Professional Documents
Culture Documents
Threedimensional Nondetection Zonefor Assessing Antiislanding Detection Schemes
Threedimensional Nondetection Zonefor Assessing Antiislanding Detection Schemes
net/publication/233042508
CITATIONS READS
8 136
4 authors, including:
Some of the authors of this publication are also working on these related projects:
New Elsevier Book on Classical and Recent Aspects of Power System Optimization - Call for Chapters View project
All content following this page was uploaded by Ahmed Saif on 23 September 2020.
To cite this article: Hatem H. Zeineldin , Ahmed Saif , Magdy M. A. Salama & A. F. Zobaa (2010)
Three-dimensional Non-detection Zone for Assessing Anti-islanding Detection Schemes, Electric
Power Components and Systems, 38:6, 621-636, DOI: 10.1080/15325000903489645
1. Introduction
Islanding detection is an essential component of the protection system when integrating
distributed generation (DG) to distribution networks. The current IEEE 1547 and UL 1741
standards necessitate that a DG should be isolated during an islanding event [1, 2]. The
621
622 H. H. Zeineldin et al.
Nomenclature
C load capacitance (F)
Cnorm normalized capacitance
cfo SFS chopping fraction parameter
fo load resonance frequency (Hz)
id DG d -axis current component
iq DG q-axis current component
Irated DG rated output current (p.u.)
k SFS gain parameter
L load inductance (H)
P load active power (MW)
PDG DG active power (MW)
Po rated load active power (MW)
QC load capacitive reactive power (Mvar)
QDG DG reactive power (Mvar)
Qf load quality factor
QL load inductive reactive power (Mvar)
Qo rated load reactive power (Mvar)
V operating voltage (p.u.)
Vo nominal voltage (p.u.)
P active power mismatch (MW)
Q reactive power mismatch (Mvar)
‚Load load phase angle
DG DG phase angle
!o nominal frequency (rad/sec)
islanding detection method could be integrated either with the DG interface control,
as in the case of active islanding detection methods, or by adding relays to monitor
a certain system parameter, as in the case of passive islanding detection methods. A
comprehensive and detailed survey on the different types of islanding detection methods
proposed in the literature could be found in [3]. Some of the methods highlighted
in this survey include over-frequency protection/under-frequency protection (OFP/UFP)
and over-voltage protection/under-voltage protection (OVP/UVP), active frequency drift
(AFD), Sandia frequency shift (SFS), and phase jump detection (PJD) methods. In
addition, recent efforts in developing more efficient and accurate islanding detection
methods could be found in [4–6]. In [4–7], the proposed methods rely on injecting
disturbances through the inverter interface control to detect islanding. In [8, 9], in order
to reduce the non-detection zone (NDZ), the proposed islanding detection methods rely on
more than one parameter to detect islanding. In [10], a new islanding detection method
is proposed that is based on proportional power spectral density. The low-frequency
elements in the voltage period of the inverter can quickly be detected by computing
proportional energy over a sliding data window. In [11, 12], the proposed techniques are
based on equipping the DG with a droop curve that forces the DG to become unstable
during an islanding condition. In [13], a power line signaling based scheme is proposed
and tested experimentally for islanding detection.
Three-dimensional Non-detection Zone for Islanding 623
Islanding detection methods are evaluated and assessed using the NDZ concept. The
NDZ defines the loading conditions for which an islanding detection method would fail to
detect an islanding condition in a timely manner. NDZs have three main representations,
including P -Q load space, Cnorm -L load space, and Qf -fo load space.
The P -Q load space representation defines the NDZ in terms of the active
and reactive power mismatch between the DG power output and islanded load demand.
In [14], the OFP/UFP and PJD methods were evaluated using the P -Q load space,
and the dominant factors that could affect the islanding detection method were highlighted
through the NDZ analysis. The main drawback of this approach is its reliance on the
load quality factor.
The Cnorm -L load space relies on representing the NDZ in terms of a normalized
capacitance (which is the ratio between the load’s capacitance and the resonant capac-
itance at grid frequency) and the load inductance [15]. This representation was used in
determining the NDZ of the AFD, SFS, and OFP/UFP islanding detection methods. As
will be highlighted later, one major disadvantage of this approach is its reliance on the
load resistance.
The Qf -fo load space for representing the NDZ of islanding detection methods was
recently proposed in [16]. This representation relies on representing the NDZ in terms of
the load’s resonance frequency (fo ) and quality factor (Qf ). The OFP/UFP, AFD, and
SFS islanding detection methods were assessed using the Qf -fo load space, and it was
concluded that this approach provides a better representation of the NDZ.
This article first analyzes and compares the different types of NDZ representations
in order to determine the effectiveness of each approach in accurately representing the
NDZ of frequency drift islanding detection methods. The analysis is conducted on three
frequency drift islanding detection methods, including the OFP/UFP, AFD, and SFS
methods. Second, based on the results of the above analysis, a three-dimensional NDZ is
proposed to overcome the shortcomings of current two-dimensional NDZ representations.
The article is organized as follows. Section 2 focuses on the different types of NDZs
and presents the mathematical derivation. Section 3 presents the different NDZ models
applied to the different islanding detection methods. Section 4 provides simulation results
to verify the mathematical derivations. Section 5 introduces the three-dimensional NDZ
concept, and finally, conclusions are drawn in Section 6.
2. NDZ Representation
As highlighted earlier, there are three possible ways to represent the NDZ of an is-
landing detection method. The representations presented in this section are considered
as two-dimensional representations, since the NDZ is quantified based on two load
parameters. The next subsections will focus on describing the three common NDZ
representations. The NDZ of the P -Q presented in this article is reformulated to
include the active power frequency dependence, which has not been addressed in previous
literature. The equations derived will be used for developing the three-dimensional NDZ
model, as will be highlighted later.
3V 2
3V Irated sin DG D QL QC D 3V 2 !C: (9)
!L
Table 1
DG phase angle for the different
islanding detection methods
OFP/UFP DG D 0
AFD DG D cfo
2
SFS DG D .cfo C k.f fo /
2
Three-dimensional Non-detection Zone for Islanding 625
By substituting the values of C and L in Eqs. (2) and (3), Eq. (9) can be rewritten
as follows:
2
!0 ! V
V Irated sin DG D .Qo C Q/ Qo : (10)
! !o Vo
By substituting Eq. (7) into Eq. (11) and rearranging, the reactive power mismatch
is calculated from Eq. (12) as follows:
!2
! P Q Q0 !
tan DG D 1 tan DG : (12)
!o Po Po Po !o2 !o
Equations (7) and (12) represent the main equations for determining the NDZ for
different islanding detection methods in the P -Q space.
The basic relationship that governs the structure of this type of NDZ could be
determined by equating the load phase angle (‚Load ) and the DG phase angle (DG ) as
follows:
1
tan DG D tan Load D R !C : (14)
!L
By substituting Eq. (13) into Eq. (14), the main equation for calculating the NDZ
could be presented in Eq. (15). It can be seen that by drawing the NDZ in the L-Cnorm
plane, the NDZ will be dependent on the load resistance:
R 1 Cnorm
tan DG D ! 2 : (15)
L ! !o
with the P -Q NDZ model. The main equation for calculating the NDZ is derived in
[16] and is given below:
One of the main problems with this type of NDZ model is that it does not take into
account the value of load resistance. This issue will be highlighted in later sections.
Figure 2. Point of common coupling (PCC) frequency with the OFP/UFP islanding detection
method with Q=Po kept fixed at 10%.
628 H. H. Zeineldin et al.
Figure 3. PCC frequency with the SFS islanding detection method for various values of L.
Figure 4. PCC frequency with the SFS islanding detection method for various values of R.
Figure 5. PCC frequency and voltage with the AFD islanding detection method for two different
active power mismatch conditions.
629
630 H. H. Zeineldin et al.
Figure 6. PCC frequency and voltage with the OFP/UFP islanding detection method.
threshold, and thus, this case is detectable. Although for both loading conditions, the
value of Qf and fo are the same, one case is detectable and the other is not. Thus,
the main drawback of this NDZ model is that it does not take into account the active
power mismatch.
An additional finding, which has not been addressed before in the literature, is
the dependence of the voltage deviation on the amount of frequency deviation for the
case where the DG is equipped with the SFS method. This was highlighted in Eq. (7).
Figures 6, 7, and 8 present the PCC frequency and voltage for an inverter-based DG
equipped with the OFP/UFP, AFD, and SFS islanding detection methods, respectively, for
a load with Qf D 7. The simulation results present the frequency and voltage for two dif-
ferent values of reactive power mismatch and with the active power mismatch set to zero.
It can be seen from both Figures 6 and 7 that for different active and reactive
power mismatches, the voltage will deviate and stabilize at the same value. Thus, for
the OFP/UFP and AFD methods, the voltage deviation is independent on the amount of
frequency deviation. For the SFS method, although the two loading conditions presented
have the same active power mismatch, the voltage stabilizes at two different values. This
is due to the dependence of voltage deviation on the amount of frequency deviation.
5. Three-dimensional NDZ
It can be seen from the previous section that each NDZ model suffers from a major draw-
back. For the P -Q NDZ representation, the active and reactive power mismatches
are not sufficient to determine whether a loading condition lies within the NDZ or not.
Figure 7. PCC frequency and voltage with the AFD islanding detection method with cfo D 0:05.
Figure 8. PCC frequency and voltage with the SFS islanding detection method with cfo D 0 and
k D 0:05.
631
632 H. H. Zeineldin et al.
The NDZ changes with the change in the Qo =Po ratio, and this parameter needs to be
taken into consideration. Thus, the P -Q NDZ model requires three parameters to
accurately model the NDZ under different loading conditions.
For the L-Cnorm representation, the NDZ region changes with the change in load
resistance. Thus, it is difficult to develop one NDZ model that takes into account all
loading conditions. This type of representation requires three parameters, which are
the load inductance normalized capacitance and load resistance to model the NDZ
accurately under different loading conditions.
The third representation, which is the Qf -fo representation, does not take into
account the active power mismatch or, in other words, the load resistance. For the same
quality factor and resonance frequency, a loading condition might be within the NDZ or
outside the NDZ, depending on the load resistive value. For accurate NDZ representation,
this model would need to take into account the load resistance, which makes it dependent
on three parameters.
It can be concluded that for accurate modeling of the NDZ, three load parameters
need to be taken into consideration. This could easily be related to the fact that the load has
three variables—the load resistance, inductance, and capacitance. The three parameters
should be chosen such that all possible loading conditions are included and well defined.
For example, for an RLC load, knowing the active and reactive power of the load is not
sufficient to calculate the load resistance, inductance, and capacitance. With an additional
parameter (Qo =Po ), the load parameters (resistance, inductance, and capacitance) could
be calculated. This also applies to the two other NDZ models presented in this article.
In order to overcome the drawbacks of previously proposed NDZ models, a three-
dimensional NDZ model is developed that relies on three parameters—P =Po , Q=Po ,
and Qo =Po . Figures 9 and 10 present the proposed NDZ model for the OFP/UFP versus
the SFS and AFD islanding detection methods, respectively. The three-dimensional NDZ
model provides a better visualization of the undetectable loading conditions. It can be
seen from Figure 9 that the SFS NDZ lies inside the OFP/UFP NDZ, thus indicating that
the SFS method has a smaller NDZ. It can also be concluded from Figure 10 that the
NDZ of the AFD method is a shifted version of the NDZ of the OFP/UFP method, thus
having no significant improvement over the OFP/UFP islanding detection method. The
three-dimensional NDZ model can clearly show the borderlines of the NDZ.
The proposed three-dimensional NDZ was tested under various loading conditions.
Load A (Qo =Po D 0:1), Load B (Qo =Po D 3:5), and Load C (Qo =Po D 7) in Figure 9
are three loading conditions with different Qo =Po but corresponding to the same active
power mismatch (P =Po D 0) and reactive power mismatch (Q=Po D 0:05).
Figure 11 presents the PCC frequency during an islanding condition for the three loading
conditions with the OFP/UFP and SFS islanding detection methods, respectively. For
Load A, the frequency drifts beyond the frequency thresholds (Figure 11), and thus, both
the OFP/UFP and SFS methods would succeed in detecting an islanding condition. For
Load B, the amount of frequency drift will depend on the type of islanding detection
method. The OFP/UFP method will not be capable of detecting this loading condition
since the frequency stabilizes at a value within the frequency thresholds. On the other
hand, the SFS islanding detection method can successfully detect this loading condition,
and thus, this loading condition is outside its NDZ (Figure 9). Finally, both the OFP/UFP
and SFS islanding detection methods would fail to detect an islanding condition with
Load C. This can also be seen from Figure 9, where Load C is within the NDZ of both
islanding detection methods. Figure 12 presents the simulated versus the analytical NDZ.
The simulated NDZ closely matches the analytical results.
Figure 9. Three-dimensional NDZ representation for the SFS method with cfo D 0 and k D 0:05
versus the OFP/UFP method.
Figure 10. Three-dimensional NDZ representation for the AFD method with cfo D 0:05 versus
the OFP/UFP method.
633
634 H. H. Zeineldin et al.
Figure 11. PCC Frequency for various loading conditions with the OFP/UFP and SFS methods,
respectively.
6. Conclusions
This article proposes a new three-dimensional NDZ model for assessing frequency drift
islanding detection methods to overcome the drawbacks of previously proposed NDZ
models. The proposed three-dimensional NDZ model relies on three parameters that
include the active and reactive power mismatch and the load quality factor (Qo =Po /.
The parameters have been chosen in such a way to cover all possible RLC loading
conditions. The three main two-dimensional NDZ models (P -Q, L-Cnorm, and Qf -
fo ) are modeled for the OFP/UFP, AFD, and SFS islanding detection methods, and the
main drawback of each NDZ model was highlighted through mathematical analysis and
confirmed through simulation results. The common drawback among two-dimensional
NDZ models is the inadequacy of relying on two parameters for modeling NDZs.
Although the Qf -fo NDZ model has been proposed in the literature to overcome the
drawbacks of previously proposed NDZ representations, it also suffers from its reliance
on the load resistance or, in other words, the active power mismatch. The proposed
three-dimensional NDZ model can accurately model all possible loading conditions and
provides a better representation for comparing different frequency drift islanding detection
methods.
References
1. IEEE, “IEEE standard for interconnecting distributed resources with electric power systems,”
IEEE Std. 1547-2003, July 2003.
2. Underwriters Laboratories Inc., Northbrook, IL, “Static inverter and charge controllers for use
in photovoltaic systems,” Standard UL 1741, 1999.
3. Ropp, M., and Bower, W., “Evaluation of islanding detection methods for photovoltaic utility
interactive power systems,” International Energy Agency Implementing Agreement on Photo-
voltaic Power Systems, Tech. Rep. IEA PVPS T5-09, March 2002.
4. Karimi, H., Yazdani, A., and Iravani, R., “Negative sequence current injection for fast islanding
detection of a distributed resource unit,” IEEE Trans. Power Electron., Vol. 23, No. 1, pp. 298–
307, January 2008.
5. Hernandez-Gonzalez, G., and Iravani, R., “Current injection for active islanding detection of
electronically-interfaced distributed resources,” IEEE Trans. Power Delivery, Vol. 21, No. 3,
pp. 1698–1705, July 2006.
6. Xiaoyu, W., Freitas, W., Dinavahi, W., and Xu, W., “Investigation of positive feedback anti-
islanding control for multiple inverter based distributed generators,” IEEE Trans. Power Syst.,
Vol. 24, No. 2, pp. 785–795, May 2009.
7. Jou, H., Chiang, W., and Wu, J., “A simplified control method for grid-connected inverter with
the function of islanding detection,” IEEE Trans. Power Electron., Vol. 23, No. 6, pp. 2775–
2783, November 2008.
8. Jang, S., and Kim, K., “An islanding detection method for distributed generations using voltage
unbalance and total harmonic distortion in current,” IEEE Trans. Power Delivery, Vol. 19,
pp. 745–752, April 2004.
9. Mahat, P., Zhe, C., and Bak-Jensen, B., “A hybrid islanding detection technique using average
rate of voltage change and real power shift,” IEEE Trans. Power Delivery, Vol. 24, No. 2,
pp. 764–771, April 2009.
10. Jun, Y., Diduch, C. P., and Liucheng, C., “Islanding detection using proportional power spectral
density,” IEEE Trans. Power Delivery, Vol. 23, No. 2, pp. 776–784, April 2008.
11. Zeineldin, H. H., and Kirtley, J., “A simple technique for islanding detection with negligible
nondetection zone,” IEEE Trans. Power Delivery, Vol. 24, No. 2, pp. 779–786, April 2009.
12. Zeineldin, H., “A q-f droop curve for facilitating islanding detection of inverter based dis-
tributed generation,” IEEE Trans. Power Electron., Vol. 24, pp. 665–673, March 2008.
636 H. H. Zeineldin et al.
13. Wencong, W., Kilber, J., and Wilson, X., “A scalable power line signaling based scheme for
islanding detection of distributed generators,” IEEE Trans. Power Delivery, Vol. 24, No. 2,
pp. 903–909, April 2009.
14. Ye, Z., Kolwalkar, A., Zhang, Y., Du, P., and Walling, R., “Evaluation of anti-islanding schemes
based on nondetection zone concept,” IEEE Trans. Power Electronics, Vol. 19, pp. 1171–1176,
September 2004.
15. Ropp, M. E., Begovic, M., Rohatgi, A., Kern, G. A., Bonn, R. H., and Gonzalez, S., “De-
termining the relative effectiveness of islanding detection methods using phase criteria and
nondetection zones,” IEEE Trans. Energy Conversion, Vol. 15, pp. 290–296, September 2000.
16. Lopes, L., and Sun, H., “Performance assessment of active frequency drifting islanding detec-
tion methods,” IEEE Trans. Energy Conversion, Vol. 21, pp. 171–180, March 2006.
17. Woyte, A., Belmans, R., and Nijs, J., “Testing the islanding protection function of photovoltaic
inverters,” IEEE Trans. Energy Conversion, Vol. 18, pp. 157–162, March 2003.
18. Wang, X., Freitas, W., Xu, W., and Dinavahi, V., “Impact of DG interface controls on the Sandia
frequency shift anti-islanding method,” IEEE Trans. Energy Conversion, Vol. 22, pp. 792–794,
September 2007.