Professional Documents
Culture Documents
18 - An Offline Penet Free Prot SCH For PV-dominated DSs - ETAP
18 - An Offline Penet Free Prot SCH For PV-dominated DSs - ETAP
18 - An Offline Penet Free Prot SCH For PV-dominated DSs - ETAP
a r t i c l e i n f o a b s t r a c t
Article history: The rapid growth of small-scale photovoltaic (PV) units in distribution grids brings new challenges in the
Received 27 May 2017 protection coordination of overcurrent relays. So far, various adaptive techniques have been addressed to
Received in revised form solve DG units’ mis-coordination problem, however, the un-predictable penetration of installed PV units
26 November 2017
highlights the importance of a penetration-free methodology. This study presents a novel offline, quick,
Accepted 29 November 2017
Available online 6 December 2017
applicable and cheap solution which guarantees protection coordination for any penetration and location
of PV units. To do so, first, under different PV penetration levels and locations, the conventional protection
performance is studied to discover the worst mis-coordination cases. Next, according to relays standards,
Keywords:
Overcurrent relay the characteristic curve of the back-up relay is modified such that it can maintain coordination in all
Coordination-saving worst cases. The proposed methodology is applied to a practical power distribution network equipped
Photovoltaic (PV) systems with numerous small-scale PV systems. The results successfully prove the effectiveness of the proposed
Penetration-free method. Since this technique only applies some changes in characteristic curves of the back-up relays,
Offline protection plan lack of programmable relay or adaptive protection requirements cannot decrease its performance. This
fact makes this method an attractive option for distribution systems with PV units.
© 2017 Elsevier B.V. All rights reserved.
https://doi.org/10.1016/j.epsr.2017.11.020
0378-7796/© 2017 Elsevier B.V. All rights reserved.
2 B. Fani et al. / Electric Power Systems Research 157 (2018) 1–9
This method uses an optimization algorithm to find the best easily applied on different protection relays, including traditional
places for installing new DG units such that there is no conflict un-programmable ones.
in traditional protection plans. These kinds of solutions are cheap The key contributions of this paper compared to the previous
and also free of protection adaption, however, they would limit the works are as follows:
penetration level of DG units sooner than others with protection
adaption. Moreover, the distributed nature of small PV systems is • The methodology has a penetration and location free nature.
not compatible with this methodology which considers DG units as • The proposed method is offline.
concentrated generations in different nodes of the network. • There is no need to any new investment, including replacement,
installing new devices, designing new online control functions.
2 Limiting the penetration level of DG units in parallel to adapting • The procedure can be applied on different kinds of old or new-
their control functions [16,17] generation relays.
Although this method can effectively resolve protection prob- The succeeding sections of this paper are organized as follows:
lems, limiting penetration is not a good idea against the significant Section 2 studies the main concept of increasing DG units’ pen-
growth of PV installation in distribution networks. Moreover, this etration in a typical distribution feeder. The proposed algorithm
solution will increase the complexity of network design and con- for recovering protection coordination is presented in Section 3. In
trol. Section 4, the three-phase, two-phase, two-phase to ground and
single-phase high-impedance faults are simulated on a practical
3 Deploying communication links in protection structure [18–21] distribution feeder for both initial protection plan and the pro-
posed algorithm to verify the effectiveness of the proposed method.
Detecting the fault, this methodology will urgently send the Finally, conclusion and future works are given in Section 5.
appropriate orders to different equipment to adapt control func-
tions against fault condition. This methodology can be used either in 2. Protection system performance in presence of PV-based
wide-area protection systems [18,19] or multi-agent ones [20,21]. DG units
However, in addition to extra expenses, the system reliability may
be affected due to communication maloperations. The fault current contribution of DG units can considerably
affect the performance of the protection system in distribution
4 Disconnecting DG units during fault condition [22] feeders. This contribution which can lose or even improve the pro-
tection coordination is mainly dependent on the three following
This technique strategically disconnects some DG units from specifications of DG units:
distribution network to reduce their contributions in fault current.
Although this method will omit DG units’ fault contributions, there • Technology:
are some disadvantages to DG units’ disconnection. For instance,
because of the timing limits for disconnecting DGs, it is not pos-
DG units’ technology can effect on the fault current contribu-
sible to use fast curves of relays or fuses. In addition, the power
tion; for instance, the PV-based DG units have a limited output
quality indices may become worse or the efficiency of DG units
current due to the control function used in the inverters [5], while
may decrease due to their frequent outages.
synchronous machine DG units have larger output current.
Fig. 2. The main and back-up relays’ characteristic curves and the line thermal limit curve.
profile will affect this controller and consequently change the But, if some PV-based DG units are supposed to be installed in
penetration of PV systems. this network, the coordination is not only dependent on the pene-
tration but also the location of DG units. Thus, three following cases
are studied on this sample network.
Meanwhile, installing PV units upstream of PD1 has a different IPD2 = (VS + IPV 21 (ZT − Z21 − ZF )) /ZT (3)
effect on coordination. Although PV units equally increase the flow-
ing current through PD1 and PD2, the coordination may be lost in where, IPV21 represents the PV system injection current, Z21 denotes
some situations. the impedance between PV21 and PD2 relay, and ZT and ZF are
Assume Fig. 3 shows the characteristic curves of PD1 and PD2 the main source and PD2 relay impedances to the location of fault,
relays. respectively.
As shown in this figure, the increase in penetration of PV units According to Eqs. (2) and (3), depending upon the capacity (IPV21 )
causes fault current exceed IFmax . As a result, the minimum required and location (Z21 ) of PV21 system, it will reinforce and reduce the
timing margin between two main and backup curves is violated and fault and upstream currents, respectively. Indeed, the increase in
the coordination is lost. PD2 current will decrease the operation time of main protection.
In sum, this example shows that the protection coordination On the other hand, with the decrease in PD1 current, the operation
accuracy depends on the location and capacity of PV units. Thus, it is time of back-up relay will increase. As a result, the CTI between PD1
necessary to present a method for making protection coordination and PD2 may increase such that it threatens the coordination plan.
independent of PV units’ characteristics. To this end, this paper pro- In general, for a large number of PV systems between PD1 and
poses a novel algorithm for modifying relay characteristic curves to PD2 (n = the number of middle PV systems and m = 0), Eqs. (2) and
reach an accurate performance for different penetration and loca- (3) can be generalized as the following equations.
tion of PV units in the distribution system. According to the fact that
the presented quick, applicable and cheap solution only needs some
n
j
changes in characteristic curves, lack of programmable relay or IPD1 = VS /ZT − IPV 2j (Z2i + ZF )/ZT (4)
adaptive protection requirements cannot decrease its performance. j=1 i=1
B. Fani et al. / Electric Power Systems Research 157 (2018) 1–9 5
350 ms), will cause the coordination be lost. To treat this misco-
ordination, the PD1 curve should be modified such that it has an
acceptable CTI at Imax−PV . Hence, the operation time of PD1 (tmin )
at the 100% penetration level should be equal to the sum of PD2
operation time and the minimum allowable CTI as shown in Eq.
(7).
In the last step, limit state ‘D’ should also be considered. Never-
theless, since upstream and middle PV units counteract each other’s
effect, state ‘D’ would not yield to new miscoordination problems.
As discussed in the previous part, two limit states of ‘B’ and ‘C’
change two limit points of the PD1 curve to [Imax−PV , tmin ] and
[Imin−PV , tmax ] which are illustrated in Fig. 4 with points b and
Fig. 4. Proposed modified curve for PD1 relay. a, respectively. According to the common characteristic curves of
relays, these two points can accurately specify the modified curve.
n
n
j For instance, assume common IEC 60255 standard as a sample
IPD2 = VS /ZT + IPV 2j − IPV 2j (Z2i + ZF )/ZT (5) standard for defining characteristic curves of relays. This standard
expresses the relay operation time as a function of fault current
j=1 j=1 i=1
with the following equation [35].
where, n is the number of PV systems, IPV2j is the injected cur- P
rent of jth PV system and Z2i denotes the impedance between two tPD1 = A × T · D/ (IPD1 /Ipickup ) − 1 (8)
consecutive PV systems (as shown in Fig. 1).
As the amount of decrease in IPD1 may cause the relay operating where, tPD1 denotes the operation time of the relay, T.D is the time
time exceed the maximum allowable thermal limit of conductors, setting coefficient, Ipickup represents the current setting threshold
it is possible to consider PD1 operating time (tPD1 ) as a criterion for and A and P are two constant coefficients related to the curve slope
situation of protection coordination. specification. According to the fact that Ipickup and T.D are depen-
According to Fig. 2, when there is no PV in the system, the dent on the current passing through the relay, they would vary with
downstream fault makes the main source to produce IS current and penetration of PV systems, that is, they are not suitable parameters
the PD2 relay to operate at tPD2 . Next, if there is an unsuccessful for modifying the curve and achieving a penetration-free algorithm.
operation, passing the minimum time interval required between On the other hand, A and P are two independent constants in Eq. (8)
operation of primary and backup relay (CTI), PD1 should operate at which can be easily used to attain the goal. Therefore, the desired
tPD1 to prevent any thermal limit violations. A and P are calculated based on two following equations made
According to this figure, at 100% penetration level of PV systems, according to the specifications of points a and b.
IPD1 decreases to Imin−PV and IPD2 increases to Imax−PV . In such a situ-
ation, although the CTI (t PD1 − t PD2 ) increases, the minimum time logA×T.D+t
Imin −PV
max
− logA×T.D+t
Imax −PV
min = logtmax tmin
Imin −PV − logImax −PV (9)
for clearing fault decreases to tmc . In other word, the PD1 should
operate sooner than tmc , whereas this relay operates at t PD1 which P = logA×T.D/t
Imin −PV
max +1
(10)
is much longer than tmc . Given that PD1 curve causes this misco-
ordination, modifying PD1 characteristic curve is the best solution Calculating the desired A and P from Eqs. (9) and (10), it is possi-
for solving this problem. ble to modify the PD1 curve just once to be completely sure about
Usually, the allowable threshold for operation of backup relay is coordination with any penetration of PV systems. In other words,
considered 1000 ms. That is, both PD1 and PD2 characteristic curves there is no need to any adaptive control or change in protection
should be below the thermal limit curve. Therefore, after consider- settings in response to variable penetrations and locations of PV
ing state ‘B’ as the worst case (Fig. 4) which changes IPD1 to Imin−PV , systems.
it is necessary to modify PD1 curve such that t PD1 transfers to a Finally, the resulted coefficients can be rounded to the nearest
point under thermal limit curve like tmax . five predefined A and P of Table 1 to find the best-fitting IEC curve
In the third step, it is necessary to consider the limit state for PD1 relay.
‘C’ where all PV systems are installed upstream of PD1 (n = 0 and
m = the number of upstream PV systems). In this situation, the cur- 3.3. Proposed algorithm procedure
rent passing through PD1 and PD2 are calculated as Eq. (6).
This procedure is triggered via a significant change in network
m
j
IPD1,2 = VS /ZT + IPV 1j (Z1i + ZF )/ZT (6) layout. Thus, after every change, the back-up relay curves should be
modified offline according to two limit states of ‘B’ and ‘C’ following
j=1 i=1
the procedure outlines in Fig. 5.
where, m is the number of upstream PV systems and Z1i represents As shown in this figure, first, corresponding to each pair of main
the impedance between two consecutive PV systems. and back-up relays, two limit states ‘B’ and ‘C’ are simulated to
Considering Eq. (6), the increase in penetration level of PV sys- verify the coordination. If the coordination is lost, the back-up relay
tems will increase both IPD1 and IPD2 and it will thereupon force curve is modified according to the method proposed in Section 3.1,
the fault current to exceed IFmax at limit state ‘C’. Thus, the resulted else, next pair of relays is examined. This process goes on till all
CTI, which is less than the minimum allowable threshold (usually relays are verified.
6 B. Fani et al. / Electric Power Systems Research 157 (2018) 1–9
Table 1
A and P coefficient according to IEC 60255 [35].
Short time inverse Normal inverse Very inverse Extremely inverse Long time inverse
Table 2
Summarized characteristics of the studied system.
Description Subject
Length of feeder 30 km
Feeder type Head way type & not transposed —
radial
Conductor size MVa line = 120 mm2 –LVb
line = cable 4 × 50 + 25 mm2
Line shape Horizontal & distance between
lines = 70, 140, 70 cm
Nominal voltage MV = 20 kVLL , LV = 400 VLL ,
3ph + N + PE
Legs altitude 9m
Transformer 630 kVA, /y grounded –
20 kV/0.4 kV
MVASC of main substation 500 MVA
PV inverter 200 kVA & tied grid without
battery for saving energy
Level of load unit 200 kVA
a
Medium voltage.
b
Low voltage.
Table 3
The PD1 and PD2 characteristics.
relation (8) is used to specify the operation time of PD1 (t PD1 ).
To make sure of correct back-up relay operation, a safety mar-
gin of 200 ms is considered between PD1 and thermal limit curves.
Therefore, the maximum allowable time for PD1 operation at 100%
Fig. 7. The initial protection plan in presence of PV systems between PD1 and PD2. penetration of middle PV systems (tmax ) is 800 ms. That is, the back-
up relay for a 1390 A fault current will operate at 800 ms instead of
1880 ms.
According to these initial input data, two case studies are The upstream PV systems will increase the fault current passing
designed to first validate the proposed method as follows: through PD1 and PD2. For instance, assume the penetration level is
10%. The fault current will increase to 1910 A. PD2 and PD1 relays
4.1. Relay coordination for zero-impedance faults will operate at 255 and 563 ms, respectively. As mentioned, the
penetration level would have a reverse relation with CTI. This fact is
Since the worst miscoordination usually happens when the shown in Table 5 for different penetration of upstream PV systems.
largest fault current is passing through relays, in this case study, As shown, although 10% penetration has decreased CTI, the PD1
the fault impedance (ZF ) is assumed zero for relations (4)–(6). relay is still successfully coordinated with PD2. However, the coor-
Assuming no PV system is installed in the network, the max- dination will be lost when the CTI becomes less than 250 ms with
imum current passing through PD2 will occur when there is a 30% penetration of PV units. This trend will continue to reach the
three-phase fault at F point. In this situation, the fault current is least CTI with 144 ms period at 100% penetration as shown in Fig. 8.
1860 A and PD2 will respond to it within 259 ms. According to the Given that the least unacceptable CTI happens at 100% pene-
fact that the passing fault currents through PD1 and PD2 are equal, tration, it is rational to modify PD1 curve for this point to reach a
the back-up relay will operate in 609 ms which is much higher than proper coordination for any penetration of upstream PV systems.
350 ms. To do so, first, according to relation (8) the operation time of PD2 is
Installing PV units between PD1 and PD2 will change the current obtained, next, 350 ms as the minimum allowable CTI is added to
passing through relays. Assume the penetration level is 10 percent, it to calculate PD1 minimum timing and create tmin . This modifica-
the fault current at F point will be 1946 A at PD2 and 1798 A at tion will change the PD1 operation time from 144 ms to 350 ms for
PD1. As a result, the main and back-up relays will operate at 253 100% upstream PV penetration with a 2188 A fault current. Draw-
and 676 ms, respectively. It is seen that the CTI is increased with ing a new curve by connecting the tmin to the tmax points will create
respect to zero penetration to have a more confident coordination. the modified curve for PD1 relay. The new curve is coordinated with
Table 4 shows CTI values for different penetrations of PV systems PD2 for any penetration and location of PV systems. Furthermore,
of Fig. 6. it is completely below the thermal limit curve with a predefined
As shown in this table, when the penetration exceeds 50%, PD2 safety margin.
operates later than tmc . For instance, for 60% penetration, PD1 acts Finally, the new curve should be fitted to a standard curve. For
within 1163 ms which is not acceptable for coordination. The worst this reason, when the range of changes in fault current and oper-
case happens when the back-up relay acts within 1880 ms for 100% ation time for PD1 relay is obtained according to the relations (9)
penetration. Fig. 7 illustrates the initial protection plan of PD1 and and (10), it is possible to calculate all unknown parameters of the
PD2. modified characteristic curve. Nevertheless, due to the changing
This figure shows the impact of the middle PV systems on relays configuration of PV systems in distribution networks, it is better
coordination. When the penetration exceeds 50%, the initial plans to fix two parameters like T.D and Ipickup to their initial values,
loses its coordination. In such a condition, first of all, the current which are 0.03 and 1.1 respectively. To avoid repetitive setting for
passing through PD1 is calculated according to relation (5). Next, these parameters, only two parameters as A and P should change to
8 B. Fani et al. / Electric Power Systems Research 157 (2018) 1–9
Table 4
The simulated operation time of PD1 in limit state ‘B’.
3 L-L-G L-L
Table 5
The simulated CTI for different fault types in limit state ‘C’.
3 L-L-G L-L
reach the new curve. As a result, for this case study the A and P are
calculated which are compatible with “short time- inverse” curve.
The validity of the proposed method for different penetrations
of middle and upstream PV systems is presented in Tables 4 and 5,
respectively. Fig. 9 also illustrates the impact of PV systems on oper-
ation times of PD1 and PD2 relays when the proposed algorithm is
applied on the network.
As shown in this figure, none of limit states, presented in Section
3, could cause any miscoordination for PD1 relay. Therefore, other
probable penetration and locations of PV systems would rationally
have a proper coordinated protection plan.
Table 6 [13] Dheeraj K. Khatod, Vinay Pant, Jaydev Sharma, Evolutionary programming
Proposed method assessment during a HIF condition (a 20 L-G fault). based optimal placement of renewable distributed generators, IEEE Trans.
Power Syst. 28 (2) (2013).
PV level (%) Middle-stream Upstream [14] Hongxia Zhan, Caisheng Wang, Yang Wang, Xiaohua Yang, Xi Zhang,
Changjiang Wu, Yihuai Chen, Relay protection coordination integrated
tPD1 tPD2 t
optimal placement and sizing of distributed generation sources in
Conventional Proposed Conventional Proposed distribution networks, IEEE Trans. Smart Grid 7 (2016) 55–65.
[15] Dhivya Sampath Kumar, Dipti Srinivasan, Thomas Reindl, A fast and scalable
0 685 693 307 313 339 protection scheme for distributed network with distributed generation, IEEE
10 778 712 306 296 336 Trans. Power Deliv. 31 (2) (2016) 67–75.
20 886 733 305 280 332 [16] Iman Sadeghkhani, Mohamad Esmail Hamedani Golshan, Josep M. Guerrero,
30 1008 751 303 265 330 Ali Mehrizi-Sani, A current limiting strategy to improve fault ride-through of
40 1137 767 302 251 327 inverter interfaced autonomous microgrids, IEEE Trans. Smart Grid 99 (2016)
50 1287 786 301 237 325 1–11.
60 1461 804 300 225 322 [17] Konstantinos O. Oureilidis, Charis S. Demoulias, A fault clearing method in
70 1666 826 299 213 319 converter-dominated microgrids with conventional protection means, IEEE
Trans. Smart Grid 31 (6) (2016) 4628–4640.
80 1914 845 298 203 317
[18] Vassilis C. Nikolaidis, Evangelos Papanikolaou, Anastasia S. Safigianni, A
90 2221 867 297 193 316
communication-assisted overcurrent protection scheme for radial
100 2610 886 296 186 315 distribution systems with distributed generation, IEEE Trans. Smart Grid 7
(2016) 114–123.
[19] Edmund O. Schweitzer, Dale Finney, Mangapathirao V. Mynam,
PV systems. Thus, this work presented an offline methodology to Communications-assisted schemes for distributed generation protection, IEEE
save overcurrent relay coordination against any penetration of PV Conference PES T&D (2012).
[20] Z. Liu, C. Su, H.K. Høidalen, Z. Chen, A multiagent system-based protection and
systems in the distribution system. To do so, first, the limit states, control scheme for distribution system with distributed-generation
where the coordination has the worst situation, are determined. integration, IEEE Trans. Power Deliv. 32 (2017) 536–545.
Next, the back-up relay characteristic curve is modified such that it [21] P.C. Maiola, J.G. Rolim, A multi-agent system for protection coordination of
radial systems in the presence of distributed generation, 11th IET
has a proper coordination in these states. Finally, the resulted curve International Conference on Developments in Power Systems Protection
is fitted to the nearest standard curve (according to IEC 60255) (2012).
to be applied on different kinds of relays. Simulation results on [22] Rajiv K. Varma, Shah Arifur Rahman, Vishwajitsinh Atodaria, Sibin Mohan,
Tim Vanderheide, Technique for fast detection of short circuit current in PV
a practical feeder successfully proved the effectiveness of the pro-
distributed generator, IEEE Power Energy Technol. Syst. J. 3 (4) (2016)
posed method. The approach used in this work can be easily applied 155–165.
on the common-used conventional non-digital relays, making this [23] Abbas Saberi Noghabi, Habib Rajabi Mashhadi, Javad Sadeh, Optimal
coordination of directional overcurrent relays considering different network
method a cheap, quick, reliable and user-friendly technique.
topologies using interval linear programming, IEEE Trans. Power Deliv. 25 (3)
(2010) 1348–1354.
References [24] Mansour Ojaghi, Zeinab Sudi, Jawad Faiz, Implementation of full adaptive
technique to optimal coordination of overcurrent relays, IEEE Trans. Power
[1] D. Ranamuka, A.P. Agalgaonkar, K.M. Muttaqi, Examining the interactions Deliv. 28 (1) (2013) 235–244.
between DG units and voltage regulating devices for effective voltage control [25] Pragnesh H. Shah, Bhavesh R. Bhalja, New adaptive digital relaying scheme to
in distribution systems, IEEE Trans. Ind. Appl. 53 (2017) 1485–1496. tackle recloser–fuse miscoordination during distributed generation
[2] Kalpesh A. Joshi, Naran M. Pindoriya, Case-specificity and its implications in interconnections, IET Gener. Transm. Distrib. 8 (4) (2014) 682–688.
distribution network analysis with increasing penetration of photovoltaic [26] Min Cheol Ahn, Tae Kuk Ko, Proof-of-concept of a smart fault current
generation, CSEE J. Power Energy Syst. 3 (1) (2017). controller with a superconducting coil for the smart grid, IEEE Trans. Appl.
[3] Hossein Hooshyar, Mesut E. Baran, Fault analysis on distribution feeders with Supercond. 21 (3) (2011).
high penetration of PV systems, IEEE Trans. Power Syst. 28 (3) (2013) [27] Hyung-Chul Jo, Sung-Kwan Joo, Kisung Lee, Optimal placement of
2890–2896. superconducting fault current limiters (SFCLs) for protection of an electric
[4] H.H. Zeineldin, Y. Rady, I. Mohamed, V. Khadkikar, V.R. Pandi, A protection power system with distributed generations (DGs), IEEE Trans. Appl.
coordination index for evaluating distributed generation impacts on Supercond. 23 (3) (2013).
protection for meshed distribution systems, IEEE Trans. Smart Grid 4 (3) [28] Lukasz Huchel, Hatem H. Zeineldin, Planning the coordination of directional
(2013) 1523–1532. overcurrent relays for distribution systems considering DG, IEEE Trans. Smart
[5] Hesam Yazdanpanahi, Yun Wei Li, Wilsun Xu, A new control strategy to Grid 7 (3) (2016) 1642–1649.
mitigate the impact of inverter-based DGs on protection system, IEEE Trans. [29] Ke Jia, Chenjie Gu, Zhenwen Xuan, Lun Li, Y. Lin, Fault characteristics analysis
Smart Grid 3 (3) (2012) 1427–1436. and line protection design within a large-scale photovoltaic power plant, IEEE
[6] Jian Liu, Hsiao-Dong Chiang, Maximizing available deliv capability of Trans. Smart Grid (99) (2017).
unbalanced distribution networks for high penetration of distributed [30] Iman Sadeghkhani, Mohamad Esmail Hamedani Golshan, Ali Mehrizi-Sani,
generators, IEEE Trans. Power Deliv. 32 (3) (2017) 1196–1202. Josep M. Guerrero, Abbas Ketabi, Transient monitoring function-based fault
[7] S. Rahman, Green power: what is it and where can we find it? IEEE Power detection for inverter-interfaced microgrids, IEEE Trans. Smart Grid (99)
Energy Mag. 1 (1) (2003) 30–37. (2016).
[8] Mohammed A. Haj-ahmed, Mahesh S. Illindala, The influence of [31] IEEE standard for IEEE recommended practice for utility interface of
inverter-based DGs and their controllers on distribution network protection, photovoltaic (PV) systems IEEE standard 929. (2000).
IEEE Trans. Ind. Appl. 50 (4) (2014). [32] Chanaka Keerthisinghe, Gregor Verbic, Archie C. Chapman, A fast technique
[9] S.A. Saleh, A.S. Aljankawey, B. Alsayid, M.S. Abu-Khaizaran, Influences of for smart home management: ADP with temporal difference learning, IEEE
power electronic converters on voltage-current behaviors during faults in Trans. Smart Grid (99) (2016).
DGUs-part II: photovoltaic systems, IEEE Trans. Ind. Appl. 51 (4) (2015) [33] Michelangelo Ceci, Roberto Corizzo, Fabio Fumarola, Donato Malerba,
2832–2845. Aleksandra Rashkovska, Predictive modeling of PV energy production: how to
[10] Ahmed E.B. Abu-Elanien, Faculty of Engineering, Alexandria University, set up the learning task for a better prediction? IEEE Trans. Ind. Inf. (99)
Increasing the penetration level of distributed generation without violating (2016).
overcurrent based protection system coordination, IEEE PES Innovative Smart [34] IEC standard for short-circuit currents in three-phase a.c. systems, IEC
Grid Technologies Europe (2014). Standard 60909.
[11] Taha Selim Ustun, Cagil Ozansoy, Aladin Zayegh, Fault current coefficient and [35] IEC standard for single input energizing quantity measuring relays with
time delay assignment for microgrid protection system with central dependent or independent time, IEC standard 60255.
protection unit, IEEE Trans. Power Syst. 28 (2) (2013) 598–606.
[12] Eric Sortomme, S.S. Venkata, Joydeep Mitra, Microgrid protection using
communication-assisted digital relays, IEEE Trans. Power Deliv. 25 (4) (2010)
2789–2796.