Vous êtes sur la page 1sur 7

Underfrequency Load Shedding Scheme for

Systems with Large Penetration Renewable


Variable Generation
Pedro Spallarossa, Alberto Del Rosso
Departamento de Eléctrica, Universidad Tecnológica Nacional
Av. Medrano 951, Buenos Aires, Argentina
pspallarossa@frba.utn.edu.ar
adelrosso1@frba.utn.edu.ar

Abstract— Under-frequency load shedding (UFLS) is well- excursions, as prolonged operation at frequencies above or
known and widely used emergency control scheme to prevent below the nominal value (60 or 50 HZ) can damage power
system instability under disturbances that cause large system equipment. An optimally designed UFLS must be
imbalance between load and generation. A properly design able to shed the minimum amount of load necessary for a
UFLS scheme should be able to shed the minimum amount of
safe and smooth transition of the system from an emergency
load necessary for a safe and smooth transition of the system
from emergency condition to a normal operation state. As situation to a normal equilibrium state [2][3].
penetration of renewable generation increases, the dynamic Load shedding schemes are usually performed over
behavior of power system changes, and consequently, the several stages, where the number of stages and the load to
performance of the UFLS is affected. This paper presents an be shed per stage is selected in such a way to minimize the
analysis of the impact of renewable generation integration on amount of load shed in excess for the less severe faults. A
the optimal design of a UFLS scheme. The analysis is proper coordination between adjacent stages has to be
performed in a generic power system model implemented in achieved in order to avoid that more load steps than
PSS-E software. Several scenarios with different levels of wind necessary are activated for lower overload conditions. The
generation are evaluated for the analysis. For each dispatch
design of UFLS scheme and tuning of its control parameters
scenario, various combinations of generation disturbances are
simulated. is carried out mainly through dynamic simulations,
considering a number of selected operating scenarios and
Resumen— Los sistemas de corte de carga por generation disturbance conditions [2].
subfrecuencia (UFLS por su sigla en inglés) son muy utilizados Detailed simulation studies as well as statistical analysis
para evitar inestabilidad de frecuencia en sistemas eléctricos on systems with large penetration of renewable generation
ante perturbaciones que originan grandes desbalances entre la show that the displacement of conventional generation with
carga y la generación. Un esquema de corte de carga por wind and solar PV significantly affects system frequency
subfrecuencia debidamente diseñado deber ser capaz de response [4][5][6]. Indeed, in case of variable-speed wind
desconectar la mínima cantidad de demanda necesaria para
turbines, because the rotational speed is decoupled from the
que el sistema se recupere desde un estado de emergencia, en
forma segura y con el menor impacto posible sobre la grid frequency by power converter, the kinetic energy stored
demanda. Es sabido que en la medida en que aumenta la in the rotating mass of turbine blades does not contribute to
inserción de generación renovable en el sistema el the inertia of the grid [7]. When wind generation cause
comportamiento dinámico del mismo cambia notablemente. some conventional synchronous generators to be
Como consecuencia de esto, el desempeño de un esquema de decommitted, the total system inertia is consequently
protección por subtensión también se verá afectado. reduced. As the amount of wind generation increases, the
En este trabajo se presenta un análisis del posible impacto frequency nadir is lower, and the rate of frequency decline
de generación renovable en el diseño óptimo de un esquema is greater. As a consequence, the performance of the UFLS
UFLS. El análisis se realiza con un modelo de sistema de
for a given load level and disturbance conditions will
potencia genérico implementado en el software de simulación
PSS-E. Se evalúan una serie de escenarios con diferentes change depending on the amount of renewable generators
niveles de generación eólica. Para cada uno de estos escenarios committed in that particular state. It is therefore very
de despacho, se analizan varias combinaciones de desconexión difficult to achieve a UFLS robust enough to operate with
de generación. the desired dependability, security and selectivity attributes
over the full range of generation mix and dispatch
I. INTRODUCTION conditions that it will encounter. Some researchers propose
Underfrequency load shedding (UFLS) is a common adaptable load shedding schemes that use frequency rate
practice for electric power utilities to ensure frequency information to change the relay parameters depending on
stability is maintained after disturbances causing severely the severity of the fault and system dynamic response, so
large imbalance between load and generation. In isolated that the security and selectivity is maintained for all possible
power systems, UFLS schemes play an important role in conditions [8][9][10]. Other researcher propose the use of
protecting the system integrity [1]. optimization techniques for a robust tuning of ULFS
The amount and timing of load shed for a particular parameters [11][12]. However, the methods proposed in
contingency event must be adequate to limit frequency
those references have not being tested for cases with large may continue to decline at a low value entering into
penetration of renewables. generator protection zones. In such a situation, generating
This paper presents an analysis of the impact of units may trip aggravating system condition, making more
generation mix change, more specifically due to the difficult system recovery, and eventually leading the system
integration of renewable, on the optimal design and tuning to frequency instability. Therefore, adequate coordination
of an underfrequency load shedding scheme. The analysis is between the UFLS program and generator off-nominal
performed in a generic power system model implemented in frequency relays is crucial to ensuring the UFLS program
the simulation software PSS-E. Several scenarios with meets its design objectives. In Argentina, the grid code for
different levels of wind generation are evaluated. For each the interconnected transmission system establishes the off-
dispatch scenario, various combinations of generation nominal frequency requirements that all generators
disconnection - combinations of conventional generators connected to the system must comply [17]. Fig. 1 illustrates
and wind plants - are considered for the analysis the generator off-frequency limits for the Argentinean
The authors are currently working on new approaches to electric system.
design robust UFLS for island systems with large
penetration of renewables. The findings of this study serve a 53
basis for that effort. PROHIBITED OPERATION
52

II. DESIGN CONSIDERATIONS OF AN UFLS 51

A. Desired attributes of the scheme 50 CONTINUOUS OPERATION

Hz
An UFLS, as any other protection system, is designed to 49
operate under conditions when its misoperation or failure RESTRICTED OPERATION
would have a significant impact on the security of the power 48

system. On the other hand, if the UFLS actuates when is not 47


require to do so, unnecessary loss of load is produced. The 0 10 20 30 40
following are basic attributes of a properly designed UFLS. Seg.
Sec.
These attributes also apply in general terms to any
protection system or scheme [13][14][15]: Fig. 1. Generator off-frequency limits according to the Argentinean Grid
 Dependability: The certainty that the UFLS scheme will Code [17]
operate when required; that is, in all cases where Also, in some cases, an ULFS needs to be coordinated
emergency controls are necessary to avoid system with voltage control equipment. Certainly, if as a
instability. consequence of a large disturbance significant amount of
 Security: The certainty that the protection scheme will not load is curtailed by the ULFS relay, it may be necessary to
operate when it is not required. A protection scheme such disconnect in-service shunt capacitors to avoid harmful
as UFLS is to actuate only under disturbances that imply overvoltage conditions. Further, in some cases, fast
risk of frequency instability. Conversely, the design connection of shunt reactors in selected locations is also
should be able to avoid incorrect operation due to internal required.
failure of its components.
 Selectivity: The ability to select the correct and minimum III. APPROACH OF THE STUDY
amount action to perform the intended function, that is, to As previously described, the objective of this study is to
avoid using disruptive controls such as excessive load assess the impact of renewable generation on the
shedding if they are not necessary to avoid system performance of an ULFS scheme, in order to identify the
collapse. necessary improvements to accomplish an optimal and
As it will be shown later in this paper, selectivity is robust scheme that fulfil all the desired attributes described
noticeable impacted when the generation mix change due to previously. The analysis is conducted through dynamic
high penetration of renewable plants. simulations on a generic yet realistic system. Several
scenarios with different levels of wind generation are
B. Coordination with other protections evaluated. For each dispatch scenario, various combinations
An UF protection scheme must shed load on proper of generation disturbances are simulated.
amount and timing to arrest frequency excursion within The UFLS scheme is first designed to be optimal for one
acceptable limits to avoid system instability and damage of particular operating scenario; that is, the number of stages
sensitive equipment. Generating units, both conventional and parameters setting are determined in such a way the
and renewable, are particularly sensitive to off-nominal scheme is secure and selective for a representative operating
frequency conditions [3][16]. Generators are normally and fault condition. Then, the UFLS scheme with the
equipped with frequency relays to trip the unit if it operates proposed settings is simulated in all the other scenarios to
at off-frequency conditions for a prolonged period of time, test its robustness.
to protect the unit from damage. The frequency setting of The design of UFLS scheme is a trial an error procedure
these relays are usually set equal to or slightly higher than intended to determine the best combination of number and
the steps of the turbine damage curve, and time setting are size of load shedding, as well as the frequency thresholds
set equal to or slightly lower than the steps of the turbine and delays of each stage. The following general criteria and
blade time restricted operating limits. If the UFLS fails to methodological aspects have been considered in this study:
arrest frequency excursion after a disturbance, the frequency
 The first step is to define the maximum overload level the TABLE I
DISPATCHED GENERATION
load shedding scheme is to protect. The amount of load
shed varies depending on the system characteristics. The Active Power Dispatch
maximum amount of load shedding is usually limited to GENERATOR Savnw0 Savnw10 Savnw20 Savnw30
about 30 to 60% of the total system demand. In this case [MW] [MW] [MW] [MW]
we considered a generation loss of 45% with respect to NUC-A 1 360 360 360 360
the initial load. NUC-A 2 345 345 345 345
 Then, the number of steps and the percentage of load to NUC-B 1 345 - - -
be shed per step are defined. The selection is somehow NUC-B 2 345 345 345 -
arbitrary. However, in actual implementation this depends HYDRO-G 355 320 270 210
upon the substation and feeder arrangements and the way MINE-G 613 651 706 778
the load is distributed among them. In general, the larger CATDOG-G 200 200 200 200
the maximum overload to protect, the larger the number URBGEN 1 345 345 - -
of stages. URBGEN 2 345 345 345 345
 The number of stages and the load to be shed per step is EOLICA 1 - 345 345 345
selected in such a way to minimize the amount of load EOLICA 2 - - 345 345
shed in excess for the less severe faults. EOLICA 3 - - - 345
 The amount of load shed in the first stages should be as TOTAL 3253 3256 3261 3273
less as possible to reduce that excess of load shed for the RESERVE 410 390 385 390
most probable contingencies.
 The pick-up frequency and the time delay setting of each
stage are determined through dynamic simulations. B. Load Modelling
 The spinning reserve is sufficient to allow full restoration The loads are represented with static load models that
of frequency after the loss of the largest generating unit include sensitivity of load with respect to bus voltage and
on-line. Such condition implies that the outage of any frequency. The loads are represented with static load models
generator should not activate the first stage of the UFLS that include sensitivity of load with respect to bus voltage
relays. and frequency. The load model is as follows:
 Frequency excursions must not penetrate the restricted P  Pload (a1v n1  a 2 v n2  a3v n3 )(1  a7 f )
operation zone of the generators off-nominal frequency
Q  Qload (a 4 v n4  a5 v n5  a6 v n6 )(1  a8 f )
protection relays.

IV. MODEL DESCRIPTION


Where Pload and Qload are the real and reactive power
The one-line diagram of the study system is shown in Fig. loads that would be drawn at nominal frequency and one per
2 in the Appendix. The system is comprised of 31 buses, 36 unit voltage. The load model parameters considered for this
branches, 8 power plants with a total of 12 generating units. study system are:
The system contains 9 loads totalling to 3200 MW.
a1  0.0 a 2  0.8 a 3  0. 2
A. Generation Dispatch Scenarios
a 4  0.0 a 5  0. 5 a6  0.5
The 12 generating units are composed of 9 conventional
units, including one hydro generator, and 3 DFIG wind n1  0.0 n 2  1. 0 n3  2. 0
farms of 345 MW nominal capacity each. n 4  0. 0 n5  1.0 n6  2.0
Four generation scenarios have been considered for the
analysis (Table I). Each scenario, 9 of the 12 total generator a7  1.5
are in service. In the first scenario (savenw0) only
conventional units are committed. The other three scenarios,
a8  0. 0
have 10%, 20% and 30% of wind generation penetration Which implies the active power is model as a
respectively. In each case, some conventional units have combination of 80% constant current and 20 % constant
been decommitted, and the dispatch of some other units impedance, while reactive power is a combination of 50%
have been backed down in order to accommodate wind constant current and 50% constant impedance.
generation. The dispatch generation for each case is shown It is necessary to represent in the simulations the
in Table I. measures to avoid large overvoltages when a significant
amount of load is shed by the UFLS scheme. For this
particular system, the control actions considered are the
switching of fixed capacitor installed in buses 154 and 205
or the reactor in bus 151.
C. Relay model
A standard underfrequency relay with five load shedding
stages is applied to all loads in the study system. The relay
model LDS3AR of the standard PSS/E library is used for
this purpose.
D. Generation Contingency TABLE IV
UFLS PARAMETERS SET - I
For each of the four generation scenarios, two sets of
generator contingency cases were simulated: In cases A-1 to Stage
Frequency set-
Load Shed Tim e Delay
A-6, contingencies are applied only to conventional RELAY-I
point

generators; while in cases B-1 to B-6, contingencies are [Hz] [%] [sec.]

considered for both, conventional and wind generators. 1 49.2 5 0.2


Tables II and III present details of generator contingencies 2 49 5 0.2
for cases A and B respectively. 3 48.8 12 0.2
TABLE II 4 48.5 12 0.6
GENERATION CONTINGENCY CASE A 5 48.2 11 1
CASE A CASE_A-1 CASE_A-2 CASE_A-3 CASE_A-4 CASE_A-5 CASE_A-6
TOTAL GEN. [MW] 3253.6 3253.6 3253.6 3253.6 3253.6 3253.6
NUC-A 1 Table V summarizes the results of this case. The case
name codification is as follows:
NUC-A 1 NUC-A 1
BUS NUC-A 1 NUC-A 1 NUC-A 2
NUC-A 1 NUC-A 2 NUC-A 2
NAME CATDOG1 NUC-A 2
CATDOG URBGEN 2
URBGEN 2
CATDOG
360
360 360
360 360 345
[MW] 360 345 345
GEN 200 345
200 345
345
TRIPPED 200
0.1
0.1 0.1
TIME 0.1 0.1 0.15
0.1 0.15 0.15
[seg.] 0.15 0.15
0.17 0.17
0.17
0.17
[MW] 360 560 705 905 1050 1250
[%] 11% 17% 22% 28% 32% 38%

TABLE III TABLE V


GENERATION CONTINGENCY CASE B RESULTS FOR RELAY PARAMETER SET -I

CASE B CASE_B-1 CASE_B-2 CASE_B-3 CASE_B-4 CASE_B-5 CASE_B-6 MIN. FINAL MIN. FREQ.
LOAD SHED
TOTAL GEN. [MW] 3253.6 3253.6 3253.6 3253.6 3253.6 3253.6 CASE FREQUENCY FREQUENCY DERIVATIVE
EOLICA1 [Hz] [Hz] [Hz/Sec] Stage [MW]
EOLICA1 EOLICA1
BUS EOLICA1 EOLICA1 NUC-A 1 CASE_A0-1 49.39 49.70 -0.37 0 0
EOLICA1 NUC-A 1 NUC-A 1
NAME CATDOG1 NUC-A 1
CATDOG1 URBGEN
URBGEN
CASE_A0-2 48.96 49.77 -0.35 2 326
CATDOG1
345 CASE_A0-3 48.41 49.87 -0.94 3 718
345 345
345 345 360 CASE_A0-4 48.44 49.81 -0.73 3 718
[MW] 345 360 360
GEN 200 360
200 345
345 CASE_A0-5 48.14 50.00 -1.08 4 1110
TRIPPED 200 CASE_A0-6 47.47 50.13 -2.61 5 1468
0.1
TIME 0.1 0.1
0.1 0.1
0.15
CASE_A10-1 49.19 49.77 -0.45 1 163
[seg.]
0.1
0.15 0.15
0.15 0.15
0.17 CASE_A10-2 48.84 49.74 -0.48 2 326
0.17 0.17
0.17 CASE_A10-3 48.38 50.29 -1.08 4 1110
[MW] 345 545 705 905 1050 1250 CASE_A10-4 48.44 50.18 -1.56 3 1110
[%] 11% 17% 22% 28% 32% 38% CASE_A10-5 47.95 50.51 -2.00 5 1469
CASE_A10-6 47.37 50.21 -2.46 5 1469
CASE A

CASE_A20-1 49.13 49.73 -0.49 1 166


V. RESULTS AND FINDINGS CASE_A20-2 48.83 49.70 -0.60 2 332
CASE_A20-3 48.43 50.41 -1.32 4 1128
As described in Section III, the approach is to design the CASE_A20-4 48.38 50.25 -1.80 4 1128
UFLS for a selected operation scenarios and contingency CASE_A20-5 47.85 50.17 -2.27 4 1128
CASE_A20-6
conditions, and test its robustness for all the other CASE_A30-1
47.24
49.03
50.42
49.65
-3.27
-0.56
5
1
1492
166
conditions. CASE_A30-2 48.71 50.10 -0.74 3 732
CASE_A30-3 48.22 50.54 -1.66 4 1132
A. Relay Parameter Set I CASE_A30-4 48.18 50.32 -1.85 4 1132
CASE_A30-5 47.29 50.43
In this case, the scenario selected for tunning the UFLS CASE_A30-6 19.42 49.45
-3.05
-96.45
5
5
1498
1453
relays is the CASE_B20. This scenario represents a dispatch CASE_B10-1 49.42 49.70 -0.37 0 0
with 20% penetration of wind generation, and contingencies CASE_B10-2 48.97 49.79 -0.31 2 326
involving both conventional and wind generators. CASE_B10-3 48.45 49.90 -0.90 3 718
CASE_B10-4 48.48 49.83 -0.58 3 718
This scenario has been selected as reference for tuning CASE_B10-5 48.05 50.00 -1.87 4 1110
the relay parameters because it represents an average CASE_B10-6 47.49 50.19 -2.09 5 1469
scenario, where the wind penetration is not at the highest CASE_B20-1 49.24 49.59 -0.40 0 0
level, and the contingency are composed of a mix of CASE_B20-2
CASE_B20-3
48.92
48.51
49.76
49.91
-0.41
-0.97
2
3
332
730
conventional and renewable generators.
CASE B

CASE_B20-4 48.44 49.79 -1.51 3 730


The basic criteria and design consideration described in CASE_B20-5 48.01 50.16 -1.94 4 1128
the previous section have followed to tune the relays. Table CASE_B20-6 47.86 50.34 -2.17 5 1492
CASE_B30-1 49.14 49.75 -0.48 1 166
IV shows a set of relay parameter (Set-I). CASE_B30-2 48.84 49.71 -0.58 2 333
CASE_B30-3 48.65 50.01 -1.33 3 732
CASE_B30-4 48.40 50.28 -1.77 4 1132
CASE_B30-5 48.08 50.17 -2.54 4 1132
CASE_B30-6 47.53 50.46 -2.65 5 1498
It can be observed in this table that the design is correct TABLE VII
RESULTS FOR RELAY PARAMETER SET -II
for the reference case (CASE_B20), but it does not comply
with the required attributes for other operation and MIN. FINAL MIN. FREQ.
LOAD SHED
contingency conditions. In some cases the outage of single CASE FREQUENCY FREQUENCY DERIVATIVE
[Hz] [Hz] [Hz/Sec] Stage [MW]
generator causes activation of the first stage (Cases A10_1, CASE_A0-1 49.39 49.70 -0.37 0 0
A20_1, A30_1, and B30_1). Also, the selectivity of the CASE_A0-2 48.96 49.77 -0.35 2 326
scheme is lost in most of the cases, especially in situations CASE_A0-3 48.41 50.23 -0.94 4 1110
where the generators that remain connected are mainly wind CASE_A0-4 48.44 50.14 -0.73 4 1110
CASE_A0-5 48.16 50.00 -1.08 4 1110
plants (Cases A10 and A20). More importantly, in some CASE_A0-6 47.52 50.13 -2.61 5 1468
cases the scheme fails to avoid frequency to decline below CASE_A10-1 49.19 49.77 -0.45 1 163
the minimum threshold (47.5 Hz), entering into the CASE_A10-2 48.84 49.74 -0.48 2 326
generator protection zone. Such condition is critical because CASE_A10-3
CASE_A10-4
48.38 50.28 -1.08 4 1110
48.44 50.18 -1.56 4 1110
it would imply that some generator will disconnect due to CASE_A10-5 48.02 49.97 -2.00 4 1110
the off-nominal protection relays, aggravating the CASE_A10-6 47.52 50.21 -2.46 5 1469

CASE A
generator/load imbalance condition, jeopardizing system CASE_A20-1 49.13 49.73 -0.49 1 166
CASE_A20-2
recovery. CASE_A20-3
48.83
48.43
49.70
50.41
-0.60
-1.32
2
4
332
1128
It was expected that by using this average scenario as a CASE_A20-4 48.38 50.24 -1.80 4 1128
reference for tuning the UFLS relay, the setting would CASE_A20-5 47.94 50.17 -2.27 4 1128
robust for other cases. However, as the result reveals, the CASE_A20-6 47.28 50.42 -3.27 5 1492
CASE_A30-1 49.03 49.65 -0.56 1 166
setting is not appropriate for several of the scenarios CASE_A30-2 48.71 50.10 -0.74 3 732
analysed. CASE_A30-3 48.22 50.54 -1.66 4 1132
CASE_A30-4 48.18 50.32 -1.85 4 1132
CASE_A30-5 47.45 50.42 -3.05 5 1498
B. Relay Parameter Set II CASE_A30-6 32.34 48.98 -99.60 5 1498
CASE_B10-1 49.42 49.70 -0.37 0 0
Several trials with different parameter setting were made CASE_B10-2 48.97 49.79 -0.31 2 326
to improve the scheme performance for other than the CASE_B10-3 48.45 50.26 -0.90 4 1110
reference conditions used for parameter Set I. Because it is CASE_B10-4
CASE_B10-5
48.48 49.83 -0.58 3 718
48.04 50.00 -1.87 4 1110
difficult to obtain a single parameter set that performs CASE_B10-6 47.50 50.19 -2.09 5 1469
optimally for all the conditions, Case A0 was used as a CASE_B20-1 49.24 49.59 -0.40 0 0
reference in this case. The revised relay setting is shown in CASE_B20-2 48.92 49.76 -0.41 2 332

Table VI. As it can be seen in this table, it change slightly CASE_B20-3 48.51 49.91 -0.97 3 730
CASE B

CASE_B20-4 48.44 50.22 -1.51 4 1128


with respect the previous case. The results of this analysis CASE_B20-5 48.05 50.15 -1.94 4 1128
are presented in Table VII. CASE_B20-6 47.91 50.34 -2.17 5 1492
CASE_B30-1 49.14 49.75 -0.48 1 166
TABLE VI CASE_B30-2 48.84 49.71 -0.58 2 333
UFLS PARAMETERS SET - II CASE_B30-3 48.65 50.01 -1.33 3 732
CASE_B30-4 48.40 50.28 -1.77 4 1132
Frequency set-point Load Shed Tim e Delay CASE_B30-5 48.11 50.17 -2.54 4 1132
Stage
CASE_B30-6 47.63 50.46 -2.65 5 1498
RELAY-II
[Hz] [%] [sec.]
1 49.2 5 0.2 It can be observed in this case table that with this
2 49 5 0.2 parameter setting, the scheme provide effective protection
3 48.8 12 0.2 for all the contingency conditions for most of the cases,
4 48.5 12 0.3
except for cases Case_A20 and Case_A30, where the timing
5 48.2 11 1
and amount of load shed is not enough to avoid the
frequency to enter into the generator protection zone.
Moreover, in the Case_A30_6 the system is unstable. It
should be notice that this is an extreme case, where the wind
penetration is at highest level and all the contingency only
involved conventional units.
Further, even in those cases where the protection is safe,
that is; the frequency excursion does not penetrate
generators protection zone, the selectivity of the scheme is
adequate.

VI. CONCLUSIONS
Under-frequency load shedding (UFLS) is well-known
and widely used emergency control option to prevent
system instability when large disturbances jeopardize
system stability. A properly design UFLS scheme should be
able to shed only the minimum amount of load necessary
for a safe and smooth transition of the system from
emergency condition to a normal operation state.
Changes in generation mix, especially high penetration of TECHNOLOGY ISSN 1007-0214 14/18 pp823-828 Volume 13,
Number 6, December 2008
renewable generation such as wind and solar PV,
[13] CIGRE, Task Force 38.02.19, “System protection schemes in power
significantly affect the system dynamic behaviour of a networks”, Technical Brochure REF. 187, 2001.
power system. Therefore, as the renewable penetration [14] Western Electricity Coordinating Council (WECC), Remedial
increases displacing conventional units, the design and Action Scheme Design Guide, prepared by Relay Work Group,
tuning of an UFLS scheme is affected. October 2005
[15] J. D. McCalley, “System Protection Schemes: Limitations, Risks,
This paper presents an analysis of the impact of and Management”, PSERC Publication 10-19 December 2010
generation mix change, more specifically due to the [16] T. Akerman (Editor), Wind Power in Power Systems, John Wiley &
integration of renewable, on the optimal settings of an Sons Ltd., 2005
under-frequency load shedding scheme. The results of the [17] CAMMESA. Los Procedimientos Versión XXV – 2014. PT Nr. 9.
http://portalweb.cammesa.com/Pages/BackupBotoneraAneriorIzquie
study reveal that it is very difficult to obtain a single relay rda/Normativa/procedimientos.aspx
parameter set that performs in an optimal fashion for all
possible operating and contingency conditions. The security
and selectivity features of the protection scheme are
severely impacted in some cases.
The authors are currently working on different
approaches to achieve robust UFLS systems for systems
with high penetration of renewable generation. The focus is
on isolated or island power system where unbalance
conditions can have severe consequences for system
stability. The system usually rely on special protection
scheme to maintain system operation integrity. One
promising option is the use adaptive relays, which change
the parameters setting based on the system frequency
gradient and other dynamic variables. The adaptable UFLS
system is expected to be a flexible and versatile
underfrequency protection scheme, which will allow
attaining a fine coordination with generator protection off-
frequency relays, reducing the risk of generator tripping
while minimizing the amount of load shed for the different
disturbances conditions.

REFERENCES
[1] IEEE C37.106, “Guide for Abnormal Frequency Protection for
Power Generating Plant”, IEEE Standards, IEEE, New York (NY,
USA), 2004.
[2] H. E. Lokay, V. Burtnyk, “Application of Underfrequency Relays
for Automatic Load Shedding”, IEEE Transactions on Power
Apparatus and Systems (Volume:PAS-87 , Issue: 3 )
[3] EPRI Power System Dynamics Tutorial, EPRI, Palo Alto, CA: 2002.
1001983.
[4] V. Singhvi, P. Pourbeik; N. Bhatt ; D. Brooks ; Yingchen Zhang ; V.
Gevorgian ; E. Ela ; K. Clark, “Impact of wind active power control
strategies on frequency response of an interconnection”, IEEE PES
General Meeting, Vancouver, BC July 2013
[5] Frequency Response Adequacy and Assessment: Global Industry
Practices and Potential Impact of Changing Generation Mix. EPRI,
Palo Alto, CA: 2012. 1024275.
[6] Doherty, R, Mullane, A., Lalor, G., Burke, D., Bryson, A. and
O’Malley, M.J. “An Assessment of the Impact of Wind Generation
on System Frequency”, IEEE Transactions on Power Systems, Vol.
25, pp. 452 –460, 2010.
[7] Undrill, J.M. 2010. Power and Frequency Control as it Relates to
Wind-Powered Generation. LBNL-4143E. Berkeley: Lawrence
Berkeley National Laboratory
[8] H. Bevrani, “Robust Power System Frequency Control”, Springer;
2009 edition - ISBN-10: 0387848770
[9] H. Bevrani G. Ledwic,; J. J. Ford, “On the use of df/dt in power
system emergency control”, Power Systems Conference and
Exposition, 2009. PSCE '09. IEEE/PES
[10] V. Terzija, “Adaptive Underfrequency Load Shedding Based on the
Magnitude of the Disturbance Estimation”, IEEE Transaction on
Power Systems, VOL. 21, NO. 3, AUGUST 2006
[11] L. Sigrist, and L. Rouco, “Design of UFLS Schemes Taking Into
Account Load Variation”, 18th Power Systems Computation
Conference, Wroclaw, Poland – August 18-22, 2014.
[12] D. Mingchui, L. Chinwang, and W. Chikong, “Adaptive Under-
Frequency Load Shedding”, TSINGHUA SCIENCE AND
APPENDIX

FIG. 2. DIAGRAM OF THE STUDY CASE

Vous aimerez peut-être aussi