Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
Series: Automatic Control and Robotics Vol. 10, N
o
2, 2011, pp. 167  180
A MULTIOBJECTIVE APPROACH TO OPTIMAL
ALLOCATION OF DISTRIBUTION AUTOMATION DEVICES
IN RADIAL MEDIUM VOLTAGE NETWORK

UDC 621.317 681.5.01 681.513.5
Miodrag Stojanović, Dragan Tasić, Aleksa Ristić
University of Niš, Faculty of Electronic Engineering, Department of Power Engineering,
Aleksandra Medvedeva 14, 18000 Niš, Serbia
Email: miodrag.stojanovic@elfak.ni.ac.rs, dragan.tasic@elfak.ni.ac.rs,
aleksa.ristic@elfak.ni.ac.rs
Abstract. In this paper a method for selection of optimal scenario for distribution
automation is proposed. For a part of distribution network that can be considered as
the smallest functional unit, the service zones are determined based on the heuristic
rules, all possible automation scenarios are searched, and the values of selected
objective functions are determined. After that, the method uses fuzzy multicriteria
evaluation and grey relational analysis in the application of evaluating a set of
alternatives in terms of decision criteria. Automation scenarios are ranked on the basis
of objective function values. The method is demonstrated on test example of real
medium voltage distribution network.
Key words: distribution automation, multiobjective optimization, medium voltage network
1. INTRODUCTION
Electric power utilities tend to run their businesses as efficient enterprises providing
energy at an acceptable level of quality. The emergence of deregulation has dramatically
changed the business environment. This radical shift in business goals, now occupying
power companies, in many countries as a result of deregulation, open access, and
privatization, is causing a significant review of network design and operating practices.
Network control and automation are playing a key role in enabling the network owners to
adapt to the changing situation and opportunities to achieve their business goals [1]. One
of the most important reasons for introducing distribution automation is efficient fault
management [2, 3].
Received October 6, 2011

Acknowledgement. The paper is a part of the research done within the projects III 44006 and TR 33035 Ministry of
Education and Science of the Republic of Serbia.
168 M. STOJANOVIĆ, D. TASIĆ, A. RISTIĆ
Increasing the automation level of distribution network, above all, affects the
reduction of outage duration time when a fault occurs. With reduction of outage duration
time, the unsupplied amount of energy to the customers is decreased. To determine the
optimal solution of this problem, it is necessary to cover a very large amount of data.
Distribution network optimization problem is an extremely complex nonlinear optimization
problem with large number of constraints. As a criterion of optimality, different parameters
can be adopted (benefit of the electric power utility over the planned time period, cost to
benefit ratio over the planned time period or the effects achieved in improving the reli
ability indicators). However, to consider the problem in total, it is not sufficient to con
sider only one objective function, but it is necessary to consider the problem as a multi
objective optimization problem.
Distribution network covers a very large territory, contains a very large number of ele
ments (lines, substations etc.) and they are developing dynamically. Considering very
rapid development of technology, the lifetime of control and communication equipment
applied in distribution automation is short compared to the lifetime of electric power
equipment. Equipment prices of different manufacturers are in a relatively wide range and
largely depend on purchased amount of the equipment, business interest of the
manufacturer for entering the particular market and other factors. In addition, it should be
noted that the installation process of medium voltage network automation devices is long
term and cannot be implemented over one year but the investments are distributed over a
period of several years. It is obvious that during the analysis of optimal device selection
for distribution network automation and location selection, it is almost impossible to
include all relevant influences, but certain simplification must be introduced.
A large number of papers presenting this issue are directed toward the optimal loca
tion selection for only one, known in advance, type of distribution network automation
device. The paper [4] presents a method for sectionalizer relocations for customer inter
ruption cost minimization which uses simple analytical computations and heuristic rules
to search for the new locations. To determine the number of sectionalizing switches and
their location, an optimization technique of simulated annealing is proposed in [5], while
in [6] a genetic algorithm procedure is proposed. In [7] immune algorithm is used to solve
the optimal placement of line switches. The objective function is formulated by consider
ing the outage cost and investment cost of line switches. The paper [8] proposes a two
stage decomposition approach to optimal remote controlled switch allocation in distribu
tion systems.
A small number of papers consider different types of automation equipment simulta
neously [3,9,10]. The paper [3] proposes methodology for optimal level of investment in
medium voltage network, based on heuristic combinatory search algorithm with simulta
neous consideration of scenarios with different types of automation equipment: local
automation and remote control. The essence of the algorithm is decomposition of an op
timal automation problem with different types of automation equipment to subproblems
with one type of equipment. The optimization problem is defined as multiobjective with
three objective functions (benefit, reliability indicators and cost/benefit). The threestate
approach inspired by the discrete version of particle swarm optimization is presented in
[9] to determine optimal number and locations of two types of switches (sectionalizes and
breakers) in radial distribution system. A technique to carry out optimized placement of
control and protective devices in the same optimization process simultaneously is
A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices... 169
presented in [10]. The problem is modelled through mixed integer nonlinear programming
model, and solved using reactive tabu search algorithm.
There is a large number of distribution network planning and exploitation problems that
can be classified as a multiobjective optimization problems, for example, the selection of
optimal distribution network restoration plan [1113]. In [11] a hybrid fuzzygrey method is
proposed to obtain satisfactory restoration plan for distribution system restoration.
In this paper a method for selection of optimal scenario for distribution automation is
proposed. For a part of distribution network that can be considered as the smallest func
tional unit, the service zones are determined based on the heuristic rules, all possible
automation scenarios are searched, and the values of selected objective functions are de
termined. After that, the method uses fuzzy multicriteria evaluation and grey relational
analysis in the application of evaluating a set of alternatives in terms of decision criteria.
Automation scenarios are ranked on the basis of objective function values.
2. METHODOLOGY
The selection of device for distribution automation and location selection for their in
stallation is a very complex problem. From the point of global optimum, it is necessary to
consider simultaneously the entire distribution network. However, the total number of
possible scenarios that should be counted when considering entire distribution network
and all types of distribution automation devices is extremely large. Considering dimension
of a problem, computational time is very long. Regardless of the remarkable progress of
computer technology, finding the optimal solution by direct searching all possible
solutions is practically impossible for a network that contains tens of thousands of
elements. The alternative to direct searching of the optimal solution is the use of heuristic
methods that do not analyze all possible combinations for allocation of different elements.
A difficulty for heuristic methods is low sensitivity of objective functions. Namely, when
location of an element is changed or one type of element is changed by the other type
within some automation scenario that contains hundreds or thousands of elements, the
values of criterion functions are almost unchanged.
Based on previous consideration, it is obvious that is necessary to decompose the
problem into subproblems that will be solved independently. For this reason, distribution
network is divided into parts which represent the smallest functional unit that can be con
sidered independently. Medium voltage feeders that supply rural areas are usually radial
and can be considered independently, or with the ability for backup supply when it is nec
essary to consider several feeders simultaneously. But that is not the case for urban cable
networks. In case of urban cable networks that are meshed, division cannot be done un
ambiguously, but taking into account the power line limitations it can still be done. Be
cause strictly division is not possible, the number of strategies is chosen in terms of
separation points within the network. A larger network, of course, corresponds to a larger
number of strategies. After the optimal solution for all strategies is found, the synthesis of
solutions should be done.
The next question is how to allocate planned investments for automation of the entire
network in certain parts of the network. One approach is to allocate investments in certain
parts with respect to annual supplied energy. However, optimization procedure is not in
170 M. STOJANOVIĆ, D. TASIĆ, A. RISTIĆ
cluded in allocating the investments. Contrary to this approach, after division of distribu
tion network into functional units, the desirable level of automation can be defined by
defining the desirable values of objective functions. For each objective function a
criterion is defined for evaluating scenarios as "good" or "bad". Considering the fact that
scenario quality is described with linguistic terms "good" or "bad", it is obvious that
application of fuzzy logic is very suitable.
For optimality criterion a maximal benefit or minimal cost to benefit ratio are usually
used. Reliability indicators can also be used as optimality criteria, but they do not take
into account the information about supplied energy and customer category. Since none of
the criteria can include all relevant factors for selection of optimal solution, the best
option is simultaneous consideration. Increasing the investments will improve reliability
indicators, which means that optimal solution cannot be found with respect to these
criteria only. Increasing the investments will also increase the benefit but only to some
boundary value and after that the benefit starts do decrease, but application of only this
criterion favours a more expensive solutions. On the other hand, cost to benefit ratio is the
most suitable for solutions with low level of investments. After analyzing all parts of the
network, total investments can be determined. If total investments are close to desirable
values, the problem is solved. Otherwise, desirable values of objective functions are
increased or decreased and the whole procedure is repeated.
Division of the network in functional units will significantly reduce the problem di
mension; however, solving the subproblems is also very demanding when considering a
large number of different devices. Namely, if there is an optimal allocation of N
eq
different devices for the network with N
line
branches, assuming that every device can be
installed at the beginning of every branch (on feeder or in the substation from which the
feeder starts), the total number of scenarios which should be considered is
eq
N
line
N . For the
60branch network that will be considered in the test example and 4 different types of
devices, there are about 13 million scenarios. Because of that, it is suitable to use some of
the heuristic methods for solving the subproblems. Since the analysis tends to include all
types of devices, dimension of subproblem can only be reduced by reducing the number
of potential locations for device installation, which can be achieved by dividing the
network into service zones. Potential locations for the installation of devices for
distribution automation are at the beginning of service zones, reducing the number of
analyzed scenarios. The criteria for service zone forming could be minimal length of
power lines, maximal number of power substations, locations of existing pole mounted
switching equipment or ring main units in distribution substations, etc. The division of the
60branch network into eight zones will reduce the total number of scenarios to 4096 for
four different types of devices.
2.1. Characteristic reliability indices in distribution systems
Nowadays, electric power utilities use a variety of different reliability indicators [1416],
which is a consequence of the absence of unique criteria for reliability assessment. System
Average Interruption Frequency Index (SAIFI) represents the average number of outages
longer than 3 minutes per customer and it is measured over the course of a year.
A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices... 171
N
N f
SAIFI
i
i i ¿
= (1)
In previous equation f
i
is the frequency of fault i, N
i
is the number of customers inter
rupted by fault i, and N is the total number of customers.
System Average Interruption Duration Index (SAIDI) represents the average duration
of outage per customer and it is measured over the course of a year.
N
N D
SAIDI
i
i i ¿
= (2)
In the previous equation D
i
is the cumulative outage duration of customers interrupted by
fault i during the year. Beside SAIFI and SAIDI, the following indicators are also used:
Customer Average Interruption Duration Index (CAIDI), Average Service Availability
Index (ASAI), Average Service Unavailability Index (ASUI), Customer Interruptions (CI),
Customer Minutes Lost (CML), Average System Interruption Duration Index (ASIDI),
Energy Not Supplied (ENS), etc.
For analysis in this paper, reliability indicators SAIFI and SAIDI will be used because
of their widespread use (except in Austria, Great Britain and Spain) [16]. The
introduction of penalties, for the case when reliability indicators are worse than the ones
prescribed by the regulator, creates the need of including these indicators as objective
functions in distribution automation optimization problem. It can be included as a
separate objective function at multiobjective optimization, or, with respect to the values
of penalties, included in the economic objective function (actualized total costs function
or cost/benefit function).
2.2. Outage duration assessment
Reliability indicators, SAIFI and SAIDI are determined on the basis of fault data and
outage data. On the basis of historical data about faults, reliability indicators of system
elements, which are usually expressed through the fault rate, can be also determined.
To determine the expected value of reliability indicator SAIFI and SAIDI for one as
sumed automation scenario, beside the value of reliability indicators of system elements,
outage duration for a fault on each element should be assessed for each customer. Outage
duration depends largely on the position of the observed customer node, with respect to
the power source and a faulted section. In general, outage duration of a customer node
due to a faulted section is a function of: network conception, organization of finding a
faulted section (isolating procedure), searching a location in the faulted section, the way
of communicating with control center [3,7].
When fault occurs on some section of a distribution feeder, the breaker in supply sub
station is opened due to activation of the protection. As consequence, outage of all cus
tomers supplied by this feeder occurs. After receiving information about fault, operator
determines feeder under fault and send field crew to isolate faulted section. Bisectional
search method is very popular isolating procedure because of its simplicity, although ap
proaches that take into account the information about configuration and network state
appeared in recent time. Because of this fact, when outage duration is estimated in this
paper, isolating procedure by the bisectional search method will be used.
172 M. STOJANOVIĆ, D. TASIĆ, A. RISTIĆ
Under this method, distribution substation or polemounted switch in the midpoint of
the faulted feeder represents the first place visited by the field crew. On arrival, the crew
opens the sectionalizer and disconnects the half of feeder. After receiving information
about opening the sectionalizer, the operator closes the breaker in supply substation. If the
protection is activated again, the fault has occurred at the first half of the feeder.
Otherwise, the fault has occurred at the second half of the feeder. After determination of
the faulted half, the crew closes the previously opened sectionalizer and moves to the
faulted half. Repeating these actions will lead to the isolation of the faulted section.
Customers supplied by the faulted feeder will start receiving electrical energy again
after a period of time. This time period depends on the fault location and their position
with respect to that location. If the customer is "in front of" the fault location, the energy
will be resupplied after time period T
A
which is the sum of alert time T
alert
, arriving time
of field crew T
arrive
, and time needed to find and isolate the faulted section T
isolate
.
isolate arrive alert A
T T T T + + = (3)
Alert time represents the time that elapses between the moment when the fault occurs
and the moment when the operator receives the information about that fault. Arriving time
represents the time interval during which the operator determines the faulted feeder and
the field crew arrives at the starting point of the fault isolating procedure. Time needed to
find and isolate the faulted section depends on the distance travelled, velocity of the field
crew and the number of performed manipulations.
If the customer is "behind" the fault location, the energy will be resupplied after time
period T
B
, which is longer than the time period T
A
. If there is no backup supply, this time
period can be expressed as:
repair micro A B
T T T T + + = (4)
In the previous expression T
micro
is time needed for finding the microlocation of the
fault on the previously isolated section, and T
repair
is repairing time of the faulted element.
The presented method for determination of the fault location corresponds to the network
without any automation.
2.3. Economic objective functions
Objective function of actualized total cost takes into account the investment in net
work automation, operation and maintenance costs as well as the costs of unsupplied en
ergy due to outages for a period of time. The total cost of the scenario S is determined by
the following expression:
¿
=
÷ ÷
+ ÷ + + + + =
n
j
S
rem
n j S
j M
S
j ENSI
S
j inv
S
inv
S
tot
I a a C C C C C
1
, , , 0 ,
) 1 ( ) 1 )( ( (5)
In previous expression n is the duration of the planned period in years, j is the
considered year during the planned time period,
S
inv
C
0 ,
is the investment cost till the
beginning of the first year,
S
j inv
C
,
is the investment cost during the jth year,
S
j ENSI
C
,
is the
outage cost during the jth year,
S
j M
C
,
is the operation and maintenance cost during the j
th year,
S
rem
I is the remaining value of the equipment at the end of the planned period, and
A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices... 173
a is the actualization rate. The remaining value of the element at the end of the planned
period can be expressed as:
¿
=
m
m rem
S
rem
I I
,
(6)
¦
¹
¦
´
¦
>
< ÷


.

\

÷ +
=
m life m life
m life m life m
m life
m life
m rem
n n I
n n I I
n
n
I
I
, ,
, , 0
,
,
,
;
); ( 1
(7)
In these expressions m is equipment index, n
life,m
is lifetime of the equipment m, I
0m
is
investment cost of the equipment m, I
rem,m
is value of the equipment m at the nth year,
and I
life,m
value of the equipment m at the end of lifetime.
When calculating the investment cost of one scenario, in addition to the equipment
cost (C
S
eq
), it is also necessary to consider the additional costs (C
S
ad
) which include design
ing costs, installation costs, functional testing costs and commissioning costs. Equipment
cost includes the prices of power equipment, control equipment and telecommunication
equipment if the object is introduced into the remote monitoring and control system.
The outage cost during the jth year is determined as follows:
¿¿
A =
i k
ik k pik j
P f c C (8)
In the previous expression, i is the customer category index, k is disturbance index,
ΔP
ik
is cutoff power of the customer i caused by disturbance k, f
k
is frequency of
disturbance k, and c
pik
is outage cost of the customer i per cutoff power.
For calculating the outage cost of customer i per cutoff power, the following expres
sion can be used:
) €/kW (
3
2 1
i
C
i i pik
d C C c + = (9)
In this expression, d is duration of outage,
i
C
1
,
i
C
2
and
i
C
3
are coefficients whose
values depend on the customer category. The coefficients also depend on the price of
electricity and differ from one country to another.
As the second economic criterion of optimality, what can be used is the cost to benefit
ratio (cost/benefit) which should be minimized. The cost includes all investment costs
actualized at the beginning or at the end of the considered time period, while benefit
represents the difference between the total cost due to the current state and the total cost
when automation equipment of one scenario is installed. All the costs must be actualized
at the same moment in time. The cost to benefit ratio for scenario S can be represented
with the following relation:
S
S
S
B
I
CB = (10)
In this expression I
S
is the investment cost for the scenario S, B
S
is benefit (total cost
reduction) due to the reduction of outage cost and reduction of operation and maintenance
costs for the scenario S.
174 M. STOJANOVIĆ, D. TASIĆ, A. RISTIĆ
2.4. Fuzzy multicriteria evaluation
In the analyses of optimal location of distribution automation equipment, the scenario
is considered more preferable if reliability indicators SAIFI and SAIDI are "better" and
total cost as well as cost/benefit indices are "smaller". Obviously, the vague terms "better"
and "smaller" are very imprecise in nature and can be conveniently handled by the fuzzy
set theory.
The fuzzy multicriteria evaluation [12] consists of a basic rule, membership func
tions, and a conclusion procedure. The basic rule formed with fuzzy rules, which describe
relationships in a linguistic sense, are written as pairs of "IFTHEN" statements. The
fuzzy rules are expressed in following forms:
IF ) ( X f
i
is Low, THEN the plan is Good.
IF ) ( X f
i
is Moderate, THEN the plan is Moderate.
IF ) ( X f
i
is High, THEN the plan is Bad.
All values of objective functions ) ( X f
i
are described using three fuzzy sets: Low,
Moderate and High. The related membership functions are shown in Fig. 1. Consequently,
the fuzzy sets Good, Moderate and Bad are clearly defined as 1, 0.5 and 0, respectively.
This can simplify the computation process and completely satisfy the needs of the study.
In the inference procedure, the real value of each objective function is firstly calculated to
get the boundary values in fuzzy IFTHEN rules. The weighted average is applied in
order to obtain real values:
Fig. 1. Fuzzy numbers which represent linguistic variables
¿
¿
=
=
µ
µ
=
F
F
N
j
j
N
j
j j
i
y
f
1
1 *
(11)
In the previous expression, µ
j
and y
j
are boundary value and fuzzy result in the jth
fuzzy rule, respectively. N
F
presents a number of fuzzy rules. The values of f
i
*
represents
the fitness degree of objective functions f
i
for each scenario.
A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices... 175
2.5. Grey relational analysis
The grey relational analysis (GRA) is an important approach of the grey system theory
in the application of evaluating a set of alternatives in terms of decision criteria. In GRA,
the data that contain same features are regarded as sequence. As a tool of quantitative
analysis, the GRA can be used to measure the relationship between two sequences by
calculating their correlative degrees, which is called grey relational grade (GRG). The
GRG is expressed by a scalar between 0 and 1. Up to now, the method has been successfully
applied in many fields and has attracted many researchers who will continue exploring this
study. The principal formulas of the GRA are briefly described in further text.
Considering a reference sequence x
0
=(x
0
(1), x
0
(2), ..., x
0
(n)) and m comparative se
quences x
i
=(x
i
(1), x
i
(2), ..., x
i
(n)), i= 1, 2, ..., m, where x
i
(k) represents the kth entry in x
i
,
k=1, 2,..., n. The grey relational coefficient (GRC) of x
i
with respect to x
0
in kth entry is
as follows:
min max
0 max
0
) (
)) ( ), ( (
A ÷ A
A ÷ A
÷ ¸
k
k x k x
i
i
(12)
) ( ) ( max max
0 max
k x k x
i
k i
÷ ÷ A
¬ ¬
(13)
) ( ) ( min min
0 min
k x k x
i
k i
÷ ÷ A
¬ ¬
(14)
) ( ) ( ) (
0 0
k x k x k
i i
÷ ÷ A (15)
The GRGs between each comparative sequence x
i
and the reference sequence x
0
can
be derived from the average of the GRC, which is denoted as:
¿
=
¸ = I
n
k
i i
k x k x
n
1
0 0
)) ( ), ( (
1
(16)
I
0i
represents the degree of relation between each comparative sequence and the refer
ence sequence. Higher degree of relation means that the comparative sequence is more
similar to the reference sequence.
3. TEST EXAMPLE
The selection of the optimal scenario for distribution automation of one small part of
the network will be demonstrated on the example of real radial medium voltage 10 kV
distribution network shown in Fig. 2. The network element data and customer data are
shown in Table 2. The analysis is made under the following assumptions. Supplying
substation is remotely controlled, annual fault rate level of lines is 0.15 1/km, arriving
time of the field crew 2 h, speed of field crew moving during the fault management
procedure is 1 m/s, manipulation time 0.15 h, time for repairing the faulted element 2 h,
and time needed for fault isolation with remote controlled switching equipment 0.25 h.
The duration of the planned time period is 10 years, actualization rate 8%, lifetime of
control equipment 10 years, lifetime of power switching equipment 30 years, value of the
equipment at the end of lifetime is 10% from the investment cost of the equipment. The
176 M. STOJANOVIĆ, D. TASIĆ, A. RISTIĆ
total cost and benefit for the planned time period will be actualized to starting year. As
objective functions, in addition to the economic functions, the reliability indicators SAIFI
and SAIDI are used.
For division of distribution network into service zones, the following criteria are used:
a zone contains maximum 5 distribution substations, the length of all branches in the zone
is more than 2 km. The locations of the already installed switches are also taken into
account when defining the service zones. When the proposed criteria are applied, eight
zones on considered feeder, marked with Roman numerals in Table 1, can be identified.
Firstly, an analysis is conducted over the current state, reliability indicators are determined,
and outage cost is evaluated.
The coefficients used for the calculation of the outage cost for different customer
categories are shown in Table 2. Table 3 shows summary data about the zones and these
data are: the number of distribution substations that belong to the zone (N
dt
), the sum of
rated power of transformers that belong to the zone (S
rT
), the average real power of all
residential category customers that belong to the zone (P
av res
) and the average real power
of all commercial customers that belong to the zone (P
av com
).
The following path of movement is assumed for the field crew, during fault isolating
in the current state network (before automation). First, the field crew should go to the
switching device S3 and open it. If the fault is not isolated, the crew should go to the
location of the switching device S5 and manipulate it. Again, if the fault is not isolated,
the crew should go to the location of the switching device S2 and manipulate it. The
analysis of distribution network before installation of the automation equipment leads to
the following values: SAIFI
10
7.7265 1/yr.custom., SAIDI
10
14.13 h/yr.custom., and total
cost
150,560 €.
Fig. 2. Distribution test system
A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices... 177
Table 1. Test system data
Line Zone Length Line Zone Length S
r
P
av res
P
av com
1 I 1220 32 I 240 160 75.175 2.067
2 I 870 33 I 500 160 14.175 0
3 I 350 34 I 115 250 55.105 1.645
4 I 360 35 I 80 100 0.581 7.611
5 I 860 36 II 60 160 16.544 2.355
6 II 405 37 II 350 160 15.333 0
7 II 690 38 II 85 160 31.771 1.568
8 II 1040 39 II 435 250 13.118 0.23
9 II 300 40 II 195 250 11.103 0
10 II 1025 41 III 50 160 6.839 0
11 II 520 42 III 540 250 22.975 1.058
12 II 30 43 III 340 250 23.669 3.2
13 III 665 44 III 885 45 7.648 0
14 III 770 45 III 20 50 1.168 0
15 III 310 46 IV 155 45 6.151 0
16 III 810 47 IV 850 160 20.479 0
17 III 550 48 IV 350 160 21.821 1.439
18 IV 365 49 V 410 160 8.876 0
19 IV 1625 50 V 280 100 17.393 0
20 IV 650 51 VI 500 160 14.412 0
21 IV 310 52 VI 520 160 5.705 0.136
22 IV 50 53 VI 900 250 37.863 0.368
23 V 1320 54 VII 120 250 39.315 3.78
24 VI 375 55 VII 70 100 13.164 0.566
25 VII 330 56 VII 45 160 23.943 1.23
26 VII 1420 57 VIII 100 160 10.761 0.24
27 VII 920 58 VIII 40 160 19.536 0.207
28 VII 290 59 VIII 290 100 12.499 0.537
29 VIII 1940 60 VIII 1295 160 21.331 0
30 VIII 510
31 VIII 810
Table 2. Coefficients for outage cost evaluation
Customer category C
1
C
2
C
3
Commercial 1.65 7 1
Residential 0.3 h ≤ d < 1 h 0.58 1.97 1
1 h ≤ d < 5 h 1.1 0.3 2.5
5 h ≤ d ≤ 48 h 15.3 0.11 2
To show the results obtained by the application of multicriteria fuzzygrey evaluation,
8 scenarios out of the set of scenarios with approximately the same automation level are
taken and shown in Table 4. In this table S represents the switch; D represents the fault
detector, C represents the recloser, while R represents remote control.
178 M. STOJANOVIĆ, D. TASIĆ, A. RISTIĆ
Table 3. Summary data about service zones
Zone I II III IV V VI VII VIII
N
ds
4 5 5 3 2 3 3 4
L
line
[km] 4.596 5.135 4.94 4.355 2.01 2.295 3.195 4.985
S
rT
[kVA] 670 980 755 365 260 570 510 580
P
av res
145.04 87.869 62.299 48.451 26.269 57.98 76.422 64.127
P
av com
11.323 4.153 4.258 1.438 0 0.504 5.576 0.984
Table 4. Distribution automation equipment for eight selected scenarios
Scenario\Zone I II III IV V VI VII VIII
1 RC S RSD S RSD S
2 RC S RSD RD S RSD S
3 RC S RSD S RSD SD
4 RC RSD RSD S RSD S
5 RC RSD RSD RD S RSD S
6 RC S RC S S
7 RC SD RC RD S SD
8 RC S RC RD S S S
Table 5 shows objective function values for eight selected scenarios, the investment
cost of the equipment for every scenario, as well as additional costs (installation cost,
technical documentation cost, functional testing and commissioning cost, training cost,
and estimated value of telecommunication equipment infrastructure cost). The table also
contains calculated values of reliability indicators SAIFI
10
and SAIDI
10
which represents
the impact of faults on 10 kV network on supply reliability, calculated values of outage
cost, obtained profit from network automation in the 10year time period and cost to
benefit ratio (cost/benefit).
On the basis of the objective functions shown in Table 5, it is obvious that total cost
function has its minimum for scenario 5, but the investment is the greatest. Scenarios 2
and 3 have approximately the same value of total cost function but scenario 2 has slightly
lower value of cost to benefit ratio. A potential solution is also scenario 8, which demands
slightly higher value of investment and has worse cost to benefit ratio, but on the other
hand, it has a lower value of reliability indicator SAIFI
10
than scenario 2. For fuzzy multi
criteria evaluation shown in Table 5, the following characteristic values of fuzzy numbers
for selected objective functions are used: f
1a
= 1.05, f
1b
= 1.15, f
1c
= 1.25, f
2a
= 0.3,
f
2b
= 0.35 and f
2c
= 0.4, f
3a
= 3, f
3b
= 4.5, f
3c
= 6, f
4a
= 6, f
4b
= 8 and f
4c
= 10. Now, grey
relational grade is calculated using expressions (13)(16) and scenarios are ranked as
shown in the last row of Table 5. The optimal solution is represented with scenario 2
which has the investment value of 20,200 €. If investment for the obtained optimal
solution deviates much from the planned value, then with a modification of the
characteristic values of fuzzy numbers, a different ranking of potential solutions can be
obtained.
A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices... 179
Table 5. Objective functions and fuzzygrey evaluation for eight selected scenarios
Scenario 1 2 3 4 5 6 7 8
C
eq
13500 14500 15500 19200 20200 13000 16000 16000
C
ad
5300 5700 6100 8100 8500 4500 5700 6550
I
rem
2220 2250 2270 2160 2640 2450 2320 2450
C
tot
95217 82873 83960 97673 76758 113375 89892 85000
f
1
(C
tot
/C
tot min
) 1.2405 1.0797 1.0938 1.2725 1 1.4771 1.1711 1.1074
f
2
(C/B) 0.3397 0.2984 0.3243 0.5162 0.3889 0.4706 0.3577 0.3401
f
3
(SAIFI
10
) 4.7265 4.7265 4.7265 4.7265 4.7265 3.4803 3.4803 3.4803
f
3
(SAIDI
10
) 8.792 8.042 8.011 8.278 6.867 10.24 8.484 8.259
f
1
*
0.0476 0.8517 0.7808 0 1 0 0.3944 0.7131
f
2
*
0.603 1 0.7567 0 0.1112 0 0.4231 0.5986
f
3
*
0.4245 0.4245 0.4245 0.4245 0.4245 0.8399 0.8399 0.8399
f
4
*
0.3 0.49 0.5 0.43 0.78 0 0.38 0.44
I
0i
0.4385 0.7856 0.709 0.3079 0.674 0.3042 0.6033 0.7409
Rank 6 1 3 7 4 8 5 2
4. CONCLUSION
A method for selection of the optimal scenario for distribution automation is proposed
in this paper. The problem of distribution automation is a complex, nonlinear and
discrete optimization problem of enormous dimensions and it is not possible to apply an
appropriate optimization procedure directly. Because of this, the proposed method firstly
determines the service zones based on the heuristic rules for a part of distribution network
that can be considered as the smallest functional unit. Then, all possible automation
scenarios are searched, and the values of the selected objective functions are determined.
After that, the method uses fuzzy multicriteria evaluation and grey relational analysis in
the application of evaluating a set of alternatives in terms of decision criteria. Automation
scenarios are ranked on the basis of objective function values. The method is
demonstrated on the test example of a real medium voltage distribution network.
REFERENCES
1. J. N. Green and R. Wilson, Control and automation of electrical power distribution systems, CRC Press
Taylor & Francis, 2006.
2. CIRED Working Group WG03, Fault management in electrical distribution systems, Final Report, 1995.
3. D. S. Popovic, Lj. R. Glamocic and M. D. Nimrihter, "The optimal automation level of medium voltage
distribution networks", Electrical Power and Energy Systems, vol. 33(3), pp. 430438, March 2011.
4. J. H. Teng and C. N. Lu, "Feederswitch relocation for customer interruption cost minimization", IEEE
Transaction on Power Delivery, vol. 17(1), pp. 254259, January 2002.
5. R. Bilington and S Jonnavithula, "Optimal switching device placement in radial distribution systems",
IEEE Transactions on Power Delivery, vol. 11(3), pp. 16461651, July 1996.
6. G. Levitin, S. MazelTov and D. Elmakis, "Optimal sectionalizer allocation in electric distribution
systems by genetic algorithm", Electric Power System Research, vol. 31(2), pp. 97102, October 1994.
7. C.S. Chen, C. H. Lin, H. J. Chuang, C. S. Li, M. Y. Huang and C. V. Huang, "Optimal placement of line
switches for distribution automation systems using immune algorithm", IEEE Transactions on Power
Systems, vol. 21(3), pp. 12091217, August 2006.
180 M. STOJANOVIĆ, D. TASIĆ, A. RISTIĆ
8. P. M. S. Carvalho, L. A. F. M. Ferreira and A. J. C Silva, "A decomposition approach to optimal remote
controlled switch allocation in distribution systems ", IEEE Transactions on Power Delivery, vol. 20(2),
pp. 10311036, April 2005.
9. A. Moradi and M. FotuhiFiruzabad, "Optimal switch placement in distribution system using trinary
particle swarm optimization algorithm ", IEEE Transactions on Power Delivery, vol. 21(3), pp. 1209
1217, January 2008.
10. L. G. W. Silva, R. A. F. Pereira, J. R. Abbad and J. R. S. Matovani, "Optimized placement of control and
protective devices in electric distribution systems through reactive tabu search algorithm", Electric
Power System Research, vol. 78(4), pp. 372381, March 2008.
11. W.H. Chen, M.S. Tsai and H.L. Kuo, "Distribution system restoration using the hybrid fuzzygrey
method", IEEE Transactions on Power Systems, Vol. 20, No. 1, February 2005, p.p. 199205.
12. S. J. Lee, S. Lim and B. S. Ahn, "Service restoration of primary distribution systems based on fuzzy
evaluation of multicriteria", IEEE Transaction on Power Systems, vol. 13(3), pp. 1156–1163, August 1998.
13. Y. P. Huang and C. C. Huang, "The integration and application of fuzzy and grey modelling methods",
Fuzzy Sets and Systems, vol. 78(1), pp. 107–119, February 1996.
14. R. Billinton, "Distribution system reliability performance and evaluation", Electrical Power and Energy
Systems, vol. 10(3), pp. 190200, July 1988.
15. J. Nahman and V. Mijailović, "Reliability of distribution systems ", Akademska misao, 2009, in Serbian.
16. Council of European Energy Regulators, 4
TH
Benchmarking Report on Quality of Electricity Supply, 2008.
VIŠEKRITERIJUMSKI PRISTUP ODREĐIVANJU OPTIMALNE
LOKACIJA UREĐAJA DISTRIBUTIVNE AUTOMATIKE U
RADIJALNOJ SREDNJENAPONSKOJ DISTRIBUTIVNOJ MREŽI
Miodrag Stojanović, Dragan Tasić, Aleksa Ristić
U radu je predstavljen jedan pristup izboru optimalne vrste opreme za automatizaciju
distributivne mreže, i njene optimalne lokacije. Problem automatizacije distributivne mreže je
složen, nelinearni diskretni optimizacioni problem izuzetno velike dimenzionalnosti zbog čega je
praktično nemoguća direktna primena optimizacionih procedura. Zbog toga je u radu najpre
predložena podela problema na podprobleme, koji su nakon toga sagledani kao višekriterijumski
optimizacioni zadatak. Za donošenje odluke primenjeni su metodi višekriterijumskog fuzzy
ocenjivanja i "grey" relacionog rangiranja. Prezetovani metod testiran je na primeru realne
radijalne srednjenaponske distributivne mreže.
Ključne reči: automatizacija, srednjenaponska distributivna mreža, višekriterijumska optimizacija
) and they are developing dynamically. The paper [4] presents a method for sectionalizer relocations for customer interruption cost minimization which uses simple analytical computations and heuristic rules to search for the new locations. known in advance. type of distribution network automation device. To determine the number of sectionalizing switches and their location. The objective function is formulated by considering the outage cost and investment cost of line switches. based on heuristic combinatory search algorithm with simultaneous consideration of scenarios with different types of automation equipment: local automation and remote control. A. the lifetime of control and communication equipment applied in distribution automation is short compared to the lifetime of electric power equipment. but certain simplification must be introduced. it is not sufficient to consider only one objective function. The paper [3] proposes methodology for optimal level of investment in medium voltage network. To determine the optimal solution of this problem. different parameters can be adopted (benefit of the electric power utility over the planned time period. A small number of papers consider different types of automation equipment simultaneously [3. However. A technique to carry out optimized placement of control and protective devices in the same optimization process simultaneously is . The paper [8] proposes a twostage decomposition approach to optimal remote controlled switch allocation in distribution systems. In addition. With reduction of outage duration time. The optimization problem is defined as multiobjective with three objective functions (benefit.168 M. while in [6] a genetic algorithm procedure is proposed. business interest of the manufacturer for entering the particular market and other factors. affects the reduction of outage duration time when a fault occurs. RISTIĆ Increasing the automation level of distribution network. In [7] immune algorithm is used to solve the optimal placement of line switches. A large number of papers presenting this issue are directed toward the optimal location selection for only one. D. The essence of the algorithm is decomposition of an optimal automation problem with different types of automation equipment to subproblems with one type of equipment. but it is necessary to consider the problem as a multiobjective optimization problem.10]. Distribution network optimization problem is an extremely complex nonlinear optimization problem with large number of constraints. it should be noted that the installation process of medium voltage network automation devices is longterm and cannot be implemented over one year but the investments are distributed over a period of several years. above all. The threestate approach inspired by the discrete version of particle swarm optimization is presented in [9] to determine optimal number and locations of two types of switches (sectionalizes and breakers) in radial distribution system. substations etc. Equipment prices of different manufacturers are in a relatively wide range and largely depend on purchased amount of the equipment. reliability indicators and cost/benefit). TASIĆ. Distribution network covers a very large territory. STOJANOVIĆ. it is necessary to cover a very large amount of data. to consider the problem in total. cost to benefit ratio over the planned time period or the effects achieved in improving the reliability indicators). Considering very rapid development of technology.9. contains a very large number of elements (lines. As a criterion of optimality. an optimization technique of simulated annealing is proposed in [5]. the unsupplied amount of energy to the customers is decreased. it is almost impossible to include all relevant influences. It is obvious that during the analysis of optimal device selection for distribution network automation and location selection.
For a part of distribution network that can be considered as the smallest functional unit. From the point of global optimum. The next question is how to allocate planned investments for automation of the entire network in certain parts of the network. Namely. all possible automation scenarios are searched. or with the ability for backup supply when it is necessary to consider several feeders simultaneously. computational time is very long. the total number of possible scenarios that should be counted when considering entire distribution network and all types of distribution automation devices is extremely large. The alternative to direct searching of the optimal solution is the use of heuristic methods that do not analyze all possible combinations for allocation of different elements. distribution network is divided into parts which represent the smallest functional unit that can be considered independently. METHODOLOGY The selection of device for distribution automation and location selection for their installation is a very complex problem. But that is not the case for urban cable networks. but taking into account the power line limitations it can still be done. In [11] a hybrid fuzzygrey method is proposed to obtain satisfactory restoration plan for distribution system restoration. There is a large number of distribution network planning and exploitation problems that can be classified as a multiobjective optimization problems. finding the optimal solution by direct searching all possible solutions is practically impossible for a network that contains tens of thousands of elements. The problem is modelled through mixed integer nonlinear programming model. it is obvious that is necessary to decompose the problem into subproblems that will be solved independently. For this reason. However. corresponds to a larger number of strategies. After that. division cannot be done unambiguously. the selection of optimal distribution network restoration plan [1113]. the service zones are determined based on the heuristic rules. the method uses fuzzy multicriteria evaluation and grey relational analysis in the application of evaluating a set of alternatives in terms of decision criteria. In this paper a method for selection of optimal scenario for distribution automation is proposed. the values of criterion functions are almost unchanged. and solved using reactive tabu search algorithm. and the values of selected objective functions are determined. Automation scenarios are ranked on the basis of objective function values. A larger network. One approach is to allocate investments in certain parts with respect to annual supplied energy. the synthesis of solutions should be done. A difficulty for heuristic methods is low sensitivity of objective functions.. However. for example. 169 presented in [10]. of course. Based on previous consideration.A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices. Considering dimension of a problem. In case of urban cable networks that are meshed. Regardless of the remarkable progress of computer technology. optimization procedure is not in . when location of an element is changed or one type of element is changed by the other type within some automation scenario that contains hundreds or thousands of elements. Medium voltage feeders that supply rural areas are usually radial and can be considered independently.. Because strictly division is not possible. the number of strategies is chosen in terms of separation points within the network. 2. it is necessary to consider simultaneously the entire distribution network. After the optimal solution for all strategies is found.
Since the analysis tends to include all types of devices. The criteria for service zone forming could be minimal length of power lines. if there is an optimal allocation of Neq different devices for the network with Nline branches. it is obvious that application of fuzzy logic is very suitable. The division of the 60branch network into eight zones will reduce the total number of scenarios to 4096 for four different types of devices. but application of only this criterion favours a more expensive solutions. For optimality criterion a maximal benefit or minimal cost to benefit ratio are usually used. For each objective function a criterion is defined for evaluating scenarios as "good" or "bad". there are about 13 million scenarios. total investments can be determined. after division of distribution network into functional units. RISTIĆ cluded in allocating the investments. it is suitable to use some of the heuristic methods for solving the subproblems. Otherwise. Characteristic reliability indices in distribution systems Nowadays. the best option is simultaneous consideration. electric power utilities use a variety of different reliability indicators [1416]. reducing the number of analyzed scenarios. Increasing the investments will also increase the benefit but only to some boundary value and after that the benefit starts do decrease. the problem is solved. After analyzing all parts of the network. System Average Interruption Frequency Index (SAIFI) represents the average number of outages longer than 3 minutes per customer and it is measured over the course of a year.1. D. eq 2. which can be achieved by dividing the network into service zones. the desirable level of automation can be defined by defining the desirable values of objective functions. the total number of scenarios which should be considered is N line . Potential locations for the installation of devices for distribution automation are at the beginning of service zones. If total investments are close to desirable values. maximal number of power substations. etc. but they do not take into account the information about supplied energy and customer category. however. Since none of the criteria can include all relevant factors for selection of optimal solution. locations of existing pole mounted switching equipment or ring main units in distribution substations. assuming that every device can be installed at the beginning of every branch (on feeder or in the substation from which the N feeder starts). Namely. Increasing the investments will improve reliability indicators. TASIĆ. desirable values of objective functions are increased or decreased and the whole procedure is repeated. STOJANOVIĆ. which is a consequence of the absence of unique criteria for reliability assessment. cost to benefit ratio is the most suitable for solutions with low level of investments. On the other hand.170 M. which means that optimal solution cannot be found with respect to these criteria only. Because of that. Division of the network in functional units will significantly reduce the problem dimension. solving the subproblems is also very demanding when considering a large number of different devices. A. dimension of subproblem can only be reduced by reducing the number of potential locations for device installation. . Reliability indicators can also be used as optimality criteria. Contrary to this approach. For the 60branch network that will be considered in the test example and 4 different types of devices. Considering the fact that scenario quality is described with linguistic terms "good" or "bad".
It can be included as a separate objective function at multiobjective optimization. outage duration for a fault on each element should be assessed for each customer. or. As consequence. For analysis in this paper. organization of finding a faulted section (isolating procedure). reliability indicators SAIFI and SAIDI will be used because of their widespread use (except in Austria. After receiving information about fault. with respect to the values of penalties. Energy Not Supplied (ENS). although approaches that take into account the information about configuration and network state appeared in recent time. 171 SAIFI i fi Ni N (1) In previous equation fi is the frequency of fault i. Outage duration assessment Reliability indicators.. the way of communicating with control center [3. included in the economic objective function (actualized total costs function or cost/benefit function). outage of all customers supplied by this feeder occurs. . Bisectional search method is very popular isolating procedure because of its simplicity. Customer Interruptions (CI). Average System Interruption Duration Index (ASIDI).A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices. In general. the breaker in supply substation is opened due to activation of the protection. Customer Minutes Lost (CML).. isolating procedure by the bisectional search method will be used.2. On the basis of historical data about faults. when outage duration is estimated in this paper. Average Service Unavailability Index (ASUI). outage duration of a customer node due to a faulted section is a function of: network conception. Outage duration depends largely on the position of the observed customer node. Ni is the number of customers interrupted by fault i. Because of this fact. beside the value of reliability indicators of system elements. searching a location in the faulted section. Beside SAIFI and SAIDI. etc. for the case when reliability indicators are worse than the ones prescribed by the regulator. SAIDI i Di Ni N (2) In the previous equation Di is the cumulative outage duration of customers interrupted by fault i during the year.7]. The introduction of penalties. Great Britain and Spain) [16]. reliability indicators of system elements. with respect to the power source and a faulted section. creates the need of including these indicators as objective functions in distribution automation optimization problem. which are usually expressed through the fault rate. System Average Interruption Duration Index (SAIDI) represents the average duration of outage per customer and it is measured over the course of a year. When fault occurs on some section of a distribution feeder. Average Service Availability Index (ASAI). To determine the expected value of reliability indicator SAIFI and SAIDI for one assumed automation scenario. and N is the total number of customers. 2. operator determines feeder under fault and send field crew to isolate faulted section. the following indicators are also used: Customer Average Interruption Duration Index (CAIDI). SAIFI and SAIDI are determined on the basis of fault data and outage data. can be also determined.
the operator closes the breaker in supply substation. If the customer is "behind" the fault location. j C ENSI . Cinv . Arriving time represents the time interval during which the operator determines the faulted feeder and the field crew arrives at the starting point of the fault isolating procedure. D. velocity of the field crew and the number of performed manipulations. TASIĆ. j C M . j is the investment cost during the jth year. The total cost of the scenario S is determined by the following expression: S S S S S S Ctot Cinv . C ENSI . which is longer than the time period TA. RISTIĆ Under this method.172 M. arriving time of field crew Tarrive. A. After receiving information about opening the sectionalizer. Customers supplied by the faulted feeder will start receiving electrical energy again after a period of time. If the customer is "in front of" the fault location. this time period can be expressed as: TB TA Tmicro Trepair (4) In the previous expression Tmicro is time needed for finding the microlocation of the fault on the previously isolated section. distribution substation or polemounted switch in the midpoint of the faulted feeder represents the first place visited by the field crew.0 is the investment cost till the S S beginning of the first year. j is the S considered year during the planned time period. On arrival.0 (Cinv . After determination of the faulted half. j )(1 a) j (1 a ) n I rem j 1 n (5) In previous expression n is the duration of the planned period in years. TA Talert Tarrive Tisolate (3) Alert time represents the time that elapses between the moment when the fault occurs and the moment when the operator receives the information about that fault. the energy will be resupplied after time period TA which is the sum of alert time Talert. the energy will be resupplied after time period TB. STOJANOVIĆ. C M . and Trepair is repairing time of the faulted element. I rem is the remaining value of the equipment at the end of the planned period. the crew closes the previously opened sectionalizer and moves to the faulted half. Economic objective functions Objective function of actualized total cost takes into account the investment in network automation. The presented method for determination of the fault location corresponds to the network without any automation. Time needed to find and isolate the faulted section depends on the distance travelled. 2. and time needed to find and isolate the faulted section Tisolate.3. If the protection is activated again. If there is no backup supply. This time period depends on the fault location and their position with respect to that location. j is the operation and maintenance cost during the jS th year. Otherwise. j is the S outage cost during the jth year. Repeating these actions will lead to the isolation of the faulted section. and . the fault has occurred at the first half of the feeder. the crew opens the sectionalizer and disconnects the half of feeder. the fault has occurred at the second half of the feeder. operation and maintenance costs as well as the costs of unsupplied energy due to outages for a period of time. Cinv .
The cost to benefit ratio for scenario S can be represented with the following relation: IS CB S S (10) B In this expression I S is the investment cost for the scenario S. I0m is investment cost of the equipment m.m is lifetime of the equipment m.m 1 nlife. All the costs must be actualized at the same moment in time. nlife. As the second economic criterion of optimality. what can be used is the cost to benefit ratio (cost/benefit) which should be minimized. The remaining value of the element at the end of the planned period can be expressed as: S I rem I rem.m n ( I I ). C2i and C3i are coefficients whose values depend on the customer category. n nlife.m n nlife. . in addition to the equipment S S cost (Ceq). I life.m is value of the equipment m at the nth year. control equipment and telecommunication equipment if the object is introduced into the remote monitoring and control system. ΔPik is cutoff power of the customer i caused by disturbance k.. BS is benefit (total cost reduction) due to the reduction of outage cost and reduction of operation and maintenance costs for the scenario S.A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices. while benefit represents the difference between the total cost due to the current state and the total cost when automation equipment of one scenario is installed. and cpik is outage cost of the customer i per cutoff power. Equipment cost includes the prices of power equipment.m value of the equipment m at the end of lifetime. installation costs. fk is frequency of disturbance k.m .m I life. The cost includes all investment costs actualized at the beginning or at the end of the considered time period.m (6) m I rem. it is also necessary to consider the additional costs (Cad) which include designing costs. C1i . Irem.. k is disturbance index. For calculating the outage cost of customer i per cutoff power.m 0 m life. functional testing costs and commissioning costs. 173 a is the actualization rate. and Ilife. The coefficients also depend on the price of electricity and differ from one country to another. When calculating the investment cost of one scenario. d is duration of outage. i is the customer category index.m (7) In these expressions m is equipment index. The outage cost during the jth year is determined as follows: C j c pik f k Pik i k (8) In the previous expression. the following expression can be used: c pik C1i C2i d C ( €/kW ) (9) 3i In this expression.
respectively. 1. Fuzzy numbers which represent linguistic variables f i* j yj j 1 NF NF j j 1 (11) In the previous expression. Moderate and High. and a conclusion procedure. NF presents a number of fuzzy rules. In the inference procedure. The fuzzy rules are expressed in following forms: IF f i ( X ) is Low. IF f i ( X ) is Moderate. membership functions.4. Obviously. The fuzzy multicriteria evaluation [12] consists of a basic rule. THEN the plan is Moderate. All values of objective functions f i ( X ) are described using three fuzzy sets: Low. D. µj and yj are boundary value and fuzzy result in the jth fuzzy rule. THEN the plan is Good. This can simplify the computation process and completely satisfy the needs of the study. THEN the plan is Bad. the vague terms "better" and "smaller" are very imprecise in nature and can be conveniently handled by the fuzzy set theory. the real value of each objective function is firstly calculated to get the boundary values in fuzzy IFTHEN rules. 1. 0. The weighted average is applied in order to obtain real values: Fig.5 and 0. . RISTIĆ 2. The basic rule formed with fuzzy rules. TASIĆ. IF f i ( X ) is High.174 M. Consequently. The related membership functions are shown in Fig. A. Moderate and Bad are clearly defined as 1. Fuzzy multicriteria evaluation In the analyses of optimal location of distribution automation equipment. are written as pairs of "IFTHEN" statements. The values of fi* represents the fitness degree of objective functions fi for each scenario. the fuzzy sets Good. STOJANOVIĆ. the scenario is considered more preferable if reliability indicators SAIFI and SAIDI are "better" and total cost as well as cost/benefit indices are "smaller". respectively. which describe relationships in a linguistic sense.
xi(n)). lifetime of control equipment 10 years. Considering a reference sequence x0=(x0(1). Up to now. speed of field crew moving during the fault management procedure is 1 m/s. . the method has been successfully applied in many fields and has attracted many researchers who will continue exploring this study. arriving time of the field crew 2 h.. time for repairing the faulted element 2 h. where xi(k) represents the kth entry in xi.25 h. n. 3. x0(n)) and m comparative sequences xi=(xi(1).. Higher degree of relation means that the comparative sequence is more similar to the reference sequence. Grey relational analysis The grey relational analysis (GRA) is an important approach of the grey system theory in the application of evaluating a set of alternatives in terms of decision criteria. .. 175 2. annual fault rate level of lines is 0. The GRG is expressed by a scalar between 0 and 1. x0(2). The analysis is made under the following assumptions. The . the GRA can be used to measure the relationship between two sequences by calculating their correlative degrees. m. actualization rate 8%. xi (k )) max (12) max min max max max x0 (k ) xi (k ) i k (13) (14) (15) min min min x0 (k ) xi (k ) i k 0i (k ) x0 (k ) xi (k ) The GRGs between each comparative sequence xi and the reference sequence x0 can be derived from the average of the GRC. As a tool of quantitative analysis. The duration of the planned time period is 10 years.15 1/km. i= 1. k=1. 2. Supplying substation is remotely controlled. In GRA. 2. and time needed for fault isolation with remote controlled switching equipment 0. . xi(2). the data that contain same features are regarded as sequence. TEST EXAMPLE The selection of the optimal scenario for distribution automation of one small part of the network will be demonstrated on the example of real radial medium voltage 10 kV distribution network shown in Fig... The network element data and customer data are shown in Table 2.. value of the equipment at the end of lifetime is 10% from the investment cost of the equipment. 2... xi (k )) k 1 n n (16) 0i represents the degree of relation between each comparative sequence and the reference sequence... The principal formulas of the GRA are briefly described in further text..5. which is denoted as: 0i 1 ( x0 (k ). which is called grey relational grade (GRG). The grey relational coefficient (GRC) of xi with respect to x0 in kth entry is as follows: 0i ( k ) ( x0 (k ).. lifetime of power switching equipment 30 years.A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices...15 h. manipulation time 0..
Distribution test system . The coefficients used for the calculation of the outage cost for different customer categories are shown in Table 2.. in addition to the economic functions. Fig.custom..7265 1/yr. First.custom. When the proposed criteria are applied. the following criteria are used: a zone contains maximum 5 distribution substations. marked with Roman numerals in Table 1. The analysis of distribution network before installation of the automation equipment leads to the following values: SAIFI10 7. and total cost 150. RISTIĆ total cost and benefit for the planned time period will be actualized to starting year. can be identified. if the fault is not isolated. The locations of the already installed switches are also taken into account when defining the service zones. Again. If the fault is not isolated. 2. the crew should go to the location of the switching device S5 and manipulate it. TASIĆ. STOJANOVIĆ.176 M. the average real power of all residential category customers that belong to the zone (Pav res) and the average real power of all commercial customers that belong to the zone (Pav com). eight zones on considered feeder. the crew should go to the location of the switching device S2 and manipulate it. the field crew should go to the switching device S3 and open it. Firstly. an analysis is conducted over the current state. For division of distribution network into service zones. and outage cost is evaluated. D. The following path of movement is assumed for the field crew. SAIDI10 14. A. the reliability indicators SAIFI and SAIDI are used. As objective functions.560 €. Table 3 shows summary data about the zones and these data are: the number of distribution substations that belong to the zone (Ndt). reliability indicators are determined.13 h/yr. the length of all branches in the zone is more than 2 km. the sum of rated power of transformers that belong to the zone (SrT). during fault isolating in the current state network (before automation).
1 15.207 0.5 2 To show the results obtained by the application of multicriteria fuzzygrey evaluation.669 7.24 0. In this table S represents the switch.355 0 1.118 11. 177 Table 1.3 C2 7 1.067 0 1.863 39.705 37.876 17.479 21. while R represents remote control.168 6.499 21.771 13.412 5. Test system data Line 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Zone I I I I I II II II II II II II III III III III III IV IV IV IV IV V VI VII VII VII VII VIII VIII VIII Length 1220 870 350 360 860 405 690 1040 300 1025 520 30 665 770 310 810 550 365 1625 650 310 50 1320 375 330 1420 920 290 1940 510 810 Line 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Zone I I I I II II II II II III III III III III IV IV IV V V VI VI VI VII VII VII VIII VIII VIII VIII Length 240 500 115 80 60 350 85 435 195 50 540 340 885 20 155 850 350 410 280 500 520 900 120 70 45 100 40 290 1295 Sr 160 160 250 100 160 160 160 250 250 160 250 250 45 50 45 160 160 160 100 160 160 250 250 100 160 160 160 100 160 Pav res 75.333 31.566 1..439 0 0 0 0.058 3.761 19.164 23.136 0.368 3.3 0.105 0. D represents the fault detector.2 0 0 0 0 1.97 0.78 0.975 23.11 C3 1 1 2.175 55.393 14.568 0.648 1.151 20.536 12.537 0 Table 2. .A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices.103 6.65 0.58 1..23 0.315 13.544 15.645 7.839 22.943 10.23 0 0 1.175 14. C represents the recloser.611 2.3 h ≤ d < 1 h 1h≤d<5h 5 h ≤ d ≤ 48 h C1 1. 8 scenarios out of the set of scenarios with approximately the same automation level are taken and shown in Table 4.331 Pav com 2.821 8.581 16. Coefficients for outage cost evaluation Customer category Commercial Residential 0.
calculated values of outage cost. A potential solution is also scenario 8. a different ranking of potential solutions can be obtained.596 5.98 76.451 26. it is obvious that total cost function has its minimum for scenario 5. f3c = 6. f3b = 4.153 4.323 4.299 48. which demands slightly higher value of investment and has worse cost to benefit ratio.295 3.258 1. grey relational grade is calculated using expressions (13)(16) and scenarios are ranked as shown in the last row of Table 5. The optimal solution is represented with scenario 2 which has the investment value of 20.195 4. it has a lower value of reliability indicator SAIFI10 than scenario 2.04 87. but on the other hand.35 and f2c = 0. and estimated value of telecommunication equipment infrastructure cost).438 0 0. f2a = 0.3.4. Summary data about service zones Zone Nds Lline [km] SrT [kVA] Pav res Pav com I II III IV V VI VII VIII 4 5 5 3 2 3 3 4 4. technical documentation cost.5.504 5. f3a = 3.422 64.01 2.985 670 980 755 365 260 570 510 580 145. RISTIĆ Table 3. but the investment is the greatest.135 4. f4b = 8 and f4c = 10.200 €.127 11. On the basis of the objective functions shown in Table 5. Scenarios 2 and 3 have approximately the same value of total cost function but scenario 2 has slightly lower value of cost to benefit ratio.25. functional testing and commissioning cost.869 62. D.576 0. the investment cost of the equipment for every scenario. f2b = 0.984 Table 4. f1c = 1. STOJANOVIĆ. A. TASIĆ.05. obtained profit from network automation in the 10year time period and cost to benefit ratio (cost/benefit).15. training cost.178 M. Distribution automation equipment for eight selected scenarios Scenario\Zone 1 2 3 4 5 6 7 8 I RC RC RC RC RC RC RC RC II S S S RSD RSD S SD S III RSD RSD RSD RSD RSD RC RC RC IV RD RD RD RD V S S S S S S S S VI RSD RSD RSD RSD RSD VII VIII S S SD S S S SD S S Table 5 shows objective function values for eight selected scenarios. then with a modification of the characteristic values of fuzzy numbers. .269 57. f4a = 6. f1b = 1. as well as additional costs (installation cost. If investment for the obtained optimal solution deviates much from the planned value.355 2. The table also contains calculated values of reliability indicators SAIFI10 and SAIDI10 which represents the impact of faults on 10 kV network on supply reliability. the following characteristic values of fuzzy numbers for selected objective functions are used: f1a = 1. For fuzzy multicriteria evaluation shown in Table 5.94 4. Now.
R. J. 16461651. . D. N.484 0. October 1994. and the values of the selected objective functions are determined.7265 8.7856 1 3 15500 6100 2270 83960 1. Glamocic and M. pp.8399 0 0. vol. 11(3).8399 0.259 0. "The optimal automation level of medium voltage distribution networks".4245 0.3401 3. August 2006.78 0.792 0. 1995.7131 0. H. H. D. M. CONCLUSION A method for selection of the optimal scenario for distribution automation is proposed in this paper.4245 0.4245 0. 3.7808 0. Popovic. G. 33(3). Because of this. J. Final Report.674 4 6 13000 4500 2450 113375 1. Teng and C.7265 8. IEEE Transaction on Power Delivery.4803 8. March 2011.4771 0. vol. S.3243 4. Electrical Power and Energy Systems. 31(2).1112 0.4385 6 2 14500 5700 2250 82873 1.38 0. the method uses fuzzy multicriteria evaluation and grey relational analysis in the application of evaluating a set of alternatives in terms of decision criteria..0938 0.24 0 0 0. vol. After that. MazelTov and D.5986 0. IEEE Transactions on Power Systems.3042 8 7 16000 5700 2320 89892 1.S. nonlinear and discrete optimization problem of enormous dimensions and it is not possible to apply an appropriate optimization procedure directly.2725 0.7567 0. Levitin. Y.3079 7 5 20200 8500 2640 76758 1 0.. January 2002.A MultiObjective Approach to Optimal Allocation of Distribution Automation Devices. Elmakis. Fault management in electrical distribution systems. Automation scenarios are ranked on the basis of objective function values. 97102.042 0.4245 0. Chuang. pp. pp. Then. pp. The problem of distribution automation is a complex. CIRED Working Group WG03. Lu.49 0.011 0. S. 2006. 4.4706 3. Li. Objective functions and fuzzygrey evaluation for eight selected scenarios Scenario Ceq Cad Irem Ctot f1 (Ctot/Ctot min) f2 (C/B) f3 (SAIFI10) f3 (SAIDI10) f1* f2* f3* f4* 0i Rank 1 13500 5300 2220 95217 1.8517 1 0. Bilington and S Jonnavithula.2984 4.5162 4. Huang and C.278 0 0 0. 21(3). Control and automation of electrical power distribution systems. 179 Table 5. Wilson. 2.7409 2 4. 12091217. Huang. 5. Lin.6033 5 8 16000 6550 2450 85000 1.867 1 0. C. C. pp. C. July 1996. Nimrihter. all possible automation scenarios are searched. 430438.7265 6. The method is demonstrated on the test example of a real medium voltage distribution network.7265 8.709 3 4 19200 8100 2160 97673 1.0476 0.44 0. Chen.1074 0. IEEE Transactions on Power Delivery. H.8399 0. "Optimal sectionalizer allocation in electric distribution systems by genetic algorithm". 6.3 0. N. R.4803 8.3944 0.2405 0.7265 8.5 0. CRC Press Taylor & Francis.3397 4. 254259.4231 0.1711 0. 17(1). S. J.0797 0.4803 10. 7. Green and R.3889 4. Electric Power System Research. "Optimal switching device placement in radial distribution systems". Lj. the proposed method firstly determines the service zones based on the heuristic rules for a part of distribution network that can be considered as the smallest functional unit.4245 0. vol. vol.3577 3. V. "Optimal placement of line switches for distribution automation systems using immune algorithm".603 0. REFERENCES 1. "Feederswitch relocation for customer interruption cost minimization".43 0.
199205.H. Abbad and J. A. p. S. 14. Ferreira and A. "The integration and application of fuzzy and grey modelling methods". vol. F.S. koji su nakon toga sagledani kao višekriterijumski optimizacioni zadatak. nelinearni diskretni optimizacioni problem izuzetno velike dimenzionalnosti zbog čega je praktično nemoguća direktna primena optimizacionih procedura. Kuo. pp. Huang. višekriterijumska optimizacija . S. L. 107–119. July 1988. 10(3). 20(2). 20. 1156–1163. March 2008. Vol. 12091217. Fuzzy Sets and Systems. Lee. February 2005. "Optimal switch placement in distribution system using trinary particle swarm optimization algorithm ". 16. February 1996. J. P. pp. P. 13(3). 13. C Silva.p. M. M. January 2008. RISTIĆ 8. srednjenaponska distributivna mreža. Ključne reči: automatizacija. D. "Reliability of distribution systems ". "Optimized placement of control and protective devices in electric distribution systems through reactive tabu search algorithm". 4TH Benchmarking Report on Quality of Electricity Supply. Pereira. "A decomposition approach to optimal remote controlled switch allocation in distribution systems ". Dragan Tasić. Ahn. 190200. A. vol. 10311036. 21(3). A. Tsai and H. Electric Power System Research. W. vol. 11. Problem automatizacije distributivne mreže je složen. IEEE Transactions on Power Systems. M. Billinton. FotuhiFiruzabad. TASIĆ. J. 78(4). J. "Distribution system restoration using the hybrid fuzzygrey method". S. 2009. Mijailović. R. "Distribution system reliability performance and evaluation". L. Aleksa Ristić U radu je predstavljen jedan pristup izboru optimalne vrste opreme za automatizaciju distributivne mreže. J. G. Nahman and V. S. S. i njene optimalne lokacije.180 M. Zbog toga je u radu najpre predložena podela problema na podprobleme. No. pp. August 1998. 15. pp. Electrical Power and Energy Systems. Prezetovani metod testiran je na primeru realne radijalne srednjenaponske distributivne mreže. R. 78(1). Carvalho. Moradi and M. 12. 10. pp. Chen. vol. Za donošenje odluke primenjeni su metodi višekriterijumskog fuzzy ocenjivanja i "grey" relacionog rangiranja. Huang and C.L. R. A. C. pp. in Serbian. IEEE Transaction on Power Systems. Council of European Energy Regulators. F. IEEE Transactions on Power Delivery. IEEE Transactions on Power Delivery. R. Matovani. April 2005. 372381. Akademska misao. Lim and B. W. "Service restoration of primary distribution systems based on fuzzy evaluation of multicriteria". Y. Silva. 9. STOJANOVIĆ. 2008. vol. vol. VIŠEKRITERIJUMSKI PRISTUP ODREĐIVANJU OPTIMALNE LOKACIJA UREĐAJA DISTRIBUTIVNE AUTOMATIKE U RADIJALNOJ SREDNJENAPONSKOJ DISTRIBUTIVNOJ MREŽI Miodrag Stojanović. 1.
This action might not be possible to undo. Are you sure you want to continue?
Use one of your book credits to continue reading from where you left off, or restart the preview.