Electrical Power and Energy Systems 70 (2015) 99–107

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

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

Stability-based minimization of load shedding in weakly interconnected

systems for real-time applications
M. El-Shimy ⇑
Electrical Power and Machines Department, Faculty of Engineering, Ain Shams University, 11517 Cairo, Egypt

a r t i c l e i n f o a b s t r a c t

Article history: The transient stability of power systems is highly affected by the changes in the power generation and
Received 21 February 2014 load levels. Due to the variability of the generating resources, the system stability can be provoked
Received in revised form 16 January 2015 and the system stability limits can be reached as a result of certain sudden drop in the power generation.
Accepted 31 January 2015
In these cases, the load shedding can act as an effective emergency corrective action for keeping system
stability; however, over-shedding of loads results in severe economical as well as social security prob-
lems. Therefore, minimization of the load shedding required for the restoration of the system stability
Keywords:
is one of the main objectives of this paper. Another critical issue related to successful load shedding is
Transient stability
Equal area criterion (EAC)
the fast assessment of the system stability, and the amount of the load shedding as well as the implemen-
Extended equal area criterion (EEAC) tation of the load shedding corrective action. Therefore, this paper presents a fast method of stability
Variability in power generation assessment and load shedding requirements in the weakly interconnected power system. The method
Renewable energy is an improved form of the extended equal area criterion (EEAC) where the required system equivalence
Load shedding is based on the availability of wide area monitoring (WAM) devices in modern power systems. The paper
also investigates the impact of the implementation duration of the required load shedding. The results
are verified through time domain simulations which confirm the accuracy of the presented method
and its suitability for real-time applications.
Ó 2015 Elsevier Ltd. All rights reserved.

Introduction From grid integration point of view, the integration of large

Generally, the electric power is produced through an energy
conversion process in which a primary energy source is converted
into electric power. The primary energy resources can be classified
according their capability of replenishment into two broad cate-
gories; non-renewable (or conventional) and renewable energy
resources [1,2]. The access to cheap energy and availability of ener-
gy resources for the future (or the energy security) is one of the
major challenges in the energy sector [3]. Non-renewable energy
resources are distributed in an uneven way throughout the world.
On the other hand, many types of renewable energy resources such
as wind and solar are available at all locations on the earth [4].
Therefore, renewable energy resources can contribute in enhancing
the worldwide energy security, reducing the energy threat, and cri-
sis. Due to their low pollution levels in comparison with conven-
tional sources of energy as well as their natural availability,
renewable energy resources are sustainable [5].
⇑ Mobile: +20 1005639589.
E-mail addresses: shimymb@yahoo.com, shimymb@gmail.com, Mohamed_be-
khet@eng.asu.edu.eg
amounts of renewable energy sources causes a significant change
in the classical energy mix and raises a number of major challenges
[6–9]. These challenges are mainly associated with the grid sta-
bility, reliability, security, power quality and behavior during fault
conditions. Elaboration of specific technical requirements or grid
codes for the connection of large amounts of variable renewable
energy has been constructed. The main conditions of these grid
codes show that grid-connected renewable sources should con-
tribute to the power system operation, control, reliability, and sta-
bility. In addition, these contributions should be as close as
possible to those contributions provided by conventional sources
[8,9].
Some major renewable energy resources such as wind and solar
are inherently variable and intermittent [1,3,6,7]. Recently, the
predictability of these renewable resources has been significantly
enhanced; however, their natural variability and intermittency
prevents their reliable large scale integration with power systems.
Successful power system operation requires the capability of the
system generating units to be included in operational programs
such as dispatch, unit commitment, security, and reliability
[7,10]. Given that it is not techno-economically feasible or may
be impossible to store bulk amounts of electric energy, variable

http://dx.doi.org/10.1016/j.ijepes.2015.01.034
0142-0615/Ó 2015 Elsevier Ltd. All rights reserved.
100 M. El-Shimy / Electrical Power and Energy Systems 70 (2015) 99–107
renewable energy sources may not provide the main operation
requirements of power grids.
From the security point of view, the sudden significant changes
in the power production (or power swings) from variable renew-
able can upset the stability of power systems especially when
the amounts of variable generation are large. Variability in the
power generation can be detected in conventional and renewable
energy sources. The main reason of this variability is the lack of
the required input primary resources such as non-renewable fuels
or renewable resources. In addition, some of the generation vari-
ability is attributed to the forced outage of generating units due
to failures. These failures can be internal (i.e. within the generating
units) or external (i.e. in the power networks). This paper focuses
on the real-time impact of the sudden changes in the power pro-
duction on the stability of power systems.
Currently, wind energy is one of the most prevalent adopted
renewable power sources [11,12]. Historically, there was an agree-
ment that wind turbines do not engage in the electromechanical
oscillations (EMOs) [13–15]. This was based on the facts that wind
power sources were considered as current sources. For example, a
DFIG was considered as a current source under the traditional
power factor control operation [13–15]. In either a weak grid or
a grid with large amounts of wind power, the ancillary voltage/fre-
quency control of wind farms are required for stabilization of pow-
er systems [8,16]. Consequently, recent wind farms react as
synchronous generators and engage in EMOs; however, there are
already many grid-connected uncontrollable wind farms that
made up with squirrel cage induction generators (SCIG) that are
directly connected to the grid. These kinds of wind farms may sig-
nificantly affect the stability of power systems as their power con-
trol capability is very limited.
Unlike conventional power sources, the placement renewable
power sources are mainly dependent on the availability of the
renewable resources. Usually, the feasible locations for renewable
energy projects are remotely located with respect to the power
grid [17–19]. Therefore, renewable sources are usually connected
to the grid via long transmission systems that presents weak links.
Generally, a transmission link (or tie-line) interconnecting two sys-
tems or two areas is said to be weak if its power capacity is smaller
than the capacity of the smaller system by about 15–20% [20].
Traditional generation loss contingency analysis considers
forced outage of generating units caused by internal or external
failures [10]; however, integration of variable renewable sources
adds additional power production contingencies [6,7]. These new
contingencies are due to the resource-based power production
loss. The intense variability and intermittency of a renewable
resources cause significant sudden changes in the power produc-
tion and stability harassment.
Following a large generation loss, the system frequency may
drop quickly if the remaining generation no longer matches the
load demand. Without adequate system response, loss of gen-
eration can produce extreme frequency excursions outside the
acceptable range of power plants, degradation of the load response,
overloading transmission lines, and may lead to system collapse
[21,22]. Therefore, real-time assessment of the impact of sudden
generation loss on the stability should be provided. In addition, fast
and effective emergency corrective actions should be taken to pre-
vent cascaded outage of generating units, instability and islanding
of the system, and even system blackout.
Depending on the size of the frequency deviation caused by a
disturbance, emergency control and protection schemes may be
required for maintaining the system frequency and stability.
Normal operation frequency deviations are small enough to be
controlled by the governor natural autonomous response (i.e.
primary control) and load frequency control (LFC). With large
frequency deviations that may be caused by the outage of
components or faults, emergency control and protection schemes
must be used to restore the system frequency and stability [22].
Under frequency load shedding (UFLS) is an effective and fast
emergency corrective action. UFLS shows success in preventing
probable system instability due to large disturbances [10,22]. The
main objective of an UFLS strategy is rapidly balancing the demand
with the available supply. For weakly interconnected systems, this
paper presents a simple and fast method for real-time estimation
of the necessary load shedding in generation drop situations. The
method is based on extended equal area criterion (EEAC) approach
which extended the application of the traditional equal area
criterion (EAC) [20,21] to multimachine systems [23–28].
The first step in the EEAC is reducing the system to a single
machine infinite bus (SMIB) system. To do so, the machines are
decomposed into two clusters or groups. The first group contains
the critical machine(s) while the second group contains the rest
of the machines. Each group is then aggregated into one equivalent
machine and these two machines are further aggregated into one
equivalent machine. Therefore, the multimachine system is
reduced finally to a SMIB system on which the EAC can be easily
applied. It is found that the EEAC system reduction method and
the resulting equivalent will not accurately represent the dynamics
of the original power system unless the machines comprising each
group are coherent [25,29,30].
The impact of sudden changes in the power production consid-
ering weakly interconnected systems is presented in this paper. In
addition, transient stability-based minimization of load shedding is
determined. The equivalence of power systems required for the use
of EEAC is improved by determination of its parameters from pow-
er system measurements available from WAMs. In addition, the
presented equivalency method is not only applicable for standard
conventional synchronous generators, but also covers any electric
power sources. In addition, the method can be applied at any bus
in a power system if the required measurements are available.
The EEAC approach is utilized for providing fast analysis and deci-
sion making for real-time applications. The impact of the delay in
the implementation of the load shedding corrective action is inves-
tigated. In addition, the presented method and results are verified
through time domain simulation.
Study system and modeling
Fig. 1 shows the study system and its equivalent. It consists, as
shown in Fig. 1(a), of two weakly interconnected areas or systems.
It is assumed that the generators in area 2 are highly variable while
area 1 comprises less variable power sources. In both areas sudden
drop in the power generation may be attributed to forced outage of
generators, or faults, or unavailability of the primary energy
resources. In area 2, the variability of the power sources causes
sudden changes in the output power of the generators. These sud-
den changes may cause emergency stability problems if the drop in
the power sources is intense and rapid. In such situations, load
shedding is implemented to ensure system stability by curtailing
sufficient system loads for matching the available generation with
the remaining loads and keeping the system stability.
Typically, load shedding is implemented to protect the system
against the decline of either the frequency or the voltage or both
of them. In this analysis, the load shedding is implemented to pro-
tect the system interconnection against frequency declines that
cause system instability. Due to the delay in the propagation of fre-
quency changes in weakly interconnected systems, there is a ten-
dency to localize the power adjustments following large
contingencies [22,31,32]. In addition, the localized power adjust-
ments have a significant improvement on the system stability
[31]. Therefore, drop of generation in a specific area will be pri-
 M. El-Shimy / Electrical Power and Energy Systems 70 (2015) 99–107
Fig. 1. The study system: (a) schematic diagram; (b) two-machine equivalence; (c) equivalent circuit of the two-machine system; (d) SMIB equivalence.
marily compensated by load shedding within that area unless
more necessary actions should be taken.
For implementation of the EEAC, the original system shown in
Fig. 1(a) should be subjected to two processes of electromechanical
reductions. In the first process shown in Fig. 1(b) and (c), each area
is reduced to one equivalent generator while a SMIB equivalent is
determined in the second process as shown in Fig. 1(d). The main
assumption here is that the dynamic coupling between the aggre-
gated generators is strong enough for accurate dynamic equiva-
lence. The first reduction process can be easily achieved through
the application of the standard Center of Area (COA) method
[24,25,28]; however, emerging and future power systems are
expected to contain numerous generation technologies. The main
issue related to these emerging technologies, especially those with
Full-Scale Converter (FSC) interface, is inertia of these generating
systems and their appropriate dynamic representation.
In the first reduction process, each area is reduced to a single
equivalent synchronous generator (see Fig. 1(a)–(c)). Each equiva-
lent generator is represented by three quantities or parameters;
the inertia constant (M), a transient impedance (Zeq), and a con-
stant voltage (Eeq). Although the assumption of the constant volt-
age during the transients is not valid in systems with automatic
excitation control, the use of this fixed voltage is acceptable within
the initial instants (t+) of the transient process [34]. In addition, in
the EEAC and EAC analysis, the impact of generator automatic con-
trols is usually neglected. The parameters of the equivalent gen-
erators are determined by measurements at the interface buses
using appropriate monitoring devices [32,33,35]. At the interface
buses, the required measurements are bus frequency, bus voltage,
active power, and reactive power. In this paper, required measure-
ments are obtained from system simulation.
101
The equivalent inertia at a specified bus can be estimated using
active power and frequency measurements at that bus [32,33,36–
40]. These measurements are performed under disturbed condi-
tions. Since this method treats the original subsystem as a black
box, it is general and can be applied to any mix of generating tech-
nologies. Conventional equivalency methods did not provide that
flexibility [29,30,41].
In [33,36–40] a simple approach is presented for the estimation
of the system inertia and time of the disturbances. The approach is
based on the fundamental swing equation which takes the form
[10,20,21,33]
Mdf ðtÞ=dt ¼ Pm ðtÞ  Pe ðtÞ ð1Þ
where M is the angular momentum at rated speed or the inertia
constant in seconds (M = 2 H), f is the electrical frequency in p.u, t
is the time, Pe is the electrical power in p.u, and Pm is the mechanical
power in p.u. Since the mechanical power changes are significantly
slow in comparison with the electrical changes, then the swing
equation for a small measurement interval (Dt) can be represented
with only measurable electrical quantities. Consequently the inertia
can be determined at the location of measurements using
-1
M ¼ ff_ ðt þ Þ  f_ ðt  Þg=ðP e ðt Þ  Pe ðtþ ÞÞ ð2Þ
where f_ is df/dt. Eq. (2) is valid for times immediately before and
after a disturbance (i.e. t and t+) reaching the location of the fre-
quency and active power measurements. Therefore, the equivalent
inertia of each area in the study system shown in Fig. 1(a)–(c) can
be determined at the interface buses. Based on the measurements
at t+, the equivalent impedances and voltages shown in Fig. 1(c)
are determined by the method presented in [34].
102 M. El-Shimy / Electrical Power and Energy Systems 70 (2015) 99–107

For simplification, all resistances are neglected in this analysis.

This is assumption is also usual in EAC based analysis
[9,10,17,21,34]. Based on the equivalent circuit shown in
Fig. 1(c), the power mismatches at the interface buses can be rep-
resented by
DP1 ¼ Pg1  Pd1  Pmax sin d12 ð3Þ

DP2 ¼ Pg2  Pd2 þ Pmax sin d12 ð4Þ

where Pmax is the maximum power across the interconnection; Pg1

and Pg2 are the power production from the equivalent generators
of area 1 and area 2 respectively; Pd1 and Pd2 are the equivalent
active power demands on area 1 and area 2 respectively; d12 is
the phase angle difference across the interconnection which is
equal to d1  d2. Therefore, the swing equation at each interface Fig. 2. Sudden drop in Pg2.
bus and the equation of the relative motion can be represented by
[20–31]
2 2 the Aacc equals to Apossdec as shown in Fig. 2. In this case dmax = dcr.
M 1 d d1 ðtÞ=dt ¼ DP1 ðtÞ ð5Þ Therefore,
2 2
Z dcr
M 2 d d2 ðtÞ=dt ¼ DP2 ðtÞ ð6Þ Pmax ðsin d  sin d1 Þdd ¼ 0 ð12Þ
do
2 2
d d12 ðtÞ=dt ¼ ðDP1 ðtÞ=M1 Þ  ðDP2 ðtÞ=M2 Þ ð7Þ
The integration in Eq. (12) results in the following equation which is
In addition, the equation of motion of the SMIB equivalence shown to be solved for d1.
in Fig. 1(d) can be represented by
ðdcr  do Þ sin d1 þ cos dcr  cos do ¼ 0 ð13Þ
2 2
M eq d d12 ðtÞ=dt ¼ Peq ðtÞ  Pmax sin d12 ð8Þ In Eq. (13), dcr = p  d1, do = sin1(Peqo/Pmax). The maximum permis-
where sible increase in Peq and the maximum generation drop in Pg2 can be
then found by
M eq ¼ M 1 M 2 =ðM 1 þ M 2 Þ ð9Þ
DPmax
eq ¼ P max ðsin d1  sin do Þ ð14Þ
 
Peq ¼ ðM 2 P g1  M 1 Pg2 Þ  ðM 2 Pd1  M 1 P d2 Þ =ðM1 þ M2 Þ ð10Þ
DPmax max
g2 ¼ f1 þ ðM 2 =M 1 ÞgDP eq ð15Þ
Eq. (10) can be rewritten in the following perturbation form
  It is clear from (14) and (15) that the permissible changes are highly
DPeq ¼ ðM 2 DPg1  M 1 DP g2 Þ  ðM2 DPd1  M 1 DPd2 Þ =ðM 1 þ M 2 Þ
dependent on the initial operating conditions of the system. As
ð11Þ
shown in Fig. 3, a generation drops higher than DPmax g2 results in sys-
It is clear from Eq. (11) that a drop in Pg2 increases the equivalent tem instability. This is can be mitigated by fast load shedding in the
power (Peq) by M1/(M1 + M2) while a reduction in Pd2 reduces Peq form of immediate reduction in Pd2. Recalling that the impact of
by M1/(M1 + M2). Therefore, changes of Pg2 and Pd2 have opposite sudden reduction in Pd2 is a reduction in Peq by M1DPd2/(M1 + M2).
impact on the value of Peq but its sensitivity to both of them is Therefore, if the generation drop jDP g2 j > jDPmax
g2 j, then the mini-
the same. This is also applicable to the impact of the changes in mum load shedding as shown in Fig. 3 is defined by
Pg1 and Pd1 on Peq; however, a drop in Pg1 reduces Peq and a drop
in Pd1 increases Peq. In this case the sensitivity factor is M2/ DPmin max
d2 ¼ DP g2  DP g2 ð16Þ
(M1 + M2).
Now, the maximum sudden drop in the power generation Pg2 It is known that the decrease in the mechanical driving power has
can be determined using the EAC. Consider that the system shown the same impact on the rotor angle swing as that of increasing
in Fig. 1 operates at initial conditions Pg1o, Pg2o, Pd1o, and Pd2o. As the electrical output power. In this analysis, Eq. (16) depicts that
shown in Fig. 2, this is equivalent to operation at Peqo and do. In
the following the symbol d12 will be written as d. Due to the vari-
ability of generation in area 2, a sudden drop in the generation
(DPg2 < 0) occurs. Consequently, Peq is suddenly increased by M1D
Pg2/(M1 + M2). The corresponding acceleration area (Aacc) is the area
(abca), the deceleration area (Adec) is the area (cdec), and possible
deceleration area (Apossdec) is the area (cdfec) are shown in Fig. 2.
On Fig. 2, do, d1, dmax, dcr indicate the initial steady state operating
angle, the final steady state operating angle, the maximum angle
of oscillations, and the critical transient angle respectively. It is
clear from the figure that dcr and d1 are dependent variables such
that dcr = p  d1. The stability of the system is governed by the ratio
of the possible deceleration area and the acceleration area. The
system is stable if this ratio is higher than one while it is unstable
if the ratio is less than one. A unity ratio indicates critical stability
conditions. The maximum or the permissible sudden drop in Pg2
can be then determined by determining the value of Peq1 at which Fig. 3. Transient stability regain by load shedding.
 M. El-Shimy / Electrical Power and Energy Systems 70 (2015) 99–107 103

Fig. 4. The study system and its equivalence: (a) detailed system; (b) first reduction process; (c) second reduction process (SMIB).

Table 1
Equivalent inertia (M) in sec, p.u transient reactances, and p.u transient emf (E0 ); all values are on 100 MVA base.

Area 1 equivalence Area 2 equivalence SMIB equivalence

Inertia Reactance E0 Inertia Reactance E0 Inertia Reactance
Scenario 1 (no wind) 29.11 0.0236 1.047 53.15 0.0253 1.062 18.81 0.0122
Scenario 2 (with wind) 29.15 0.0237 1.034 47.61 0.0243 1.050 18.08 0.0120

the change electric power demand shows that the same impact as nario becomes 550 MW in the second scenario. The wind farm is
the changes in the mechanical power. formed of 100 identical squirrel-cage induction-generator (SCIG)
based wind turbine generators (WTGs) each of 2 MVA rating. The
parameters of the wind turbines are available at [42–44]. The
Case study
Table 2
The IEEE two-area weakly interconnected system [21,34] Study cases for verification.
shown in Fig. 4(a) is used to represent the study system shown
Scenario 1 (no wind) Scenario 1 (with wind)
in Fig. 1(a).
The parameters of the generators and network are taken from Case 1 5.0 MW drop in the 4.0 MW drop in the
generation of area 2 generation of area 2
[34] while the loads, generator active power settings, and reactive
Case 2 5.6 MW drop in the 4.9 MW drop in the
power compensation values are taken from [21]. All loads are rep- generation of area 2 generation of area 2
resented by a constant impedance model. With these values, area 1 Case 3 6.0 MW drop in the 6.0 MW drop in the
is exporting 400 MVA to area 2. Two structural scenarios are con- generation of area 2 generation of area 2
sidered. In the first scenario, no wind power is included in the sys- Case 4 6.0 MW drop in the 6.0 MW drop in the
generation of area 2 and generation of area 2 and
tem generation mix. In the second scenario, 150 MW of the 0.4 p.u simultaneous load 1.1 p.u simultaneous load
conventional power generated in area 2 is replaced by wind power. shedding shedding
Therefore, the 700 MW power generations from G4 in the first sce-
104 M. El-Shimy / Electrical Power and Energy Systems 70 (2015) 99–107

not dependent on the nature of the power sources. Therefore, it has

the same values in both study scenarios. Eq. (13) is solved for the
two scenarios. The value of d1 is 64.5° in the first scenario while
its value is 64.7° in the second scenario. Consequently, DP max
eq which
is represented by Eq. (14) is found to be 6.32 p.u in the first scenar-
io and 6.2 p.u in the second scenario. This clarifies the stability
reduction caused by the presence of the wind power in the system.
Using Eq. (15), the DPmaxg2 is found to be approximately 5.6 p.u in
the first scenario while its value is 4.9 p.u in the second scenario.
If the sudden drops in the generation are higher than these values,
then Eq. (16) is used to determine the minimum amount of to be
dropped. Eqs. (17) and (18) represent the p.u load shedding
requirements for scenario 1 and scenario 2 respectively.

DPmin
d2 ¼ DP g2 þ 5:6 ð17Þ

DPmin
d2 ¼ DP g2 þ 4:9 ð18Þ

The verification of the estimated maximum drop in power gen-

eration and minimum load shedding requirements is performed
through time domain simulation of the two machine equivalent
system (shown in Figs. 1(b) and 4(b)). The perturbation cases shown
in Table 2 are considered for each scenario. Case 2 in both scenarios
represents the critical drop in the generation in area 2 while case 3
present situation where the drop of generation is higher than the
critical values. Case 4 presents mitigation by the minimum amount
of load shedding. The dynamic performances of the system are
shown Figs. 5 and 6 for scenario 1 and scenario 2 respectively. These

Fig. 5. Dynamic performance associated with sudden changes in the generation and
load – Scenario 1; (a) cases 1, 2, and 4; (b) case 3.

equivalents of the system (shown in Fig. 4(b) and (c)) are deter-
mined by the presented equivalency method. Evaluation of the
equivalency results is performed by the comparing the transient
performance of the detailed system and its equivalent SMIB sys-
tem. The following section presents these results as well as the
use of EAC for minimum load shedding estimation.

Results and discussion

Considering the two stated structural scenarios, the equivalent

inertia and transient reactances of the study system are deter-
mined using the presented method. The disturbance is a temporary
3-cycle three-phase fault started at t = 1 s. The fault is applied on
bus 8. Table 1 shows the determined inertia values.
It is depicted from Table 1 that the presence of the wind power
reduces the equivalent inertia of area 2 by about 10.42% while
the inertia of area 1 is insignificantly increased. These changes
are attributed to the low inertia of the WTGs in comparison with
conventional synchronous generators and also to the interaction
of these generation technologies with the transient disturbances.
The SMIB inertia is reduced only by 3.88%. This is because of
the impact combined inertia of the two areas.
The equivalent SMIB system is used to estimate the impact of
generation drop as well as the minimum required load shedding.
The analysis is based on the model presented by Eqs. (3)–(16). Peqo Fig. 6. Dynamic performance associated with sudden changes in the generation and
is determined using Eq. (10) and its value is 4.330 p.u. This value is load – Scenario 2; (a) cases 1, 2, and 4; (b) case 3.
 M. El-Shimy / Electrical Power and Energy Systems 70 (2015) 99–107 105

Fig. 7. Impact of load shedding delay; (a) simultaneous load shedding; (b) delayed load shedding.

investigated through the EAC and time domain dynamic simula-

tion. In comparison with Fig. 3, the simultaneous load shedding
with the values presented by Eq. (16) will not cause changes in
the acceleration area (area abca) shown in Fig. 7(a). Therefore,
the system will be stable after the perturbation in the power gen-
eration and load. On the contrary, the delayed load shedding, as
shown in Fig. 7(b), causes an increase in the acceleration area by
the amount bfghb. Therefore, the minimum amounts of load shed-
ding determined by Eq. (16) will not effectively restore the system
stability. Therefore, the P max
eq (Eq. (14) and Fig. 7) must be reduced
to P 0max
eq (shown in Fig. 7); Pmax
eq is reduced by an amount that

Fig. 8. Impact of load shedding delay on the minimum required load shedding
considering both scenarios – EAC approach.

figures show the rotor angle swing of area 2 relative to area 1. The
time-domain simulations are performed using PSAT 1.8 [45] on
Matlab 2012 [46]. These figures ensure the accuracy of the present-
ed method in predicting the system transient response and
minimization of the load shedding requirements.
From a practical point of view, the simultaneous drop in the
power generation and the required load shedding is technically
impossible. Therefore, the impact of delayed load shedding on Fig. 9. Impact of shedding delay time on the minimum amount of load shedding –
the system stability and the validity of the presented results are time-domain simulation.
106 M. El-Shimy / Electrical Power and Energy Systems 70 (2015) 99–107
Fig. 10. Dynamic response of properly reduced loads with and without time delay.
increase the possible deceleration area by an amount jcekj such
that the increase in the acceleration area is precisely compensated
by the increase in the possible deceleration area i.e. area jcek-
j = area bfghb.
In Fig. 7, dshed denotes the angle at which the load shedding
1
takes place while dh is the angle defined by sin ðP eq1 =Pmax Þ where
Peq1 is the equivalent power due to the reduction in Pg2 (see Eq.
(10)). The load shedding delay time is nonlinearly related to the
difference between the load shedding angle (dshed) and the initial
angle (do). Fig. 7(c) and (d) show two situations defined according
to the relation between dshed and dh. These situations are dshed < dh
(Fig. 7(c)) and dshed P dh (Fig. 7(d)). The acceleration and possible
deceleration areas are clearly illustrated in the figures. It can be
easily shown that the angle d1 can be determined for both situa-
tions by solving Eq. (19) for d1. Eq. (19) is determined by equalizing
the acceleration and deceleration areas in each situation.
ðdcr  dshed Þ sin d1  cos d1  cos do þ ðdshed  do Þ sin dh ¼ 0 ð19Þ
It is clear that Eq. (19) will be reduced to Eq. (13) if simultaneous
load shedding (i.e. dshed = do) is applied. Based on the value of d1,
P0max (shown in Fig. 7) can be determined using P0max ¼ Pmax sin d1 .
eq eq
Consequently the minimum reduction in peak (denoted by DP min
eq )
is equal to Peq1  P0max
eq . As a result, the minimum load shedding
 
DPd2 can be determined using ð1 þ M2 =M1 ÞDP min
min
eq . It is depicted
from Eq. (19) that the value of d1 is dependent on the initial operat-
ing conditions of the system represented by do and the shedding
delay which is represented by (dshed  do).
Considering both scenarios and for the same DPg2 of 6.0 p.u,
Fig. 8 illustrates the impact of delayed load shedding on the mini-
mum load shedding requirements In both scenarios, the results
show that with the increase in the shedding delay, the minimum
load shedding requirements are increased. Since the same amount
of sudden reduction in Pg2 is considered for both scenarios, com-
parative analysis can be performed. With a significant amount of
wind power (scenario 2), the minimum required amounts of load
shedding are significantly high in comparison with scenario 1 (no
wind) for all shedding delay. In addition, the slop of the curve asso-
ciated with scenario 2 is high in comparison with that associated
with scenario 1. This indicates that the system with only conven-
tional generators has lower sensitivity to shedding delay in com-
parison with the system with a significant amount of wind power.
The relation between the minimum load shedding and the
shedding time delay (rather than (dshed  do)) can be determined
by solving the nonlinear mathematical model representing the sys-
tem. This is can be easily adopted in contingency analysis/correc-
tive action schemes in the system security center. Considering
scenario 1 as an example, the relation between the minimum load
shedding and the shedding time delay is determined as shown in
Fig. 9. The results shown in this figure are determined through
time-domain dynamic simulation of the two-machine equivalence
of the system considering a drop of 6.0 p.u in the power generation
of area 2. Recalling that the maximum drop of generation in this
area is found to be 5.6 p.u. Therefore, Eq. (17) show that at least
0.4 p.u of area 2 load must be immediately disconnected. With
the delayed load shedding, the results shown in Fig. 9 show that
the minimum value of the load shedding must be increased for
restoring the system stability. This is in agreement with the EAC
based analysis shown in Figs. 7 and 8; however, with the time
domain analysis, the shedding delay is represented in time rather
than angles. In addition, the application and the accuracy of the
EAC is limited to the initial stages of the dynamic processes; how-
ever, the EAC shows significant potential and accuracy in the fast
estimation of the transient stability of power systems.
The time domain responses of the two-machine equivalence
considering zero cycles (case 5) and eight cycles (case 6) load shed-
ding delay times are shown in Fig. 10. With the delayed shedding,
the minimum load shedding is increased from 0.4 p.u (no delay) to
0.48 p.u. The transient associated with the load shedding action is
clearly shown on the figure.
Conclusion
Fast assessment of the transient stability of power systems for
use in real-time emergency conditions is presented in this paper.
Restoration of the system stability due to critical amounts of gen-
eration drop is achieved by load shedding. The presented method
provides a fast and accurate procedure for estimating the minimum
amount of load shedding by the use of EEAC. The system equivalen-
cy needed for the EEAC is determined based on the available WAMs
available at modern power systems. In addition, the determination
of the system equivalents is fast enough to be implemented in real-
time. The impact of delayed load shedding is effectively modeled
and simulated by use of the EAC and time domain simulations.
The validity of the presented work is evaluated by numerous
numerical examples and time-domain simulations.
References
[1] Maczulak A. Renewable energy: sources and methods. New York, USA: Facts
On File, Inc.; 2010.
[2] Kreith F, Goswami DY, editors. Handbook of energy efficiency and renewable
energy. USA: CRC Press; 2007.
[3] Sovacool BK, Brown MA. Competing dimensions of energy security: an
international perspective. Ann Rev Environ Resour 2010;35:77–108.
[4] Surface meteorology and solar energy. NASA renewable energy resource
website. <http://eosweb.larc.nasa.gov/sse>. [accessed: 20.02.14].
[5] Difference between renewable and sustainable energy. <http://
www.solarpowernotes.com/difference-between-renewable-and-sustainable-
energy.html#>. [accesses 20.02.14, UnAsrlPnURQ].
[6] Cochran J, Bird L, Heeter J, Arent DA. Integrating variable renewable energy in
electric power markets: best practices from international experience. National
Renewable Energy Laboratory, US Department of Energy; 2012.
[7] Hand MM, Baldwin S, DeMeo E, Reilly JM, Mai T, Arent D, Porro G, Meshek M,
Sandor D, editors. Renewable electricity futures study NREL/TP-6A20-
52409. Golden, CO: National Renewable Energy Laboratory; 2012.
[8] Tsili M, Papathanassiou S. A review of grid code technical requirements for
wind farms. IET Renew Power Gener 2009;3(3):308–32.
[9] El-Shimy M. Modeling and analysis of reactive power in grid-connected
onshore and offshore DFIG-based wind farms. Wind Energy 2014;7(2):279–95.
[10] Wood AJ, Wollenberg BF. Power generation, operation, and
control. Canada: John Wiley & Sons; 2012.
[11] EL-Shimy M. Probable power production in optimally matched wind turbine
generators. Sustain Energy Technol Assess 2013;2:55–66.
[12] Arántegui RL, Corsatea T, Suomalainen K. 2012 JRC wind status report:
technology, market and economic aspects of wind energy in Europe –
JRC77895. Joint Research Centre of the European Commission; 2012.
 M. El-Shimy / Electrical Power and Energy Systems 70 (2015) 99–107
[13] Yuanzhang S, Lixin W, Guojie L, Jin L. A review on analysis and control of small
signal stability of power systems with large scale integration of wind power.
In: IEEE International conference on power system technology (POWERCON),
2010 Hangzhou; 2010. p. 1–6.
[14] Slootweg J, Kling W. The impact of large scale wind power generation on
power system oscillations. Electr Power Syst Res 2003;67(1):9–20.
[15] Domínguez-García JL, Gomis-Bellmunt O, Bianchi FD, Sumper A. Power
oscillation damping supported by wind power: a review. Renew Sustain
Energy Rev 2012;16(7):4994–5006.
[16] Jadhav H, Roy R. A comprehensive review on the grid integration of doubly fed
induction generator. Int J Electr Power Energy Syst 2013;49:8–18.
[17] Ackermann T, editor. Wind power in power systems. USA: Wiley Online
Library; 2005.
[18] El-Shimy M. Optimal site matching of wind turbine generator: case study of
the Gulf of Suez region in Egypt. Renew Energy 2010;35(8):1870–8.
[19] El-Shimy M. Viability analysis of PV power plants in Egypt. Renew Energy
2009;34(10):2187–96.
[20] Venikov VA. Transient processes in electrical power systems. Moscow: Mir
Publishers; 1977.
[21] Kundur P, Balu NJ, Lauby MG. Power system stability and control. New
York: McGraw-hill; 1994.
[22] Bevrani H. Robust power system frequency control. Springer; 2008.
[23] Xue Y, Van Custem T, Ribbens-Pavella M. Extended equal area criterion
justifications, generalizations, applications. IEEE Trans Power Syst
1989;4(1):44–52.
[24] Xue Y, Pavella M. Extended equal-area criterion: an analytical ultra-fast
method for transient stability assessment and preventive control of power
systems. Int J Electr Power Energy Syst 1989;11(2):131–49.
[25] Dong Y, Pota H. Transient stability margin prediction using equal-area
criterion. IEE Proc C (Generat, Transm Distrib) 1993;140(2):96–104.
[26] Chang H-C, Wang M-H. Another version of the extended equal area criterion
approach to transient stability analysis of the Taipower system. Electr Power
Syst Res 1992;25(2):111–20.
[27] Chiodo E, Lauria D. Transient stability evaluation of multimachine power
systems: a probabilistic approach based upon the extended equal area
criterion. IEE Proc Gener, Transm Distrib 1994;141(6):545–53.
[28] Choo YC, Kashem MA, Negnevitsky M. Transient stability assessment of a small
power system subjected to large disturbances. Australasian Universities Power
Engineering Conference (AUPEC); 2006. p. 1–5.
[29] Abuel-wafa AR, EL-Shimy M. Coherency-based electromechanical equivalents
for use in power system stability studies. Sci Bull – Facul Eng – Ain Shams Univ
2003;38(3):475–87.
[30] Abuel-wafa AR, EL-Shimy M. Treatment of large-scale power systems in
transient stability studies. Sci Bull – Facul Eng – Ain Shams Univ
2003;48(4):675–87.
107
[31] You H, Vittal V, Yang Z. Self-healing in power systems: an approach using
islanding and rate of frequency decline-based load shedding. IEEE Trans Power
Syst 2003;18(1):174–81.
[32] Tsai S-J, Zhang L, Phadke AG, Liu Y, Ingram MR, Bell SC, et al. Frequency
sensitivity and electromechanical propagation simulation study in large power
systems. IEEE Trans Circ Syst I: Regul Papers 2007;54(8):1819–28.
[33] Wall P, González-Longatt F, Terzija V. Demonstration of an inertia constant
estimation method through simulation. In: 45th International universities
power engineering conference (UPEC), Cardiff, Wales: IEEE; 2010. p. 1–6.
[34] Chow JH, Chakrabortty A, Vanfretti L, Arcak M. Estimation of radial power
system transfer path dynamic parameters using synchronized phasor data.
IEEE Trans Power Syst 2008;23(2):564–71.
[35] Ota Y, Hashiguchi T, Ukai H, Sonoda M, Miwa Y, Takeuchi A. Monitoring of
interconnected power system parameters using PMU based
WAMS. Lausanne: IEEE Power Tech; 2007. p. 1718–22.
[36] Inoue T, Taniguchi H, Ikeguchi Y, Yoshida K. Estimation of power system
inertia constant and capacity of spinning-reserve support generators using
measured frequency transients. IEEE Trans Power Syst 1997;12(1):136–43.
[37] Chassin DP, Huang Z, Donnelly MK, Hassler C, Ramirez E, Ray C. Estimation of
WECC system inertia using observed frequency transients. IEEE Trans Power
Syst 2005;20(2):1190–2.
[38] Wall P, Gonzalez-Longatt F, Terzija V. Estimation of generator inertia available
during a disturbance. In: Power and energy society general meeting. San
Diego, CA: IEEE; 2012. p. 1–8.
[39] Gonzalez-Longatt FM. Effects of the synthetic inertia from wind power on the
total system inertia: simulation study. In: 2nd International symposium on
environment friendly energies and applications (EFEA). Newcastle: IEEE; 2012.
p. 389–95.
[40] Zhang Y, Bank J, Wan YH, Muljadi E, Corbus D. Synchrophasor measurement-
based wind plant inertia estimation. In: IEEE Green technologies conference.
Denver: IEEE; 2013. p. 494–9.
[41] Milano F, Srivastava K. Dynamic REI equivalents for short circuit and transient
stability analyses. Electr Power Syst Res 2009;79(6):878–87.
[42] El-Shimy M, Badr M, Rassem O. Impact of large scale wind power on power
system stability. In: Power system conference, 12th international middle-east
power conference (MEPCON). Aswan, Egypt: IEEE; 2008. p. 630–6.
[43] El-Shimy M, Ghaly N, Abdelhamed M. Parametric study for stability analysis of
grid-connected wind energy conversion technologies. In: 15th International
middle-east power conference (MEPCON). Alex., Egypt: IEEE; 2010. p. 1–8.
[44] EL-shimy M, Ghaly B. Grid-connected wind energy conversion systems:
transient response. In: Encyclopedia of energy engineering and Technology.
2nd ed., vol. IV. CRC Press, Taylor & Francis Group; 2014. p. 2162–83.
[45] PSAT 2.18, Jan. 6; 2013. <http://faraday1.ucd.ie/psat.html>. [accessed:
20.02.14].
[46] MATLAB; 2012. <http://www.mathworks.com/>. [accessed: 20.02.14].