598 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO.

2, MAY 2013

Fault Current Coefficient and Time Delay

Assignment for Microgrid Protection System
With Central Protection Unit
Taha Selim Ustun, Cagil Ozansoy, and Aladin Zayegh

Abstract—With growing numbers of distributed generators alter the grid structure and jeopardize safe and reliable opera-
(DGs) getting connected to the network, new protection schemes tion [4]. These are only a few of the issues caused by the con-
are required. These schemes are aimed at responding to the
changing fault current values, bi-directional current flow and
nection of DGs to grids. In order to tackle them a new concept
distributed generation (DG) at all levels of the grid. For this called “Microgrid” is proposed. A microgrid is a collection of
purpose these schemes utilize comprehensive protection systems loads and microgenerators along with some local storage and
with extensive communication and coordination between DGs and behaves just like a model-citizen [5] from grid side thanks to
protection devices such as relays. This paper details the assign-
intelligent control [6]. Although a microgrid has many gener-
ment of two parameters which are critical for proper operation
of a microgrid protection system. The first parameter, the fault ators and loads, it appears as a net load or a net generator to
current coefficient, represents the fault current supplied by any the broader grid with well-behaved characteristics [7], i.e., with
DG to any point inside the network whereas the second parameter stable voltage and frequency.
is the adjustment of relay hierarchy for selective operation of
Through microgrids the gigantic conventional utility network
relays. The automated assignment of these parameters serves the
notion of self-operating microgrid protection system. Further- can be divided into smaller networks which manage distributed
more, elimination of manual design and calculation facilitates generators, loads, storage and protection devices in their own
deployment of new DG units and thus makes it possible to design grid [8]. In microgrids these generators are handled in smaller
plug-and-play DGs and protection devices.
quantities and, thus, DGs can be connected to the grid and side-
Index Terms—Distributed generation, fault currents, microgrid effects on the grid operation can be eliminated.
control, relay programming, selectivity. Despite their advantages, microgrids bring along some tech-
nical challenges in control, management and protection fields
I. INTRODUCTION [9], [10]. Since the very basics of conventional network opera-
tion do not hold any more (such as grid’s being radial and pas-

T HE search for cleaner and more efficient systems drew

engineers’ attention to distributed generators (DGs). As
a result of recent developments in technology and concerns for
sive) revolutionary changes are required for safe operation. Pro-
tection systems, in particular, are affected heavily by the current
contribution of DGs.
global warming, the new tendency in the power engineering There are different solutions proposed in the literature for
field is to generate electricity from cleaner energy sources and the issues stated above. Some of these solutions have been de-
closer to the consumption areas [1]. This implies that the share scribed in [11]–[19]. The use of anti-islanding frequency relays
of DG in the power generation shall increase substantially. is proposed in [11] for frequency dependent protection. One
These generators may be renewable energy (RE) based such as concern with this technique is that it is not realistic to assume
wind turbines, solar systems, etc. that all relays can be replaced [12]. An alternative method, given
However, existing transmission and distribution networks are in [13], calculates operating points of relays with modified par-
not suitable for large scale DG connections since they were tra- ticle swarm optimization. There are more ground-breaking pro-
ditionally designed with the assumption of a passive network. tection systems which have been designed and developed to em-
The DG interconnection to such networks changes their funda- ploy unprecedented communication and coordination methods
mental characteristics and create unprecedented technical prob- to follow the system parameters and carry out necessary calcu-
lems [2]. Moreover, in the case of a fault, DG systems contribute lations [14], [15]. The drawback of these methods is the need
to the fault currents and the transient characteristics of the net- for new communication devices and the limited versatility for
work become completely different [3]. Since the inverter inter- new deployments. Realizing this gap, authors have proposed a
faced DGs (IIDGs) have highly variable characteristics, they protection system which is versatile [16] and can be extended
for different topologies such as fault current limiters [17]. Fur-
Manuscript received March 29, 2011; revised August 01, 2011, September thermore, the protection system proposed in [16] is able to ac-
13, 2011, November 04, 2011, and January 02, 2012; accepted August 14, 2012. commodate new deployments with plug-and-play approach [18]
Date of publication September 20, 2012; date of current version April 18, 2013.
Paper no. TPWRS-00279-2011. and it can be implemented based on the international commu-
The authors are with the School of Engineering and Science, Victoria Uni- nication standards such as IEC 61850 [19]. Other approaches
versity, Melbourne, Australia (e-mail: [Link]@[Link]). have the disadvantage of either requiring a set microgrid struc-
Color versions of one or more of the figures in this paper are available online
at [Link] ture (not adaptive) or requiring new communication/protection
Digital Object Identifier 10.1109/TPWRS.2012.2214489 modules for each device [13], [15]. This would increase the cost

0885-8950/$31.00 © 2012 IEEE

USTUN et al.: FAULT CURRENT COEFFICIENT AND TIME DELAY ASSIGNMENT 599

and the complexity of the protection system since every new de-
ployment will bring along a communication device.
How to calculate the new fault currents and fault levels for
any change occurring in the system is a major concern. Current
systems use some sort of database or event table to search the
current status and take pre-determined precautions [13], [15].
However, since microgrids are designed to accommodate new
generators and loads, these schemes are not practical. Some sort
of algorithm which dynamically calculates fault currents and
manages to adjust the protection scheme to the new state of mi-
crogrid is direly needed. This paper contributes to the literature
by satisfying this demand.
The organization of this paper is as follows: Section II
explains the new generation protection systems and their
operation principles, Section III outlines the derivation and
calculation of the fault current coefficient and Section IV details
the adjustment of selectivity hierarchy. Section V shows the
adaptation of the proposed system for an exceptional fault case.
Section VI investigates the reliability of the proposed scheme,
and Section VII draws the conclusions. Finally, Section VIII
gives an insight about further development of the system.

II. NEW GENERATION PROTECTION SCHEMES Fig. 1. Centralized protection system with communication.
WITH CENTRAL PROTECTION UNIT
In conjunction with the changing conditions in microgrids, be a communication network which offers cost-effective and
fault currents and fault levels need to be recalculated and power readily available communication. With these inputs, the MCPU
ratings of protection equipments should be rearranged accord- calculates the tripping current value for each relay. In other
ingly. Furthermore, IIDGs have highly nonlinear and varying terms, the MCPU assigns the current value above which the
characteristics. Consequently, it is a challenge in itself to calcu- relay should interrupt the connection. For a given relay “ ” in
late the fault contributions of IIDGs. It is accepted, in the liter- the network, the operating current is calculated as in (1). This
ature, that under fault conditions the maximum contribution of equation considers the grid’s and all DGs’ fault contribution on
an IIDG is 1.5 to 2 times its rated current value [4], [20]. For that particular relay:
DGs with rotating machines such as diesel generators this coef-
ficient is taken as 5 [4]. Since many of the storage devices are
connected to the grid through an inverter interface, they can be
modeled as an IIDG and their fault current contribution is cal- (1)
culated accordingly.
Microgrids are dynamic entities where DGs, loads, storage where m is the total number of DGs in the microgrid,
devices etc. are constantly connected and disconnected. Oper- is the impact factor of distributed generator on the
ating conditions vary rapidly since the inertia of the system is fault current of the relay ,
much less than that of conventional networks. Weather condi- is the maximum fault current contribution of
tions, resource availability and consumption affect the operating DG
conditions of the equipment. In order to ensure safe operation of If the microgrid is operating in islanded mode, then the grid’s
such a system, it is required to monitor all elements and make fault contribution will be multiplied by the “Operating Mode”
the necessary changes in their settings as the operating condi- bit which will be equal to 0. Likewise, the fault contribution of
tions of the microgrid change. Since the conventional protection a DG which is not in operation will be annulled by its status bit.
schemes that rely on large inertia and long transient periods are The contribution of the fault grid is calculated, as shown in
insufficient, new protection schemes are necessary [16]. In this (2), by taking the Thevenin equivalent of the electric network
paper, one such protection system is being proposed by the au- as in traditional fault current calculations [21], [22]:
thors, as shown in Fig. 1 [17].
In this system, the Microgrid Central Protection Unit (2)
(MCPU) communicates with every single relay and distributed
generator in the microgrid. The communication with relays is As shown in Fig. 2, denotes the Thevenin equivalent of
necessary to update the operating currents of the relays and to the utility grid and remains the same for different parts of the
detect the direction of fault currents and thus isolate the fault microgrid.
properly. DGs, on the other hand, are monitored to follow their However, which denotes the Thevenin impedance be-
status and include/disregard their fault contribution if they tween the utility grid and point of calculation varies with respect
are ON/OFF, respectively. The communication scheme will to distance. Therefore, in the above equation. is a constant
600 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 2, MAY 2013

Fig. 3. Fault contribution of DG at different locations.

Fig. 2. Thevenin equivalent circuit taken from point A.

Fig. 4. Symmetrical components scheme.

value while is a function of distance, i.e., . In this

case, the fault grid contribution also becomes a function of the fault current contribution of rotating DGs such as diesel genera-
distance and can be defined as tors or wind generators. In fault current calculations, the impact
factors of IIDGs can be adjusted to fine tune the fault current cal-
culated in (4). This introduces extra versatility to the proposed
(3)
system, if in the future, better fault current estimation schemes
are developed for different inverters with different topologies
The signal indicates whether that DG is in opera- and control schemes. However, in this study the impact factors
tion or not. Depending on the type of the DG, i.e., whether it is are taken as 1 for IIDGs and their fault contributions are worked
inverter interfaced or with rotating machines, is calcu- out from (4).
lated as in (4) and (5), respectively. In literature, the fault con- Calculation of k for rotating DGs is a crucial part of this pro-
tributions of the Inverter interfaced DGs is reported to range be- tection system and is detailed in Sections III–VIII.
tween 1.2 and 2 [4], [20]. In order to be practical and more real-
istic, authors selected the coefficient 1.5 since the extreme con- III. CALCULATION OF DG IMPACT FACTOR-K
ditions (where the fault contribution is 1.2 or 2) will not occur
Fig. 3 shows a section of a microgrid where DG1 and Relay
as frequent as conditions which require fault contribution coef-
4 are separated by a distance of “x”. At point A, i.e., the output
ficient to be 1.5:
of DG1 (which may be a diesel generator or a wind generator),
(4) the fault current IfaultDG1 occurs. At a distance of x, at point
B, this current decreases to .
(5) In order to calculate the impact coefficient, consider the sym-
metrical representation of this system under fault conditions in
The impact factor, namely k, is introduced to calculate the
Fig. 4. This representation is based on the symmetrical compo-
fault contribution of DGs at distant points in the microgrids. It
nents calculations [24].
takes a value between 0 and 1 and represents the decrease in the
Since these faults are steady faults, is assumed to be zero
fault current due to inductance and resistance of low voltage dis-
[25]. From the electrical equation for the direct loop voltage the
tribution lines. The relays closer to the DG under consideration
following can be written:
will have higher “ ” levels whereas those which are further
downstream or upstream will have lower coefficients. In this (6)
way not only the effect of distribution lines can be taken into
account but also a more flexible and versatile protection system where is the direct voltage output of DG, is the output
is designed. This is especially desirable for growing networks impedance, and is the rated output voltage of DG which is
with potential new deployments. regulated by the stability and safety code of the grid. is the
The fault currents of IIDGs do not vary with distance but with fault current supplied, is the impedance of the transmission/
their control loops [23]. Therefore, the impact factor k does not distribution line per length, and is the distance between DG
apply to IIDGs. It is introduced to account for the variation of and the relay under consideration.
USTUN et al.: FAULT CURRENT COEFFICIENT AND TIME DELAY ASSIGNMENT 601

The fault current supplied is the maximum fault current This equation can be written in compact form as in (14):
available, i.e., , becomes the rated voltage of the
(14)
DG. Different fault current capacities of DGs are represented
by this output impedance. Rearranging (6), the fault current can MCPU will assign each row of the vector Irelay to the related
be written in terms of circuit parameters and the distance as relay in the network. With the help of communication lines, the
network will be monitored and these values will be constantly
(7) re-calculated and updated.
In this paper, only balanced three-phase to ground faults are
There are two unknown parameters in this equation. In order considered. It is assumed that the microgrids are considerably
to work them out, an additional equation is required. This is smaller than the conventional electrical networks. This means
the voltage relation between , the voltage that needs to be that the size of a microgrid is suitable for protection with de-
sustained according to the stability and safety code of the grid, layed-type instantaneous relays which implement definite-time
and . It can be written as in (8): grading technique. In case of communication failure relays sur-
(8) vive until communication is restored. The latest fault current set-
tings shall be kept in the relay until the link reconnects. Thanks
Replacing in (6), can be expressed as to the local decision making scheme of the proposed method,
the relays will be able to operate based on these fault current
(9) settings. In this fashion, though it may not have the most de-
sirable settings the protection system will be active in case of a
In this fashion, a DG’s fault current contribution at a distance communication failure and the microgrid will be protected from
x can be calculated. Note that all the parameters on the right catastrophic conditions.
hand side of (9) are known before any fault occurs. This allows The authors only consider the faults occurring in the micro-
the MCPU to calculate all fault contributions and assign tripping grid. Faults in other feeders and/or grids are not considered in
current levels to the relays for proper operation. Once the fault this work. Since the microgrids under consideration are assumed
current contribution is calculated the impact factor can be to be sufficiently small, the relays are close to the fault locations
expressed as the ratio of to : in each branch. That is to say, the fault current magnitude does
not differ significantly between the actual location of fault and
(10) the location of the relay for which the calculations are carried
out. This protection system is developed for radial electrical net-
where and denote the DG and the relay under consideration. works. Should an algorithm be developed to assign proper fault
The protection system will create a matrix by calculating the current settings and adjust relay hierarchy, the system can be ex-
impact factors of many DGs on many relays: tended to non-radial networks such as ring or meshed networks.

IV. RELAY HIERARCHY ADJUSTMENT FOR SELECTIVITY

(11) Selectivity is a vital concept in protection systems. It refers
to isolating the fault with the nearest circuit breaker so that the
rest of the system will not be affected in case of a fault, the cir-
cuit breakers should react according to a hierarchy. The circuit
From (4) and (5), the maximum fault current contribution of breakers which are downstream and closer to the fault point are
every DG is calculated and the vector , having dimen- required to operate first. However, if the fault current is very
sions can be represented as shown in (12): large and the downstream circuit breaker fails to interrupt, then
the upstream circuit breaker with larger capacity should operate
and isolate the fault. With the introduction of DGs, downstream
(12) and upstream circuit breaker concepts are prone to change ac-
cording to the status of the microgrid.
Consider the system shown in Fig. 5. This network has
generation at all branches. Through the combination of circuit
The tripping currents of relays which represent the thresholds breakers, various alternative network structures can be formed.
for relay operation are calculated by the cross product of matrix Throughout this section, the authors will refer to relays being
and vector . The result is the relay tripping current closed or open as a shortcut to indicate the open/closed status of
vector of dimension , : the Circuit Breakers (CBs) associated with each relay. Assume
that R1, R2, R3, R4, R6, and R7 are closed whereas R5 remains
open. When a fault occurs at the terminals of Load 2, then the
most downstream relay will be Load 2’s own relay (represented
by the little box) and selectivity implies that its CB should in-
terrupt the fault. If Load 2’s CB fails to achieve that in a prede-
termined time (delay), then the proper selectivity requires that
(13) R6 operates.
602 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 2, MAY 2013

Fig. 6. 2-pair selectivity method algorithm.

Fig. 5. Sample microgrid. TABLE II

CRITICAL RELAY HIERARCHY FOR VARIOUS CASES

TABLE I
2-PAIR SELECTIVITY METHOD DATA

Therefore, the time delay is selected to be between these min-

The proposed algorithm employed by the MCPU determines imum and maximum time values.
the network structure according to the status of critical relays. To further clarify the operation of the algorithm four different
The status of a critical relay changes the structure of the net- structures of the system shown in Fig. 5 are studied. These are:
work. Following this definition R1, R2, R3, R4 and R6 are 1) R2, R3, R4, and R6 are closed. The middle bus is fed
all critical relays whereas Load 2’s relay and DG1’s relay are through R6 connection.
non-critical relays. After determining the microgrid structure 2) R2, R3, R4, and R5 are closed. The middle bus is fed
MCPU assigns 2-pair selectivity couples. Table I shows the through R5 connection.
2-pair selectivity data for the left branch of the example given 3) R2, R3, R5, and R6 are closed. The middle bus and the right
above. bus are fed through R5 and R6 connections, respectively.
The method owes its name to the fact that every relay is 4) R2, R4, R5, and R6 are closed. The left bus and the middle
paired up with its downstream relay. The information provided bus are fed through R5 and R6 connections, respectively.
includes; the operating current of that particular relay and the It is required that these cases are studied by engineers and
operating current of its downstream relay. Whenever a fault oc- proper hierarchy of the critical relays should be determined and
curs, the relay which is closest to the fault operates to isolate saved in the MCPU. While monitoring the network, MCPU will
the fault. If it proves to be unsuccessful in a predetermined time also monitor the status of critical relays. In accordance with the
(i.e., time delay of the upstream relay), the upstream relay rec- changes in their status, MCPU will compare the present struc-
ognizes the fault and isolates the fault. Fig. 6 shows the flow ture with the predetermined structures saved in the memory.
chart implemented by each relay for 2-pair selectivity method. After determining the present structure, MCPU will retrieve
Let us consider the case in Table I; the pairs are R2-R3 and the critical relay hierarchy pertaining to it and assign 2-pair
R3-I.R (Individual Relay). Should a fault occur at Load 1, I.R. selectivity.
should isolate the fault. If unsuccessful, when the predetermined Alternatively, the graph theory-based approach developed by
time expires (e.g., 200 ms), its upstream relay, i.e., R3 isolates the authors [18] can be utilized to extract the relay hierarchy.
the fault. In this manner, the proposed system can be imple- In this fashion, the system acquires higher independence and
mented in simple and complex microgrids. plug-and-play concept can be realized.
The value of the time delay (i.e., 200 ms) is assigned by con- For each case given above, the proper critical relay hierarchy
sidering different factors. It is enforced by the stability param- is given in Table II. As shown, these cases have independent
eters that a fault shall be cleared before a maximum time when branches in microgrid. That means, critical relays in these
the network becomes unstable. At the same time, practical con- branches are elements of discrete sets and do not belong to
siderations for over-current relay communication and operation same 2-pair distribution. For instance, in case 1, R3 and R4
imply that there is a minimum time that these CBs can operate. are discrete from each other and thus they are not considered
USTUN et al.: FAULT CURRENT COEFFICIENT AND TIME DELAY ASSIGNMENT 603

TABLE III TABLE IV

RELAY PAIRING FOR CASE 1 RELAY PAIRING FOR CASE 3

also, change. R6 for instance is at 2nd level of 2nd branch in

Case 1 while it is placed at 3rd level in case 3.
As shown in Table II, for Case 3 there is only one branch with
5 hierarchy levels. Therefore, 2-pair assignment for this case is
as shown in Table IV.
When Tables III and IV are compared, it becomes evident that
the pairing of the relays change and they need to be updated.
A reliable communication is vital for this purpose. In similar
fashion, the protection system will recognize any new structure
and repeat pairing process according to Table II. If an automatic
relay hierarchy detection algorithm such as in [18] is not uti-
lized, then it has utmost importance that the network operators
have their power engineers study possible network structures of
the microgrid, list the proper hierarchy of the relays and pre-
pare Table II. The MCPU will automatically operate according
to these data.

V. FAULTS OUTSIDE THE MICROGRID

One exceptional case arises when a fault occurs outside the
microgrid, i.e., in the utility grid. Since the fault is in the utility
grid, the microgrid should be isolated from the system by
opening R1. In this case, the interfacing relay R1, which is at
the top of the relay hierarchy, is required to operate first. Fig. 8
shows the fault currents and the inversed selectivity hierarchy
under these circumstances. In order to address this issue, it is
required that relays shall operate bidirectionally as proposed
earlier [15], [26]. The proposed system, which is a combination
of fault current and relay hierarchy assignment, is implemented
bidirectionally. DGs’ fault contribution will be calculated for
the critical relays through which they are connected to the
utility grid.
Referring to the system in Fig. 8, for instance, reverse trip-
ping current of R2 will include the contributions of DG1 and
Fig. 7. Network structure change from Case 1 (a) to Case 3 (b). DG2. Once the reverse tripping currents are calculated the relay
hierarchy order will be exactly the opposite of order given in
Table II. This means, when a fault outside the microgrid occurs
simultaneously for 2-pair distribution. R3 is considered for 1st and the fault current flows in reverse direction (i.e., upstream)
branch whereas R4 is considered for 2nd branch. R1 will act first and isolate the microgrid from the fault. If, for
Case 3 can be considered as a single branch since, due to its any reason, it fails to interrupt the current reverse selective hi-
single line structure, 3rd branch covers 1st and 2nd branches. In erarchy imposes that R2 interrupt reverse fault current.
these cases, none of the relays are discrete and they are all con- In order to realize this in a systematic way, relay should have
sidered in the same time for 2-pair distribution. If the network four different parameters, as shown in Table V. In this table each
operates under case 1 then, 1st and 2nd branches will have the relay has two relay-pairing (forward and reverse) and fault cur-
2-pair assignments shown in Table III. rent parameters, one for forward (downstream) and one for re-
As shown in Fig. 7, when R4 opens and R5 closes the struc- verse (upstream) fault current flows. Two relay currents shall
ture changes from that of case1 to that of case 3. For this new be calculated individually by considering fault current contri-
microgrid structure, the relay hierarchy and 2-pair assignments butions of various DGs. Pairing of the relays for 2-pair selec-
shall be updated. The place of relays in the relay hierarchy may, tivity, on the other hand, can be calculated at once. The reverse
604 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 2, MAY 2013

Fig. 9. Local decision making scheme in relays.

failures are classified and new parameters are introduced to de-

sign a more comprehensive reliability assessment scheme [27],
[31].
These systems consider issues related to relay hardware, relay
Fig. 8. Faults in the utility grid. software, ancillary equipment, communication units and human
errors [29], [31]. All of them consider the probability of a failure
TABLE V occurrence and its impact on system dependability and secu-
BIDIRECTIONAL PARAMETERS FOR CASE 1 rity as defined by IEEE standard C37.100-1992 [32]. The prob-
ability of a failure is defined by [30], [33]

(15)

where all the individual probability components are added to

find the overall value. If there is more than one event that is
required for a failure to occur then the combined probability can
be expressed as in (16):

(16)

where n is the total number of independent events.

As mentioned in earlier sections, in case of a communication
failure the protection system proposed in this paper implements
a “last-setting-valid” approach. Fig. 9 shows the local decision
relay hierarchy is extracted by inverting the forward hierarchy making mechanism employed to active the circuit breakers. As
and the assignment is trivial. shown, the operation is independent of the communication link
This additional feature can easily be implemented in MCPU if and this ensures that relays are always set to operate, though it
the relays have the capability to operate in both directions. Two may not be the most desirable operation point.
different tripping currents will be calculated, namely forward However, this requires two events: 1) a failure in the commu-
and reverse fault currents, and should the fault current flow in nication line, 2) an electrical fault in the network before the com-
reverse direction the Critical Relay Hierarchy will be reversed. munication line is restored. Therefore the probability of a relay
operating following “last-setting-valid” method is the multipli-
cation of the probabilities of communication link failure and
VI. RELIABILITY CONSIDERATIONS electrical fault:
The protection systems in electrical networks play a vital (17)
role in sustaining stability and ensuring safe operation. With
the development of new technologies, protection systems be- where P[comm] is the probability of a communication failure
came more complex. Significant progress has been recorded in occurring in the microgrid while P[fault] is the probability of
power system reliability modeling and computation by applying an electrical fault occurring in the same microgrid. Equations
quantitative analysis based on probability theory [27]. In re- (15) and (16) hold for all conditions except for extreme climate
cent years, different reliability assessment methods have been conditions where power system fault follows power law distri-
proposed such as Reliability Block Diagram (RBD) [28], 3RF bution instead of exponential distribution [34]. Extreme weather
Technique [29], and Software Reliability Allocation [30]. The conditions are not considered in this study.
USTUN et al.: FAULT CURRENT COEFFICIENT AND TIME DELAY ASSIGNMENT 605

It is clearly mentioned in the literature that with the im- VIII. FUTURE WORK
proving technology the hardware failures are less frequent [30]. The system is very versatile and can be developed to suit
Consequently, the value of probability of communication is much larger and more complex microgrids. An automated al-
very small. The values assumed for these probability values are gorithm may be run to determine the structure of the network
around on the order of or [31], [33]. For short term and deduce correct order of the relays. If that is realized by the
analysis these values may drop as low as [35]. Accord- MCPU then the Critical Relay Hierarchy Table can be created
ingly, replacing these characteristic figures in (17) yields by the system. This will eliminate the human input and new de-
ployments can be easily done.
(18) Special attention must be paid to the communication protocol
utilized for the communication of data within the system and a
As a result, the protection system proposed in this paper standard packet-type for the relay. This will also have significant
proves to be reliable in accordance with the requirements of effect on the amount of communication delay experienced. This
the standards [32]. Considering the value of , inevitable delay can be deducted from the overall delay assigned
it is concluded that the possibility of relays operating when to the relays and thus its effect can be eliminated on the system.
there is a communication failure is very small. Even if they Finally, simulation of the proposed system shall provide more
are forced to operate before the communication is restored, the realistic information about the dynamic response, calculation
implementation of “last-setting-valid” method ensures that no time and communication delay.
catastrophic condition occurs.
As long as the communication speed is in permissible levels, REFERENCES
the communication speed is not vital. Communication is only [1] C. M. Colson and M. H. Nehrir, “A review of challenges to real-time
required for updating relay settings and this occurs when a DG power management of microgrids,” in Proc. IEEE Power & Energy
Society General Meeting, PES ’09, 2009, pp. 1–8.
connects to or disconnects from the grid. In case of a fault, no [2] R. A. F. Currie et al., “Fundamental research challenges for active man-
communication is required as relays operate based on Local De- agement of distribution networks with high levels of renewable gener-
ation,” in Proc. 39th Int. Universities Power Engineering Conf., UPEC
cision Making. If, as an exceptional case, a fault occurs just 2004, 2004, vol. 2, pp. 1024–1028.
after a Relay Setting Update process has started, then the reg- [3] T. Boutsika et al., “Calculation of the fault level contribution of dis-
ular communication speed is around 80 ms [36] and this ensures tributed generation according to IEC Standard 60909,” in Proc. CIGRE
Symp., Athens, Greece, Apr. 2005.
that relay settings will be updated and the fault will be cleared [4] P. P. Barker and R. W. De Mello, “Determining the impact of dis-
before maximum allowed time expires. tributed generation on power systems. I. Radial distribution systems,”
in Proc. IEEE Power Engineering Society Summer Meeting, 2000, vol.
3, pp. 1645–1656.
VII. CONCLUSION [5] B. Lasseter, “Microgrids,” in Proc. IEEE 2001 WM Panel, Role of Dis-
tributed Generation in Reinforcing the Critical Electric Power Infra-
This paper has presented two procedures to assign two key structure, 2001.
[6] M. Barnes et al., “Real-world microgrids—an overview,” in Proc.
parameters in a microgrid protection system with a MCPU. The IEEE Int. Conf. System of Systems Engineering, 2007.
first of these parameters is the DG impact factor “k” which is [7] B. Lasseter, “Microgrids [distributed power generation],” in Proc.
useful in accurately anticipating the fault current contributions IEEE Power Engineering Society Winter Meeting, 2001, 2001, vol. 1,
pp. 146–149.
of DGs’ over a relay. The calculation of the impact factor in- [8] F. Katiraei et al., “Microgrids management,” IEEE Power and Energy
volves parameters which are known beforehand. Hence, “k” Mag., vol. 6, pp. 54–65, 2008.
[9] N. Hatziargyriou et al., “Microgrids,” IEEE Power and Energy Mag.,
can be calculated before the fault actually occurs. Secondly, this vol. 5, pp. 78–94, 2007.
paper presents an algorithm to adjust the hierarchy of relays for [10] B. Kroposki et al., “Making microgrids work,” IEEE Power and En-
ergy Mag., vol. 6, pp. 40–53, 2008.
proper selective operation in microgrids. The concepts of crit- [11] J. C. M. Vieira et al., “Performance of frequency relays for dis-
ical relay and 2-pair selectivity are introduced. The procedure tributed generation protection,” IEEE Trans. Power Del., vol. 21, pp.
is automated as much as possible by decreasing human input to 1120–1127, 2006.
[12] T. S. Ustun et al., “Recent developments in microgrids and example
minimum. cases around the world—A review,” Renew. Sustain. Energy Rev., vol.
Finally, the faults occurring outside the microgrid were con- 15, pp. 4030–4041, 2011.
[13] H. H. Zeineldin et al., “Protective relay coordination for micro-grid
sidered. It is explained how this situation necessitates the pro- operation using particle swarm optimization,” in Proc. Power Engi-
tection system to reverse its operation. The proposed protection neering, 2006 Large Engineering Systems Conf., 2006, pp. 152–157.
system is adapted accordingly and the assignment of two pa- [14] H. Laaksonen, “Protection principles for future microgrids,” IEEE
Trans. Power Electron., to be published.
rameters under consideration is modified to suit all cases. As [15] A. Oudalov and A. Fidigatti, “Adaptive network protection in micro-
a result, a complex centralized microgrid protection system, its grids,” in More Microgrids Europe. [Online]. Available: [Link]
[Link]/documents/[Link].
operation principles, and calculation/assignment procedures are [16] T. S. Ustun et al., “A microgrid protection system with central protec-
explained in detail. tion unit and extensive communication,” in Proc. 10th Int. Conf. Envi-
ronment and Electrical Engineering (EEEIC), 2011, 2011, pp. 1–4.
The main contribution of this work is to develop an algorithm [17] T. S. Ustun et al., “A central microgrid protection system for networks
to assign necessary operating parameters to a microgrid protec- with fault current limiters,” in Proc. 10th Int. Conf. Environment and
tion system which is already in literature. In doing so, the paper Electrical Engineering (EEEIC), 2011, 2011, pp. 1–4.
[18] T. S. Ustun et al., “Implementation of Dijkstra’s algorithm in a dynamic
extends the scope of the said protection system and automizes microgrid for relay hierarchy detection,” in Proc. IEEE Int. Conf. Smart
the calculation process. This is important to follow the dynamic Grid Communications, Belgium, 2011.
[19] T. S. Ustun et al., “Modeling of a centralized microgrid protection
changes occurring in the microgrid and also, to easily accom- system and distributed energy resources according to IEC 61850-7-
modate new deployments. 420,” IEEE Trans. Power Syst., to be published.
606 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 2, MAY 2013

[20] B. Hussain et al., “Integration of distributed generation into the grid: Taha Selim Ustun received the B.E. degree in elec-
Protection challenges and solutions,” in Proc. 10th IET Int. Conf. De- trical and electronics engineering from Middle East
velopments in Power System Protection (DPSP 2010), pp. 1–5. Technical University, Cankaya, Turkey, in 2007 and
[21] J. Talaq, “Fault calculations using three terminal Thevenin’s equivalent the Master of Engineering Science degree from the
circuit,” Int. J. Elect. Power & Energy Syst., to be published. University of Malaya, Kuala Lumpur, Malaysia in
[22] M. Dilek et al., “Computing distribution system fault currents and volt- 2009. He is currently pursuing the Ph.D. degree in
ages via numerically computed Thevenin equivalents and sensitivity Victoria University, Melbourne, Australia.
matrices,” in Proc. IEEE PES Power Systems Conf. Expo., 2004, vol. His research interests are power systems, power
1, pp. 244–251. electronics, distributed generation, and microgrids.
[23] C. A. Plet et al., “Fault models of inverter-interfaced distributed gen-
erators: Experimental verification and application to fault analysis,” in
Proc. IEEE Power and Energy Society General Meeting, 2011, 2011,
pp. 1–8.
[24] B. D. Metz-Noblat, “Analysis of three-phase networks in disturbed op-
erating conditions using symmetrical components,” in Schneider Elec-
tric, Cahier technique no. 18.
[25] D. Penkov et al., “DG impact on three phase fault location. DG use for Cagil Ozansoy received the [Link]. degree in elec-
fault location purposes?,” in Proc. Int. Conf. Future Power Systems, trical and electronic engineering (Hons.) from Vic-
2005, 2005, p. 6. toria University, Melbourne, Australia, in 2002. In
[26] R. J. Best et al., “Communication assisted protection selectivity for 2006, he received the Ph.D. degree in power system
reconfigurable and islanded power networks,” in Proc. 44th Int. Uni- communications.
versities Power Engineering Conf. (UPEC), 2009, 2009, pp. 1–5. He is now working as a lecturer and researcher
[27] K. Jiang and C. Singh, “New models and concepts for power system re- in the School of Engineering and Science, Victoria
liability evaluation including protection system failures,” IEEE Trans. University. His major teaching and research focus
Power Syst., to be published. is on electrical engineering, renewable energy tech-
[28] M. I. Ridwan et al., “Reliability and availability assessment of trans- nologies, energy storage, and distributed generation.
mission overhead line protection system using reliability block dia- He has successfully carried out and supervised many
gram,” in Proc. 2010 IEEE Int. Conf. Power and Energy (PECon), sustainability related studies in collaboration with local governments in the
2010, pp. 964–969. past. These include the assessment of solar and wind electric energy potential
[29] Z. Dai and Z. Wang, “Protection dynamic reliability analysis system in Melbourne’s west. He has over 25 publications detailing his work and
based on 3RF technique,” IEEE Trans. Power Syst., vol. 26, pp. contributions to knowledge.
1137–1144, 2011.
[30] D. S. Roy et al., “Software reliability allocation of digital relay for
transmission line protection using a combined system hierarchy and
fault tree approach,” IET Softw., vol. 2, pp. 437–445, 2008.
[31] A. H. Etemadi and M. Fotuhi-Firuzabad, “New considerations in
modern protection system quantitative reliability assessment,” IEEE Aladin Zayegh received the B.E. degree in electrical
Trans. Power Del., vol. 25, pp. 2213–2222, 2010. engineering from Aleppo University, Aleppo, Syria,
[32] IEEE Standard Definitions for Power Switchgear, IEEE Std. C37.100-
in 1970 and the Ph.D. degree from Claude Bernard
1992, 1992.
University, Lyon, France, in 1979.
[33] J. Sykes et al., “Reliabilty of protection systems (what are the real con-
cerns),” in Proc. 2010 63rd Annu. Conf. Protective Relay Engineers, He has held lecturing position at several univer-
pp. 1–16. sities, and since 1991, he has been at Victoria Uni-
[34] X. Duan and S. Su, “Self-Organized criticality in time series of power versity, Melbourne, Australia. He has been head of
systems fault, its mechanism, and potential application,” IEEE Trans. School and research director where he has conducted
Power Syst., vol. 25, pp. 1857–1864, 2010. research, supervised several Ph.D. students, and pub-
[35] L. Shenghu et al., “Short term reliability analysis for protection system lished more than 250 papers in peer reviewed inter-
with average unavailability,” in Proc. Power and Energy Engineering national conferences and journals. He is currently an
Conf., Asia-Pacific, 2009, pp. 1–6. Associate Professor at the School of Engineering and Science, Faculty of Health,
[36] D. Laverty et al., “Internet based loss-of-mains detection for distributed Engineering and Science at Victoria University. His research interest includes
generation,” in Proc. 42nd Int. Universities Power Engineering Conf., renewable energy, embedded systems, instrumentation, data acquisition and in-
2007, pp. 464–469. terfacing, sensors and microelectronics for biomedical applications.