] M. B. Anwar, M. S. E. Moursi, and W. Xiao, “Novel power smoothing
and generation scheduling strategies for a hybrid wind and marine current
turbine system,” IEEE Trans. Power Syst., vol. 32, no. 2, pp. 1315–1326,
Mar. 2017.
[72] L. Pinto, J. Szczupak, and L. Nogueira, “An optimal less vulnerable re-
newable portfolio,” in Proc. IEEE Int. Energy Conf., Dubrovnik, Croatia,
2014, pp. 340–345.
[73] N. Menemenlis, M. Huneault, and A. Robitaille, “Computation of dy-
namic operating balancing reserve for wind power integration for the
time-horizon 1–48 hours,” IEEE Trans. Sustain. Energy, vol. 3, no. 4,
pp. 692–702, Oct. 2012.
[74] E. S. Karapidakis and N. D. Hatziargyriou, “Online preventive dynamic
security of isolated power systems using decision trees,” IEEE Trans.
Power Syst., vol. 17, no. 2, pp. 297–304, May 2002.
[75] N. Hatziargyriou et al., “Artificial intelligence techniques applied to
dynamic security assessment of isolated systems with high wind power
penetration,” CIGRE, Paris, France, Aug 27– Sept. 1, 2000.
[76] A. G. Isemonger, “The evolving design of RTO ancillary service markets,”
Energy Policy, vol. 37, no. 1, pp. 150–157, 2009.
[77] F. Bouffard, F. D. Galiana, and A. J. Conejo, “Market-clearing with
stochastic security—Part II: Case studies,” IEEE Trans. Power Syst.,
vol. 20, no. 4, pp. 1827–1835, Nov. 2005.
[78] H. Ye, Y. Ge, M. Shahidehpour, and Z. Li, “Uncertainty marginal price,
transmission reserve, and day-ahead market clearing with robust unit com-
mitment,” IEEE Trans. Power Syst., vol. 32, no. 3, pp. 1782–1795, May
2017.
[79] “Flexibility in the energy transition: A toolbox for electricity DSOs,” Euro-
pean Distribution System Operators’ Association for Smart Grids (EDSO
for smart grids), Brussels, Belgium, Tech. Rep., Feb. 2018. [Online]
https://www.edsoforsmartgrids.eu/wp-content/uploads/Flexibility-in-the-
energy-transition-A-tool-for-electricity-DSOs-2018-HD.pdf
[80] G. E. Asimakopoulou and N. D. Hatziargyriou, “Evaluation of economic
benefits of DER aggregation,” IEEE Trans. Sustain. Energy, vol. 9, no. 2,
pp. 499–510, Apr. 2018.
[81] E. Ela, V. Gevorgian, A. Tuohy, B. Kirby, M. Milligan, and M. O’Malley,
“Market designs for the primary frequency response ancillary service–
Part I: Motivation and design,” IEEE Trans. Power Syst., vol. 29, no. 1,
pp. 421–431, Jan. 2014.
[82] F. Rahimi and A. Ipakchi, “Demand response as a market resource under
the smart grid paradigm,” IEEE Trans. Smart Grid, vol. 1, no. 1, pp. 82–88,
Jun. 2010.
[83] F. Bouffard and F. D. Galiana, “Power system security and short-term
electricity market equilibrium,” in Proc. Power Syst. Conf. Expo., Atlanta,
GA, USA, Nov. 2006, pp. 90–95.
[84] E. Saiz-Marin, J. Garcia-Gonzalez, J. Barquin, and E. Lobato, “Economic
assessment of the participation of wind generation in the secondary regu-
lation market,” IEEE Trans. Power Syst., vol. 27, no. 2, pp. 866–874, May
2012.
[85] I. D. Margaris, S. A. Papathanassiou, N. D. Hatziargyriou, A. D. Hansen,
and P. Sørensen, “Frequency control in autonomous power systems with
high wind power penetration,” IEEE Trans. Sustain. Energy, vol. 3, no. 2,
pp. 189–199, Apr. 2012.
[86] M. Hedayati-Mehdiabadi, K. W. Hedman, and J. Zhang, “Reserve policy
optimization for scheduling wind energy and reserve,” IEEE Trans. Power
Syst., vol. 33, no. 1, pp. 19–31, Jan. 2018.
[87] Y. Wang, H. Bayem, M. Giralt-Devant, V. Silva, X. Guillaud, and B.
François, “Methods for assessing available wind primary power reserve,”
IEEE Trans. Sustain. Energy, vol. 6, no. 1, pp. 272–280, Jan. 2015.
[88] P. Moutis and N. D. Hatziargyriou, “Decision trees-aided active power
reduction of a virtual power plant for power system over-frequency miti-
gation,” IEEE Trans. Ind. Inform., vol. 11, no. 1, pp. 251–261, Feb. 2015.
MOHANDES et al.: REVIEW OF POWER SYSTEM FLEXIBILITY WITH HIGH PENETRATION OF RENEWABLES
Baraa Mohandes (M’09) received the B.Sc. degree
(Hons.) from Petroleum Institute—UAE, Abu Dhabi,
UAE, in 2010, and the M.Sc. degree (Hons.) in 2015.
He joined Khalifa University in 2015, where is work-
ing toward the Ph.D. degree in engineering working
on power system operation and economics and also
maintains and honors with distinction status. He has
worked with the oil and gas industry between 2010
and 2015, as a Technical Support and Projects En-
gineer. His research interests are diverse and include
feedback control systems, optimization, data science,
game-theory and power systems. He has authored in various areas. He was the
recipient of the Abu-Dhabi National Oil Company’s scholarship, as well as a
number of local and regional student awards.
Mohamed Shawky El Moursi (M’12–SM’15) re-
ceived the B.Sc. and M.Sc. degrees from Mansoura
University, Mansoura, Egypt, in 1997 and 2002, re-
spectively, and the Ph.D. degree from the University
of New Brunswick (UNB), Fredericton, NB, Canada,
in 2005, all in electrical engineering. He was a Re-
search and Teaching Assistant with the Department
of Electrical and Computer Engineering, UNB, from
2002 to 2005. He joined McGill University as a Post-
doctoral Fellow with the Power Electronics Group.
He joined Vestas Wind Systems, Arhus, Denmark, in
the Technology R&D with the Wind Power Plant Group. He was with Transco,
Abu Dhabi, UAE, as a Senior Study and Planning Engineer and seconded to
an Associate Professor position in the Faculty of Engineering, Mansoura Uni-
versity, Mansoura, Egypt and is currently on leave. He is currently a Professor
with the Electrical and Computer Engineering Department, Khalifa University
of Science and Technology, Abu Dhabi, UAE. He was a Visiting Professor with
the Massachusetts Institute of Technology, Cambridge, MA, USA. He is cur-
rently an editor for the IEEE TRANSACTIONS ON POWER DELIVERY, an editor for
the IEEE TRANSACTIONS ON POWER SYSTEMS, an Associate Editor for the IEEE
TRANSACTIONS ON POWER ELECTRONICS, a Guest Editor of the IEEE TRANSAC-
TIONS ON ENERGY CONVERSION, an Editor for the IEEE POWER ENGINEERING
LETTERS, Regional Editor for IET Renewable Power Generation, and an Asso-
ciate Editor for IET Power Electronics Journals. His research interests include
power system, power electronics, FACTS technologies, VSC-HVDC systems,
microgrid operation and control, renewable energy systems (Wind and PV) in-
tegration and interconnections.
Authorized licensed use limited to: Universidade Federal de Sergipe. Downloaded on January 25,2024 at 02:30:41 UTC from IEEE Xplore. Restrictions apply.
3155
Nikos Hatziargyriou (S’80–M’82–SM’90–F’09) is
a Faculty Member with the Electrical and Computer
Engineering School, National Technical University of
Athens, Athens, Greece, since 1984. Since 2015, he
is the Chairman of the Hellenic Distribution Network
Operator. He has authored the book “Microgrids:
Architectures and Control” and of more than 200
journal publications and 500 conference proceedings
papers. He is included in the 2016 and 2017 lists of
the top 1% most cited researchers. He was the Past
Chair of the Power System Dynamic Performance
Committee. He is the Honorary Member of CIGRE and Past Chair of CIGRE
SC C6 “Distribution Systems and Distributed Generation.” He is the Chair of
the EU Technology and Innovation Platform on Smart Networks for Energy
Transition. He has participated in more than 60 R&D projects performed for
the EU Commission, electric utilities, and manufacturers in Europe, for both
fundamental research and practical applications.
Sameh El Khatib (M’03) received the bachelor’s,
master’s, and Ph.D. degrees in electrical engineer-
ing from McGill University, Montreal, QC, Canada,
where he pursued his research in power system anal-
ysis and economics, electricity market restructuring
and planning, as well as decision-making in the en-
ergy sector. He worked as a Strategy Consultant with
Booz & Company. During his time as a Faculty
with the Engineering Systems & Management De-
partment, Masdar Institute of Science and Technol-
ogy (MIT), Abu Dhabi, UAE, his research focused
on the development and analysis of integrated engineering systems as well as
demand side management and demand response programs to meet national
energy sustainability targets. During his time with MIT as a Visiting Profes-
sor, he participated in the Multidisciplinary MIT-Harvard Research Team for
explorations in cyber international relations. His work was recognized by the
academic community in Cambridge, and featured in the Interdisciplinary Con-
sortium for Improving Critical Infrastructure Cybersecurity 3. In 2017, He was
the Founder and CEO with SmartWatt company for Energy Solutions, estab-
lishing collaborations and joint ventures with various organizations such as GE,
Etihad Airlines, ICAD, and the University of Tokyo.