>Yahoo! is not affiliated with the authors of this page or responsible for its content.
A NEW SCHEME FOR IMPROVED DISTRIBUTION PROTECTION EMPLOYING NUMERICAL OVERCURRENT RELAYS A NEW SCHEME FOR IMPROVED DISTRIBUTION PROTECTION EMPLOYING
NUMERICAL OVERCURRENT RELAYS
D. T. W. Chan*, C. Z. Lu**
*School of EEE, Nanyang Technological University
**School of EEE, Nanyang Technological University
Abstract
A new protection scheme utilizing digital overcurrent relays is proposed in this paper. Using the
concept of characteristic vector, the proposed protection scheme can locate a fault by comparing
the corresponding fault current vector with characteristic vectors in a pre-built characteristic vector
database before relays operate according to their own settings. Trip command is then sent to
appropriate relays from a host computer based on a pre-built relay trip table. Fault can thus be
cleared without unnecessary loss of loads. By updating the characteristic vector database and the
relay trip table, the proposed protection scheme can be applied to the changed system, thus to
achieve adaptive feature. No complex optimization technique is required. A typical distribution
system is employed to test the proposed protection scheme and its adaptive approach. Test results
show that the scheme has very good performance.
1.

INTRODUCTION
The normal protection practice for distribution
systems requires a unit protection, often of the pilot-
wire current differential type for feeders, and
supplemented with overcurrent (O/C) relays as back-
up protection. With unit protection installed, fast fault
clearance can be achieved and fault identification is
relatively easy. However, there are applications where
pilot-wire differential protection may not be
economically justifiable, and O/C protection may have
to be the sole and main protection. For example, the
unit protection for the feeder can not protect buses in
the distribution systems, and the bus fault can only be
cleared by O/C relays. Nevertheless, under such
circumstances, it will be extremely difficult to achieve
ideal protection that only the fault part is isolated from
the healthy system, particularly when the coordination
of O/C relays can not be maintained due to changes in
system configuration requiring parallel or inter-
connected operations.
In this paper, a new scheme for distribution system
protection with only O/C relays as protection devices
is proposed. The concept of
characteristic vector is
proposed in this scheme to symbolize a fault location.
Based on a pre-built
characteristic vector database,
fault location can be deduced by comparing fault
current vector and characteristic vectors in the
database before relays operate, and trip command can
be sent directly to the desired relays from a host
computer via communication channel between them.
A system fault thus can be isolated swiftly without
unnecessary loss of loads. The relays should be
numerical type, which can send and receive data under
the control of the host computer. These relays are in
wide spread use nowadays.
Simulation tests have been carried out with
satisfactory resulted noted. The scheme can even
identify busbar faults thus providing improved
performance in terms of selectivity.
2.

A POTENTIAL SOLUTION
Suppose there is a fault at location F in the network
(Fig.1), where
Z
f
and
Z
g
are the impedance between
phases and phase to ground respectively. Based on the
three-component method [1], fault current for different
types of faults consists of three components, namely
positive, negative, and zero sequence current, although
in some cases some of them are zero. These
components are the currents flowing in positive,
negative, and zero sequence networks respectively.
Therefore, at a healthy feeder
i in the faulted system,
there are also three sequence currents corresponding to
the three sequence networks. They are positive
sequence feeder current
I
ia</i>1
, negative sequence feeder
current
I
ia</i>2
, and zero sequence feeder current
I
ia</i>0
.
In general, distribution systems are linear constant
systems. Based on sequence networks of the faulted
system and ignore all the loads, we should be able to
find the relation between
I
ia</i>0
and
I
a</i>0
,
I
ia</i>1
and
I
a</i>1
,
I
ia</i>2
and
I
a</i>2
. These relations can be expressed as follows:
I
ia</i>0
=
K
i</i>0
I
a</i>0
I
ia</i>1
=
K
i</i>1
I
a</i>1
(1)
I
ia</i>2
=
K
i</i>2
I
a</i>2
Where I
a</i>0
,
I
a</i>1
,
I
a</i>2
are zero, positive and negative sequence
fault currents at fault location
I
ia</i>0
,
I
ia</i>1
,
I
ia</i>2
are zero, positive and negative sequence
fault currents at feeder
i in the system
K
i</i>0
,
K
i</i>1
,
K
i</i>2
are zero, positive and negative sequence
Feeder Coefficient of feeder i respectively
with regard to fault location F
Fig 1. Four most common fault types in power system
Feeder Coefficient is a new concept proposed here.
When we calculate fault current at a specified location
in a network, if the fault current is so large that load
currents have very small influence on it, we can ignore
load currents and assume the system is an idle system.
In this case, for a particular feeder corresponding to a
particular fault location in the system, these three
Feeder Coefficients are constants and can be
calculated on the basis of circuit theory and the three
sequence networks.
Feeder Coefficient may vary from
feeder to feeder and from fault location to fault
location. For a particular feeder
i, it has different
Feeder Coefficients for different fault locations. And
for a particular fault location, different feeders may
have different
Feeder Coefficients.
If the three
Feeder Coefficients of a feeder
corresponding to a fault location are same, namely
K
i</i>1
=K
i</i>2
=K
i</i>0
=
K
i
,
K
i
is the general
Feeder Coefficient
of feeder
i with regard to fault location F, we can
have:
I
ia</i>1
=
K
i</i>1
I
a</i>1
I
ia</i>2
=
K
i</i>2
I
a</i>2
(2)
I
ia</i>0
=
K
i</i>0
I
a</i>0
Thus, at feeder
i, we have












=












=






2
1
0
2
2
2
1
0
2
2
1
1
1
1
1
1
1
1
1
1
a
i
a
i
a
i
ia
ia
ia
ic
ib
ia
I
K
I
K
I
K
a
a
a
a
I
I
I
a
a
a
a
I
I
I






=












=
c
b
a
i
a
a
a
i
I
I
I
K
I
I
I
a
a
a
a
K
2
1
0
2
2
1
1
1
1
1
(3)
Therefore






=






c
b
a
i
ic
ib
ia
I
I
I
K
I
I
I
(4)
Equation (4) shows that the fault current value at
feeder
i is K
i times that at the fault location, i.e.
f
i
if
I
K
I
=
(5)
In equation (5),
I
if
denotes fault current flowing
through feeder
i, I
f
denotes fault current at fault
location in the faulted system. Equation (5) shows
that, no matter what type of fault it is, the magnitude
of fault current at feeder
i is proportional to that at
fault location. We therefore can build a
characteristic
vector with fault current magnitudes at all relaying
locations corresponding to a fault location. By
normalizing these fault current magnitudes (dividing
them by the maximum fault current magnitude), all of
the normalized fault currents will be less than or equal
to 1.0. This will make it easier to compare fault
currents under different fault types. If we draw a curve
of normalized fault current versus relaying location,
the shape of the curve for a particular fault location
will be identical despite different fault types.
Furthermore, curves for different fault locations will
be different from each other since fault current
distribution in the system will not be the same. We
can therefore make a conclusion that for such
distribution systems, each fault location has its own
characteristic vector and different fault locations have
different
characteristic vectors. Therefore, the
characteristic vector, which consists of fault current
magnitudes at all relaying positions, can be used to
describe a fault location.
B
I
b
=0 I
c
=0
A
C
Z
f
I
a
B
I
b
I
c
A
C
Z
f
I
a
=0
B
I
b
I
c
A
C
Z
f
I
a
=0
B
A
C
Z
f
I
a
Z
f
Z
g
I
b
+I
c
Z
f
Z
f
I
b
I
c
Z
g
I
a
+I
b
+I
c
1-phase-ground fault
2-phase fault
2-phase-ground fault
3-phase fault
F
F
F
F Fig 2. Curves for different faults at Bus 3 of the
simplified distribution network
Fig.2 shows the curves of normalized fault currents
versus relaying locations (RLY01 to RLY14) for four
basic fault types at bus 3 of a simplified distribution
network, (which is used to test the proposed protection
scheme and is described later). No impedance is
applied to the fault. We can see these curves match
well.
Therefore, when a fault occurs, current magnitudes at
all relays are retrieved to build a fault current vector,
in which the data items have the same format as those
in
characteristic vectors. We then compare it with
those
characteristic vectors. A decision on where the
fault location is can be made from the comparison.
Trip commands are sent to appropriate relays before
the relays operate based on their PS and TMS settings.
The fault thus can be cleared without unnecessary loss
of loads.
3.

FAULT DETECTION
To compare the fault current vector with characteristic
vectors in the database, a coefficient of correlation
(Pearson