960 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 4, NO.

4, OCTOBER 2013

Small Signal Stability Assessment of Power

Systems With Increased Penetration of
Photovoltaic Generation:
A Case Study
Sara Eftekharnejad, Member, IEEE, Vijay Vittal, Fellow, IEEE, Gerald Thomas Heydt, Life Fellow, IEEE,
Brian Keel, Senior Member, IEEE, and Jeffrey Loehr

Abstract—The present paper investigates the effect of high pen- The distinctive characteristics of the PV systems has resulted
etration of photovoltaic (PV) systems on the small signal stability in new concerns regarding the reliability and security of the
of a large power system. Reduced system inertia and altered power power systems with high levels of PV generation. Reduced
flow patterns as a result of the addition of the utility scale and res-
idential rooftop PVs that replace a portion of conventional gen- system inertia and lack of reactive power support for residential
eration resources, may lead to decreased damping of the critical rooftop units are the main contributors that could impact the
modes of the system. To identify the critical modes of the system response of power systems with increased PV penetration
and the effect of the high PV penetration on these modes, eigen- levels. Displacing part of the conventional generators with
value analysis is carried out on the aforementioned system under more PV resources translates to less rotating mass within the
various PV penetration levels. To substantiate the results observed
from the small signal analysis, transient analysis is carried out on system, hence decreasing overall system inertia. Although this
the system under various PV penetration levels. The simulation re- phenomenon is not much of a concern in terms of the steady
sults effectively identify the impact of high PV penetration on small state behavior of the power systems, it will be a major source
signal stability of the studied system. of concern during system transients. The ability of the system
Index Terms—Distributed power generation, photovoltaic (PV) to adjust the system operating conditions in response to various
generation, power transmission, sensitivity, small signal stability, disturbances, and performing the required tasks in a timely
transient stability. manner is essential for a reliable system operation.
Presently practiced standards and recommended practices,
e.g., IEEE 1547 [2] and UL 1741 [3], suggest that residential PV
I. INTRODUCTION
inverters should not actively regulate the voltage at the point of
common coupling (PCC). Consequently, with the exception of

T HE present renewable portfolio standards (RPS) which

have been developed to address the need for clean, sus-
tainable, and renewable energy, are expected to structurally alter
the utility scale units, PV systems are mainly utilized as sources
of active power and no reactive power is generated by these re-
sources. Therefore, supplying adequate reactive power support
the characteristics of the power systems [1]. The RPS mandates
to the customers is a problem that needs to be addressed as the
have required some of the states in the U.S. to increase their
PV penetration level increases.
level of renewable generation to up to 20% by the year 2020.
While addition of the PV resources are certain to occur in
Hence, large-scale power generation from photovoltaic (PV) re-
the future, the exact effect of these systems, particularly on the
sources is no longer a vision but a foreseeable reality. As more
transmission systems, are not adequately studied. The effects of
PV systems are being installed in existing grids, more conven-
high PV penetration on distribution systems have been studied
tional generation units may be displaced due to their high oper-
in [4]–[7]. As these studies suggest, high PV penetration can
ating costs and other factors such as aging. With these unprece-
affect the voltage profile depending on the loading conditions
dented rapid changes to the power grids, special measures may
and the amount of PV penetration. Study results of these works
be required to accommodate the needs of the future power sys-
suggest that presently available voltage regulation equipment is
tems.
not capable of mitigating the adverse effects of PV system tran-
sients. These works have identified some problems associated
Manuscript received September 27, 2012; revised January 12, 2013 and with the PV systems that directly affect the distribution systems.
March 26, 2013; accepted April 18, 2013. Date of publication May 03, 2013;
date of current version September 16, 2013. This work was supported by the With increased PV penetration, the effect of these systems
grants from the Salt River Project (SRP). on the transmission systems can no longer be neglected. Re-
S. Eftekharnejad, V. Vittal, and G. T. Heydt are with the Department of Elec- searchers [8]–[10] have investigated the transient behavior of
trical, Computer and Energy Engineering, Arizona State University, Tempe, AZ
85287 USA (e-mail: [email protected]; [email protected]; [email protected]). the transmission system in response to various disturbances re-
B. Keel and J. Loehr are with the Salt River Project, Phoenix, AZ 85281 USA lated to PV generation. It has been shown in [8] that depending
(e-mail: [email protected]; [email protected]). on the amount of PV solar generation and the point of inter-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. connection (POI), PV generation can have both beneficial and
Digital Object Identifier 10.1109/TSTE.2013.2259602 detrimental impacts on transient voltages. However, the studied

1949-3029 © 2013 IEEE

Authorized licensed use limited to: TULANE UNIVERSITY. Downloaded on September 16,2024 at 23:01:39 UTC from IEEE Xplore. Restrictions apply.
EFTEKHARNEJAD et al.: SMALL SIGNAL STABILITY ASSESSMENT OF POWER SYSTEMS 961

cases are not actual representations of a large interconnected power output of a generator. In a system with high PV pene-
power system. Achilles et al. [10] have conducted various case tration, some of the synchronous generators are replaced with
studies on an IEEE 39 bus system to analyze transmission PV units. The authors in [16] have proposed a general approach
system performance with high PV penetration. for integration of variable energy resources. The method sug-
The aforementioned works have analyzed the system be- gests that for every 3-MW addition of renewable generation to
havior in terms of the steady state and transient stability. Due to the system, there would be a 2-MW reduction in conventional
the reduced system inertia, small signal stability of the systems generator commitment and 1-MW reduction in their dispatch.
with high PV penetration could also be affected. In [11], small While the choice of the cited “1/3–2/3 rule” could be quite arbi-
signal stability of a two area power system under increased PV trary, note that the overall system inertia is decreased, which can
generation is investigated to observe both the detrimental as lead to potential small signal stability problems. Consequently,
well as the beneficial impacts of PV generation on the interarea with displacing/rescheduling of conventional units as a result of
modes of the system. the addition of PV generation, it is advantageous to determine if
The authors in [13] study the voltage stability as well as angle a particular generator’s inertia has significant impact on a partic-
stability of an IEEE 14 bus system while equipped with utility ular inertial oscillation mode. This could be determined by per-
scale PV systems. The effect of PV units on inter-area modes forming sensitivity analysis with respect to generator inertia and
of oscillation is also investigated in this work. In another work performing decommitment/rescheduling using the sensitivity to
[12], the effect of PV systems on a power system represented as inertia as a constraint.
a single machine and an infinite bus is investigated. This work
derives a mathematical model based on the – characteristics B. Proposed Approach
of the PV systems which is suited for the distribution system Small signal stability analysis is based on deriving a linear
rather than a large-scale transmission system. It is also not ap- model of the nonlinear system model around a certain operating
parent that considering the – characteristics significantly im- condition. The stability of the linearized system is determined
pacts electromechanical oscillations. The present paper investi- by the eigenvalues of the state matrix while the participa-
gates the small signal stability of a study system which is a por- tion of each system state in a specific eigenvalue is determined
tion of the Western Electricity Coordinating Council (WECC) by the right eigenvector and the left eigenvector . The th
under various PV penetration levels. The approach developed eigenvalue of the system matrix and its corresponding eigen-
is proposed for any large realistic system without loss of gen- vectors are defined as
erality even though the conclusions derived depend on the spe-
cific characteristics of the system analyzed. A comprehensive (1)
eigenvalue analysis of the system is carried out to identify the (2)
critical modes of the system. The results are compared to assess
the stability of the system with increased PV penetration levels. For a complex eigenvalue that corresponds to an oscillatory
The results derived from the small signal analysis of the system mode of the system, the frequency of the oscillation in Hz
are further expanded by conducting transient simulations on the and the damping ratio are expressed as
system. To perform the studies presented in this work, software
packages such as DSATools and PSLF have been utilized. (3)
This paper is organized as follows. Section II describes the (4)
basic premises of small signal stability analysis and the motiva-
tions for this study. In Section III, a description of the studied (5)
system is presented followed by the modeling aspects of the PV
systems. The analysis of small signal stability is presented in The damping ratio shown in (5), which is related to the real part
Section IV. Conclusions derived from these analyses are pre- of the eigenvalues, determines the rate at which the amplitude
sented in Section V. of the oscillations decreases [15]. Hence, a positive real part
corresponds to oscillations with increased amplitudes while a
II. SMALL SIGNAL STABILITY ANALYSIS negative real part corresponds to a damped oscillation. As the
complex pole moves towards the right half plane (RHP), the
damping of the system worsens.
A. Motivation for the Study
The objective is to investigate whether the critical modes
Small signal stability is defined as the ability of the system of the system are detrimentally or beneficially affected by in-
to maintain synchronism when it is subjected to small distur- creased PV penetration. The following steps are taken in the
bances [14]. In this context, there are two types of instability proposed approach:
that can occur: the steady state rotor angle increase due to lack 1) Identify the most critical, i.e., poorly damped, modes of the
of synchronizing torque and the increasing rotor oscillations due system by performing eigenvalue analysis on the base case
to insufficient damping torque are the two major types of small with no PV generation.
signal instability [15]. Generator–turbine inertia generally plays 2) Perform eigenvalue analysis for the cases after introducing
a key role in providing synchronizing capability to the syn- various levels of residential rooftop and utility scale PVs.
chronous generators whenever a disturbance results in a mis- 3) Compare the results of the eigenvalue analysis under dif-
match between the mechanical power input and the electrical ferent PV penetration levels to investigate the impact of
Authorized licensed use limited to: TULANE UNIVERSITY. Downloaded on September 16,2024 at 23:01:39 UTC from IEEE Xplore. Restrictions apply.
962 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 4, NO. 4, OCTOBER 2013

TABLE I TABLE II
SUMMARY OF THE STUDIED AREA PV SYSTEM DESCRIPTION AND CAPABILITIES

TABLE III
SUMMARY OF PV PENETRATION LEVELS

high PV penetration on small signal stability of the system

under study.
4) Perform eigenvalue sensitivity analysis with respect to
the displaced generators’ inertia to validate the results
achieved from the eigenvalue analysis.
5) Analyze the transient stability performance of the system negative loads. Additional details of these units can be found
and examine whether the identified critical modes can be in [10]. Utility scale PVs are fixed at 600 MW of active power
excited by a large disturbance and substantiate the results generation while rooftop PVs are varied under various PV gen-
obtained by eigenvalue analysis. eration levels. The PV penetration definition used is based on
the available generation in the base case and is calculated as
III. ATTRIBUTES OF THE TEST BED SYSTEM
(6)
A. Description of the Test Bed System
Small signal stability analysis as well as transient analysis is
Alternate definitions of PV penetration levels such as the def-
carried out on a test bed that represents the entire Western Elec-
inition based on the system peak load [9] or by the amount of
tricity Coordinating Council (WECC) with transmission voltage
energy served (e.g., in one year) can also be found in the liter-
levels ranging from 34.5 to 500 kV. The PV systems are added
ature. The equivalent active power generation for each gener-
to an area of the WECC system that has a high potential for PV
ation scenario is illustrated in Table III. Calculation of the PV
installations and has a relatively large amount of power export to
penetration percentages is carried out based on (6). For compar-
the other areas. The synchronous generators within the WECC
ison purposes, PV penetration levels are also defined based on
system are equipped with power system stabilizers (PSS), ex-
the other methods in Table III. Note that 50% PV penetration
citation systems, as well as governors for frequency regulation.
level defined based on (6) approximately corresponds to 12.5%
Each area is equipped with a single slack bus for power flow
PV penetration by energy, which is below the RPS requirements
analysis. Table I summarizes the attributes of the studied area
for some states in the U.S. The definition of the PV penetration
which has been chosen for incorporating a high concentration of
by energy is an approximation that calculates the PV penetration
utility scale and rooftop PVs. The active load in the case studied
by energy based on (7). The capacity factor of the PV systems
is about 13 GW while the summer peak load of the studied area
is assumed to be 25% throughout this work
is close to 20 GW. Therefore, these operating conditions are
valid representations of a case with peak PV outputs and light
loads. Under these conditions, specifically during the daylight
(7)
off peak hours, the load at a particular location might be less
than the PV output generated. Hence, the excess PV generation As the level of PV penetration increases, more conventional
can contribute to the reversal of the power flow from the load generators are displaced by rooftop PV resources. However, the
side to the transmission system and disruption in system oper- critical generators in terms of the reactive power support are
ating conditions may occur. retained in service. In order to keep the generation and load
Two types of PV systems are included in the system consid- balanced, the output of these critical generators is scaled down
ered. Table II lists important characteristics of these units [10]. to accommodate the addition of the PV resources to the system,
“Rooftop” PVs are aggregated at the transmission/subtransmis- for PV penetration levels of higher than 30%.
sion voltage level of 69 kV based on actual penetration provided
using a zip code delineation across the studied area. Rooftop B. Modeling of the PV Systems
units in a given zip code are lumped at the closest 69 kV bus and Presently practiced standards such as IEEE 1547 [2] do
are modeled on the “system side” (upstream) of the inverters as not allow PV inverters to regulate the voltages at the point of
Authorized licensed use limited to: TULANE UNIVERSITY. Downloaded on September 16,2024 at 23:01:39 UTC from IEEE Xplore. Restrictions apply.
EFTEKHARNEJAD et al.: SMALL SIGNAL STABILITY ASSESSMENT OF POWER SYSTEMS 963

common coupling. This limitation has resulted in PV instal-

lations being operated at a fixed power factor. Although the
PV inverters can be operated at any fixed power factor, the
majority of those systems hold a unity power factor. Therefore,
these systems are modeled at unity power factor throughout this
work. Being operated at unity power factor, residential rooftop
PVs are modeled as constant current loads corresponding to
negative power injections. This choice of modeling is justi-
fied by the fact that rooftop units are typically located at the
distribution level. The rooftop units are also limited in their
rating compared to utility scale units and in general there does
Fig. 1. Schematic of the PSLF model used to represent utility scale PVs.
not exist any coordinated control among the inverters of units
located in the same neighborhood. As a result, the dynamics
of each individual unit is unlikely to affect system dynamic
behavior but the large penetration of rooftop units in a given
neighborhood is likely to alter the power flows at the subtran-
mission level. The model considered accurately represents the
manner in which the rooftop units alter the power flows.
Unlike the rooftop PVs, utility scale PV systems are equipped
with converters that can regulate the voltage. Hence, utility scale
PV units are not operated at unity power factor and possess the
capability to generate reactive power. As a result, a detailed
model including the representation of the inverter is required
for utility scale PV system representation for both power flow
as well as dynamic studies. Details of the power flow modeling Fig. 2. Reactive power control model in PSLF used to represent a reactive
can be found in [10] and [17]. Although utility scale PV systems power controller at utility scale PVs.
are only a small portion of the total PV generation, due to the fact
that small signal analysis of the studied system necessitates dy-
namic modeling of the PV systems, a brief description of these
dynamic models is presented. The PV model used for the small
Fig. 3. Converter model used to represent the dynamics of the PV inverter.
signal analysis is the same as the models used for the transient
analysis mainly because the DSATools utilize the same dynamic
data file for both the transient stability analysis and small signal block. The basic structure of the reactive power controller
stability analysis. is presented in Fig. 2. This controller consists of two major
To better represent the dynamic behavior of the utility scale parts: the voltage regulator and the power factor controller.
PV units, a representation of the converter model is required. Depending on the required control task, each of these control
As suggested by NERC and WECC working group on PV mod- units can be activated in PSLF by specifying a proper flag.
eling, [17], [18], a converter model suitable for electromechan- The details of the model presented in Fig. 2 can be found in
ical transients that is available both in DSATools and PSLF [19]. In this work, voltage regulation has been selected for the
can approximately represent the full converter models. In the controller options.
model of the converter considered, the dynamics are dominated Various works have investigated the models for representing
by the rapid response due to the electrical controls through the the inverter dynamics. The authors in [20] have proposed a first-
converter. However, the parameters of the aforementioned con- order transfer function to model the inverters. This model has
verter model should be modified to fully represent the utility been used in this study and is presented in Fig. 3.
scale PV resources [18]. It should be noted that due to the time
frame of the stability analyses, the dc side dynamics, which
IV. IMPACTS ON POWER SYSTEM SMALL SIGNAL STABILITY
generally exhibit slower response, are neglected in this model.
Fig. 1 presents a schematic of the WECC generic model pro- The objective of the small signal stability analysis is to
posed for representation of the converter characteristics. As seen examine the stability of the system under various PV pen-
from Fig. 1, two main components contribute to the dynamic be- etration levels. In order to locate the critical modes of the
havior of the full converter: converter model and the converter system, an eigenvalue analysis is conducted for the modes
control. The converter model represents the dynamic behavior within the frequency range of 0.01–2 Hz and damping of
of the utility scale PV systems. The control model has the re- less than 10%. This analysis is performed utilizing the Small
quired modules for representing the active and reactive power Signal Analysis Toolbox (SSAT) which is part of the DSATools
control, in addition to representing current control capabilities software package [21]. The analysis is conducted for various
of the converters. PV penetration levels defined in Table III. The critical modes
The reactive power control block shown in Fig. 1 determines of the base case that are also present in all PV penetration cases
the reactive power reference value for the electrical control studied are presented in Table IV.
Authorized licensed use limited to: TULANE UNIVERSITY. Downloaded on September 16,2024 at 23:01:39 UTC from IEEE Xplore. Restrictions apply.
964 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 4, NO. 4, OCTOBER 2013

TABLE IV TABLE VI
CRITICAL MODES PRESENT IN THE BASE CASE EIGENVALUE SENSITIVITY CORRESPONDING TO THE INERTIA OF THE
DISPLACED GENERATORS IN THE 20% PV PENETRATION CASE

TABLE V
CRITICAL MODE DETRIMENTALLY AFFECTED BY HIGH PV GENERATION

TABLE VII
EIGENVALUE SENSITIVITY CORRESPONDING TO THE INERTIA OF THE
DISPLACED GENERATORS IN THE 40% PV PENETRATION CASE

Among the modes presented in Table IV, the second mode,

i.e. , is detrimentally affected by increasing the
is computed. These computations are performed with the aid
level of PV penetration. For comparison purposes, the charac-
of the SSAT software that calculates sensitivity of a particular
teristics of the aforementioned mode for various PV penetration
mode to variations of a system parameter. Inertia of the conven-
levels are shown in Table V. As seen from the results presented
tional generators is the perturbed system parameter in this case.
in Table V, the oscillatory mode has been adversely impacted
Tables VI and VII present a summary of the sensitivity of the
by increasing the level of PV penetration. This is evident by
critical mode to inertia variations of the conventional generators
comparing the real parts of the eigenvalues that have moved
being displaced by rooftop PVs in the 20% and 40% penetration
closer to the RHP as the PV penetration increases. This is also
levels, respectively. Considering the fact that the damping of the
reflected in the reduction in the damping ratio. The significant
system modes is determined by the real part of the eigenvalues,
reduction of the damping ratio from 30% to 40% PV penetra-
the real part sensitivity of the critical mode is presented in the
tion is mainly caused by displacing a number of large conven-
aforementioned tables.
tional generators that were in service prior to 40% PV penetra-
The negative real part sensitivities of the eigenvalues shown
tion level. By displacing those generators, the overall inertia of
in Tables VI and VII illustrate the adverse impact of high PV
the studied system dramatically reduces and hence the damping
penetration on system damping. By displacing with rooftop PVs
ratio of the adversely impacted mode, which is closer to those
or reducing the inertia of the conventional generators listed in
units, decreases.
those tables, the real part of the studied mode moves closer to the
In order to substantiate the results observed by the eigenvalue
RHP. Therefore, the sensitivity analysis results corroborate the
analysis, a sensitivity analysis is carried out corresponding to the
results derived from the full eigenvalue analysis of the system
mode which is detrimentally impacted with increased PV pen-
under various PV penetration levels.
etration. As more conventional generators are displaced with
Similar to the critical mode being studied, the fourth mode
distributed PV resources, the overall system inertia is reduced.
presented in Table IV, i.e., the mode with a frequency of
Sensitivity assessment of the critical modes of the system, with
1.72 Hz, has also shown to be detrimentally impacted by the
respect to inertia, is a means of explaining the detrimental im-
increased PV penetration levels. A comparison of the eigen-
pacts of the PV generation on small signal stability of the power
values corresponding to this mode for different PV levels is
systems. The eigenvalue sensitivity with respect to inertia of the
presented in Table VIII. These results illustrate that care should
th generator is expressed as [22]
be taken in selecting the displaced generators since the loss of
their inertia could result in unsatisfactory damping ratios of
(8) certain low frequency oscillatory modes.

The sensitivity of the detrimentally impacted critical mode of V. TRANSIENT ANALYSIS

the system with respect to the inertia of the system generators The results of the modal analysis presented in Section IV
that are being displaced due to the higher PV penetration levels point out the fact that reduced system inertia, which is a side
Authorized licensed use limited to: TULANE UNIVERSITY. Downloaded on September 16,2024 at 23:01:39 UTC from IEEE Xplore. Restrictions apply.
EFTEKHARNEJAD et al.: SMALL SIGNAL STABILITY ASSESSMENT OF POWER SYSTEMS 965

TABLE VIII
CRITICAL MODE DETRIMENTALLY AFFECTED BY HIGH PV GENERATION

Fig. 5. Single line diagram of the system near the generator at bus number
2102.

Fig. 4. Participation factor of the machines participating in the critical mode

detrimentally affected by high PV penetration. Fig. 6. Generator 2103 speed (p.u.) following a three phase fault at bus 2104.

effect of increasing distributed rooftop PVs, could result in de- aforementioned generators, the speed of either of these gener-
creased system damping. Through extensive studies of systems ators is monitored during the transients. The one line diagram
with rooftop PVs, it has been found to be advisable to examine of the system in vicinity of the two participating generators is
the transient response of the large scale transmission system to presented in Fig. 5.
identify problematic damping. This is illustrated here through a The two participating generators are located in the same area
continuation of the case study. and are connected to a 345 kV level bus with a step-up trans-
To illustrate the adverse effects of increased PV generation former. To identify the location of the disturbance that excites
on small signal stability of the systems, the detrimentally im- the detrimentally affected critical mode, various faults are simu-
pacted critical modes are scrutinized in the time domain. A dis- lated in the vicinity of the generator located at bus number 2103.
turbance that excites the most detrimentally impacted critical The types of the faults considered for this study are mainly three
mode with respect to the PV penetration level is simulated. The phase faults on the transmission system and single phase faults
purpose of this case study is to show how the system is im- are not considered for this study. Among the simulated fault sce-
pacted with decreased system inertia under various PV gener- narios, a three phase fault at the 345 kV bus of 2104, which
ation levels. Therefore, the cases studied are solely those that is cleared after four cycles, has shown to excite the mode ob-
excite the modes analyzed by eigenvalue analysis. Additional served in full eigenvalue analysis results. Later in this paper,
transient stability studies with high PV penetration can be found the validity of the studied transient case is verified with the aid
in [23]. of Prony analysis [24].
In order to identify the transient case that would excite a detri- Fig. 6 illustrates the speed of the generator located at bus 2103
mentally impacted critical mode, the dominant machine with the following the aforementioned three phase fault. As observed
highest participation factor in that mode needs to be identified. from Fig. 6, at higher PV penetration levels, more oscillations
The participation factors of various machines participating in are observed in speed of this generator. In other words, the de-
the first studied mode, i.e., , are presented in creased damping observed in modal analysis is verified by the
Fig. 4. As seen from these results, the generators located at bus transient analysis results.
2102 and 2103 have the highest participation factor in this mode. At a PV penetration level of 40% and above, the generator
In view of the fact that this mode is related to the speed of the speeds attain higher peaks and deeper dips. Higher oscillations
Authorized licensed use limited to: TULANE UNIVERSITY. Downloaded on September 16,2024 at 23:01:39 UTC from IEEE Xplore. Restrictions apply.
966 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 4, NO. 4, OCTOBER 2013

TABLE IX
PRONY ANALYSIS RESULTS OF THE GENERATOR 2103 SPEED FOR VARIOUS
PV PENETRATION LEVELS

small signal stability analysis and illustrate the need to care-

fully select the conventional generators that are displaced when
increased PV penetration occurs.
Fig. 7. Generator 2103 relative rotor angle (deg) following a three phase fault
at bus 2104.
VI. CONCLUSION
In this paper, the impact of high PV penetration on power
system small signal stability is investigated. Two different types
of PV generation units, i.e., rooftop PVs as well as utility scale
PV units, are added to a large system to simulate a case with high
PV penetration. Various levels of PV penetration are examined
for their effect on system small signal stability. Modal analysis
is performed on the system with no PVs as well as the cases after
different PV levels were introduced. The approach illustrated
via a case study is offered as an approach for the identification
of the impact of large-scale PV penetration on power systems.
Eigenvalue analysis performed on the studied system iden-
tified the locations of the critical modes of the system. These
modes are examined in the frequency range of 0.01–2 Hz with
Fig. 8. Active power output of the generator located at bus 2103 following a
three phase fault at bus 2104.
a damping ratio of less than 10%. Reduced system inertia as
a result of displacing conventional generators with distributed
rooftop PVs may result in a reduction in the damping torques
are also observed in the relative rotor angle of the generator lo- of the system modes. In order to illustrate this effect, results of
cated at bus 2103, under various PV penetration levels. Varia- the modal analysis were compared to cases with PV penetration
tions of the relative rotor angles are presented in Fig. 7. Detri- levels of up to 50%. The results of these comparisons identified
mental impact of the PV penetration levels, which result in in- the detrimental impacts of high PV penetration on decreasing
creased generator speeds, is also evident in the active power the damping of the critical modes of the system.
output of the generator located at bus 2103. Variations of the To highlight the adverse effects of reduced inertia on electro-
generator active power output are shown in Fig. 8. mechanical modes of oscillation, eigenvalue sensitivities to in-
Transient simulation results presented so far corroborate the ertia of the displaced generators are calculated. The sensitivity
results achieved by the small signal stability analysis. However, analyses pinpoint the fact that the eigenvalues are detrimentally
to reassert the premise of these simulations, a measure is re- impacted as the inertia of conventional generators is reduced.
quired to ascertain the simulated disturbance in fact excites the The negative sensitivities of the real part of the eigenvalue to
detrimentally impacted mode. Prony analysis is carried out to inertia show that the eigenvalues move closer to the right half
ensure the presence of the studied mode in the transient re- plane with a reduction in generator inertia and, therefore, result
sponse of the generator 2103 speed. This method is an exten- in degraded damping performance during system transients.
sion of Fourier analysis that directly estimates the frequencies, It is recommended that in cases of high PV penetration, time
damping, amplitude, and relative phase of the modes presents domain analysis be performed as well: time domain analysis is
in a given signal [24]. performed to examine the detrimental impacts of high PV pene-
The result of Prony analysis following a three phase fault at tration during system transients. Using the concept of participa-
bus 2104 is presented in Table IX. As presented in this table, the tion factors and by utilizing Prony analysis, a transient case that
same mode is observed in the generator speed variable as was excites a detrimentally impacted critical mode was identified in
observed in the eigenvalue analysis. The difference between the an example test bed. Transient studies carried out for various
results presented in Tables V and IX is related to the inaccuracy PV penetration levels substantiate the results of small signal
of the discrete models used in the Prony analysis. However, as analysis together with eigenvalue sensitivity analysis. Based on
expected, with an increase in PV penetration levels, the damping the simulation results presented in this work, PV generation is
of the studied mode is reduced. The transient stability results found to have a detrimental impact on small signal stability of
coupled with Prony analysis reinforce the results obtained by the studied system with increased penetration of PV systems.
Authorized licensed use limited to: TULANE UNIVERSITY. Downloaded on September 16,2024 at 23:01:39 UTC from IEEE Xplore. Restrictions apply.
EFTEKHARNEJAD et al.: SMALL SIGNAL STABILITY ASSESSMENT OF POWER SYSTEMS 967

In view of the fact that this detrimental impact of PV penetra- [15] P. Kundur, Power System Stability and Control, ser. EPRI Power
tion is caused by reduced system inertia, maintaining the critical System Engineering Series. New York, NY, USA: McGraw-Hill,
1994.
generators in service can be helpful in mitigating the impact of [16] GE Energy, Western Wind and Solar Integration Study: February
poor damping resulted from high PV penetration. The critical 2008–February 2010, NREL Rep. SR-550-47434, May 2010.
generators can be identified by examining the sensitivity of the [17] A. Ellis, PV System Models for System Planning and Interconnection
Studies, Sandia National Laboratories, Oct. 2009.
critical modes to the inertia of the generators. Although the con- [18] North American Electric Reliability Corporation, Standard Models for
clusions are derived only from the studied system, due to large Variable Generation, NERC special report, draft Feb. 2010.
size of the system and the fact that the system is representative [19] K. Clark, N. W. Miller, and R. Walling, Modeling of GE Solar Photo-
voltaic Plants for Grid Studies, Schenectady, NY, USA, General Elec-
of most of North American power systems, the conclusions can tric International report, Apr. 2010.
be safely generalized without loss of generality to the systems [20] F. Fernandez-Bernal, L. Rouco, P. Centeno, M. Gonzalez, and M.
of same size and characteristics. In addition, the procedure to Alonso, “Modelling of photovoltaic plants for power system dynamic
studies,” in Proc. 5th Int. Conf. Power System Management and
perform the study is general and applies to all systems. Control, Apr. 2002, pp. 341–346.
[21] “SSAT User’s Manual,” ver. 11.0, Powertech Labs Inc., Vancouver,
BC, Canada.
[22] T. Smed, “Feasible eigenvalue sensitivity for large power systems,”
REFERENCES IEEE Trans. Power Syst., vol. 8, no. 2, pp. 555–561, May 1993.
[23] S. Eftekharnejad, V. Vittal, G. T. Heydt, B. Keel, and J. Loehr, “Impact
[1] U.S. Environmental Protection Agency, Renewable Portfolio Stan-
of increased penetration of photovoltaic generation on power systems,”
dards Fact Sheet 2009 [Online]. Available: http://www.epa.gov/chp/
IEEE Trans. Power Syst., vol. 28, no. 2, pp. 893–901, May 2013.
state-policy/renewable_fs.html
[24] J. F. Hauer, C. J. Demeure, and L. L. Scharf, “Initial results in Prony
[2] Interconnecting Distributed Resources With Electric Power Systems,
analysis of power system response signals,” IEEE Trans. Power Syst.,
Standard 1547, IEEE [Online]. Available: http://groupe.ieee.org/
vol. 5, no. 1, pp. 80–89, Feb. 1990.
groups/scc21/1547/1547/index.html
[3] Standard for Inverters, Converters, Controllers and Intercon-
nection System Equipment for Use With Distributed Energy Sara Eftekharnejad (S’12–M’13) received the B.Sc. degree from University
Resources, UL 1741 [Online]. Available: http://ulstandardsin- of Tehran, in 2006, and the M.Sc. degree from West Virginia University, Mor-
fonet.ul.com/scopes/1741.html gantown, WV, USA, in 2008, both in electrical engineering. She received the
[4] Y. Liu, J. Bebic, B. Kroposki, J. de Bedout, and W. Ren, “Distribution Ph.D. degree in electrical engineering from Arizona State University, Tempe,
system voltage performance analysis for high-penetration PV,” in Proc. AZ, USA, in 2012.
IEEE Energy 2030 Conf., Atlanta, GA, USA, Nov. 2008, pp. 1–8. She is currently working as a transmission planning engineer at Tucson Elec-
[5] K. Turitsyn, P. Šulc, S. Backhaus, and M. Chertkov, “Distributed con- tric Power (TEP) Company, Tucson AZ, USA.
trol of reactive power flow in a radial distribution circuit with high
photovoltaic penetration,” in Proc. IEEE Power and Energy Society
General Meeting, Minneapolis, MN, USA, Jul. 2010, pp. 1–6.
[6] J. W. Smith, W. Sunderman, R. Dugan, and B. Seal, “Smart inverter Vijay Vittal (SM’78–F’97) received the B.E. degree in electrical engineering
volt/var control functions for high penetration of PV on distribution from the B.M.S. College of Engineering, Bangalore, India, in 1977, and the
systems,” in Proc. IEEE Power Systems Conf. Exposition, Phoenix, Ph.D. degree from Iowa State University, Ames, IA, USA, in 1982.
AZ, USA, Mar. 2011, pp. 1–6. He is currently the director of the Power Systems Engineering Research
[7] P. P. Barker and R. W. De Mello, “Determining the impact of dis- Center (PSerc) and is the Ira A. Fulton Chair Professor in the Department of
tributed generation on power systems: Part I radial distribution sys- Electrical Engineering at Arizona State University, Tempe, AZ, USA.
tems,” in Proc. IEEE Power and Engineering Society Summer Meeting, Dr. Vittal is a member of the National Academy of Engineering.
Seattle, WA, USA, Jul. 2000, vol. 3, pp. 1645–1656.
[8] Y. Zhang, C. Mensah-Bonsu, P. Walke, S. Arora, and J. Pierce, “Tran-
sient over-voltages in high voltage grid connected PV solar intercon-
nection,” in Proc. IEEE Power and Energy Society General Meeting, Gerald Thomas Heydt (S’62–M’64–SM’80–F’91–LF’08) is from Las Vegas,
Minneapolis, MN, USA, Jul. 2010, pp. 1–6. NV. He received the Ph.D. degree in electrical engineering from Purdue Uni-
[9] General Electric Co., Reliable, Low Cost Distributed Generator/Utility versity in 1970.
System Interconnect 2001 Annual Report, GE Corporate Research and His industrial experience is with the Commonwealth Edison Company,
Development, NREL/SR-560-34634, Aug. 2003 [Online]. Available: Chicago, IL, USA, and E.G.&G., Mercury, NV, USA. He is presently the site
http://www.nrel.gov/docs/fy03osti/34634.pdf director of a power engineering center program at Arizona State University in
[10] S. Achilles, S. Schramm, and J. Bebic, Transmission System Per- Tempe, AZ, USA, where he is a Regents’ Professor.
formance Analysis for High-Penetration Photovoltaics, National Dr. Heydt is a member of the National Academy of Engineering. He is the
Renewable Energy Laboratory, Rep. SR-581-42300, Feb. 2008 [On- recipient of the IEEE 2010 Richard H. Kaufmann award.
line]. Available: http://www.nrel.gov/docs/fy08osti/42300.pdf
[11] H. Liu, L. Jin, D. Le, and A. A. Chowdhury, “Impact of high pene-
tration of solar photovoltaic generation on power system small signal
stability,” in Proc. Int. Conf. Power System Technology, Oct. 2010, pp.
1–7. Brian Keel (M’98–SM’08) recieved the Bachelors and Masters degrees from
[12] W. Du, H. F. Wang, and L.-Y. Xiao, “Power system small signal the University of Illinois, Champaign, IL, USA, in electrical engineering, in
stability as affected by grid-connected photovoltaic generation,” Eur. 1988 and 1989, respectively.
Trans. Elect. Power, DOI: 10.1002/etep.598. He has 20 years of experience in the power industry and is the Manager of
[13] B. Tamimi, C. Canizares, and K. Bhattacharya, “Modelling and perfor- Transmission System Planning at Salt River Project, Phoenix, AZ, USA.
mance analysis of large-scale photovoltaic generation on voltage sta-
bility and inter-area oscillations,” in Proc. IEEE PES GM, Detroit, MI,
USA, 2011.
[14] P. Kundur, J. Paserba, V. Ajjarapu, G. Andersson, A. Bose, C. Jeffrey Loehr received the Bachelors degree in geography from Arizona State
Cañizares, N. Hatziargyriou, D. Hill, A. Stankovic, C. Taylor, T. V. University (ASU) in 2003. He is working toward an Electrical Engineering de-
Custem, and V. Vittal, “Definition and classification of power system gree at ASU.
stability,” IEEE Trans. Power Syst., vol. 19, no. 2, pp. 1387–1401, He has 10 years of experience in the power industry and is an Analyst in
May 2004. Transmission System Planning at Salt River Project, Phoenix, AZ, USA.

Authorized licensed use limited to: TULANE UNIVERSITY. Downloaded on September 16,2024 at 23:01:39 UTC from IEEE Xplore. Restrictions apply.