You are on page 1of 6

Prioritization of Load Points in Distribution System

considering Multiple Load Types using Fuzzy


Theory
Galiveeti Hemakumar Reddy, Pranju Chakrapani, *Arup Kumr Goswami and Nalin B Dev Choudhury
Department of Electrical Engineering
National Institute of Technology Silchar, Silchar
Assam, India, 788010
Email: *gosarup@gmail.com

Abstract—The power shortages and/or outages in the distri- is used to illustrate the priority of load in different feeders.
bution system enables the need of load shedding or curtailment, Loads in a microgrid are classified into three different groups
where the important loads are given high priority. This paper namely must run (health care), discretionary (heating and
presents a novel method to solve the load point prioritization
problem. The load points in the distribution system contains air-conditioning etc.) and emergency shedding (residential,
different types of loads and each load type has certain importance commercial etc.) loads [7].
over other load type. The inclusion of load types in the load Authors in [8], solved restoration problem as a multiobjec-
point ranking, increases the complexity in the problem. The tive problem and priority customers is one of the constraints.
proposed ranking method based on the fuzzy theory resolves Load curtailment is applied on in-service customers to improve
this complexity and evaluates the weights of load points and
thereupon ranking. The proposed method is validated on a test the service restoration and increase the number of customers
system considering a natural disaster situation. served [9]. Multi-agent systems is presented in [10] for restora-
tion including the load prioritization.
I. I NTRODUCTION From the literature, it is found that the load prioritization is
Equipment failures in the distribution system may discon- very important in many ways. The load weights and/or load
nect the loads from the grid supply and which leads to priority levels are used for the purpose of system restoration
formation of microgrids by using the distributed generation during the system failures, load shedding and/or curtailment
available in that part of the system. Load demand in the newly during power shortages and during restoration process. These
formed islanded microgrid exceeds the generation, it requires are increasing the necessity of methods to evaluate the weights
load shedding and/or load curtailment. While applying load of load points.
shedding/curtailment, important (or critical) loads are given The evaluation of weights of load points is unclear in [3],
high priority. Distribution system faces multiple outages i.e. [4], [5], [6] and no formal method is presented. A method
affected by natural disasters, the restoration of the system is is presented in [11] to evaluate the weights of load points
to be done in such a way that important loads are restored based on the current rating of a load point. The weights, as
first. All this process depends on the prioritization of loads as mentioned in [12], are used in an objective function to increase
per their importance. the critical load pick up. There are, however, certain limitations
The concept of priority loads is used for load shedding, in the aforementioned method, for e.g. the volume of load
load curtailment and restoration of the system. Authors in cannot ascertain its importance and depends on what type of
[1] have considered two types of loads i.e. critical loads load it is. The importance of a particular load type over the
and controllable/adjustable loads during the restoration of the other types of loads cannot be accumulated in this method. In
system. Where, the critical loads are given high priority during general, the load points (under one distribution transformer)
restoration process and load scheduling and/or curtailment is contains several types of loads, it further increases the load
applied on controllable loads based on the available power point prioritization problem.
generation. The same concept is used in [2] to prevent stalling To overcome all the complexities in finding the priority of
of distributed generation in a microgrid. load points, fuzzy theory based approach is presented in this
The critical load restoration problem is solved in [3] con- paper i.e. fuzzy multi criteria decision making, here, each load
sidering natural disasters where weights are assigned for each type is a criterion to decide the importance of a load point and
load point. In ref [4], [5], customers in the distribution system its prioritization.
are distinguished load points based on their priority level
and each priority level is associated with a weight factor. II. L OADS IN DISTRIBUTION SYSTEM
Load prioritization concept is used to evaluate the reliability Distribution systems serves low voltage single and/or three
of island mode microgrids [6], where a prioritization matrix phase consumers. Consumers differ from each other based on

978-1-5090-6034-4/17/$31.00 2017
c IEEE
their role in the daily life. The different types of loads are as TABLE I
follows S UBJECTIVE OPINIONS AND RELATED POSITIVE NUMBERS FOR LOAD
TYPE WEIGHTS
∙ Health care, this type of loads includes hospitals, nursing
houses, diagnosis centers and blood bank etc. As a Subjective opinion Positive number
concern of human life, these loads are very important Not preferred (Np) 0.2
and are served without load shedding or curtailment. Less preferred (Lp) 0.4
∙ Public utilities, these type of loads provide common Equally preferred (Ep) 1
facilities/services to the people like fuel stations, banks, More preferred (Mp) 2.5
water supply systems and govt. offices etc. Highly preferred (Hp) 5
∙ Grocery, these loads which are associated with food falls
under this category which includes, vegetable markets,
food storage and super markets etc. types. The diagonal elements in matrix 𝐵 are equal to 1 and
∙ Residential loads, the load curtailment is easily applicable non-diagonal elements are calculated as follows
for these type of loads as consumers can be compromised 𝐾
with their usage to get the incentives. 1 ∑ 𝑡
𝑏𝑖𝑗 = 𝑏
∙ Commercial loads, this category of loads includes com- 𝐾 𝑡=1 𝑖𝑗
(2)
mercial buildings, shopping malls and entertainment ac- 1
tivities etc. These loads are easily scheduled during any 𝑏𝑗𝑖 =
𝑏𝑖𝑗
emergency situations like natural disasters.
here, 𝑏𝑡𝑖𝑗 is the subjective opinion given by 𝑡𝑡ℎ expert.
III. P RIORITIZATION OF LOAD POINTS The weights of each load type is obtained by using a
The load points in the distribution system are prioritized geometric mean method to define the fuzzy mean number
based on their importance/significance. During a power short- corresponding to a row 𝑒𝑗 of 𝐵 matrix and determination of
age/outage, based on the priority of load points, load shedding weights are as follows:
and/or curtailment is applied to match the balance between
demand and generation in the system to ensure smooth opera- 𝑒𝑗 = (𝑏𝑗1 (.)𝑏𝑗2 (.)...(.)𝑏𝑗𝑚 )1/𝑚
(3)
tion. If a load point has only one type of load then load point 𝑤𝑗 = 𝑒𝑗 /(𝑒1 (+)𝑒2 (+)...(+)𝑒𝑚 )
prioritization can be done easily, but, in practical distribution
system, most of the the load points are a mix of different The weights are represented as triangular fuzzy number
load types hence load prioritization becomes complicated. The ˜𝑗 = (𝑤𝑗𝑙 , 𝑤𝑗𝑚 , 𝑤𝑗𝑟 )) and the fuzzy number is determined
(𝑤
importance of a load type over other load types varies with by
different situations and people has different opinions on them. 𝑤𝑗𝑚 = 𝑤𝑗
All these increases the complexity in prioritization of load 𝑤𝑗𝑙 = 𝑤𝑗𝑚 − 𝜀𝑙 (4)
points. 𝑤𝑗𝑟 = 𝑤𝑗𝑚 + 𝜀𝑟
A. Proposed fuzzy ranking approach here, 𝜀𝑙 and 𝜀𝑟 are the spreads to the left hand and right
In this paper, the load point prioritization problem is consid- hand side of triangular membership function respectively.
ered as multi criteria decision making, where, each load type is In second stage, the ranking of load points is solved using
a criterion. The preference of a load type over other load types fuzzy multi criteria decision making (FMCDM) [14]. A load
and importance of a load point from the view of each load point is rated with respect to the load types by a set of 𝐾
type are subjective in nature, hence fuzzy approach is taken experts as subjective ratings such as: not applicable, very
to include the subjective opinions/ratings. Now, the solution is less, less, fair, high and very high. Each subjective rating is
named as fuzzy multi criteria decision making (FMCDM). associated with a triangular fuzzy number as given in Table
The load point prioritization problem is solved in two stages. II.
In the first stage, the preference of a load type over other
load types are collected by a set of 𝐾 experts, the opinions TABLE II
are as follows, not preferred, less preferred, equally preferred, S UBJECTIVE RATINGS AND ASSOCIATED FUZZY NUMBERS FOR LOAD
POINT RATING
more preferred, and highly preferred. Each subjective opinion
is associated with a positive number as given in Table I. Subjective rating Fuzzy number
The positive number associated with subjective opinions are Not Applicable (NA) (0, 0, 0)
used to evaluate the fuzzy reciprocal matrix (𝐵) [13] Very Low (VL) (0, 0, 3)
Low (L) (1, 3, 5)
𝐵 = [𝑏𝑖𝑗 ]𝑚×𝑚 (1) Fair (F) (3, 5, 7)
High (H) (5, 7, 9)
where 𝑏𝑖𝑗 is a positive number given to assign priority of 𝑖𝑡ℎ
load type over 𝑗 𝑡ℎ load type. 𝑚 is the total number of load Very High (VH) (7, 10, 10)
𝐾
1 ∑ 𝑡 IV. N UMERICAL EXAMPLE
𝑥
˜𝑖𝑗 = 𝑥
˜
𝐾 𝑡=1 𝑖𝑗 (5) No existing test system has been found suitable to validate
˜ = [˜
𝐷 𝑥𝑖,𝑗 ]𝑙×𝑚 the proposed load point prioritization method. In this paper, to
validate the proposed method, a sample system is developed
here, 𝑙 = total number of load point
with necessary data and is shown in Fig. 1. It is a radial feeder
˜𝑡𝑖𝑗 = rating given by 𝑡𝑡ℎ expert on 𝑖𝑡ℎ load point from the
𝑥
having six load points (LP1, LP2, LP3, LP4, LP5 and LP6).
view of 𝑗 𝑡ℎ load type.
˜ is normalised by a linear scale transformation The loading details of each load point is given in Table III and
The matrix 𝐷
the values represents the percentage of each load type over the
[14] as follows.
total load in that load point.
𝑅˜ = [˜
𝑟𝑖𝑗 ]𝑙×𝑚
( )
𝑥𝑙𝑖𝑗 𝑥𝑚 𝑟
𝑖𝑗 𝑥𝑖𝑗 (6)
𝑟˜𝑖𝑗 = ∗ , ∗, ∗
𝑐𝑗 𝑐𝑗 𝑐𝑗
∗ Substation
where, 𝑥 ˜𝑖𝑗 = (𝑥𝑙𝑖𝑗 , 𝑥𝑚 𝑙 𝑟
𝑖𝑗 , 𝑥𝑖𝑗 ) and 𝑐𝑗 = 𝑚𝑎𝑥(𝑥𝑖𝑗 ) LP1 LP2 LP3 LP4 LP5 LP6

An ideal solution is generated for each load type 𝑟˜0𝑗 =
˜
(0, 1, 0). Now, the Hamming distance (𝑑𝑖𝑗 ) is calculated by Fig. 1. Sample radial distribution feeder

𝑑˜𝑖𝑗 = (˜∗
𝑟0𝑗 − 𝑟˜𝑖𝑗 ) (7)
TABLE III
The Fuzzy Grey Relational Coefficient (FGRC) of each load L OAD POINT LOADING DETAILS IN PERCENTAGE
point is obtained as follows [14]:
𝑚
∑ Load Health Public Food
𝑚𝑖𝑛𝑖 𝑚𝑖𝑛𝑗 𝑑˜𝑖𝑗 + 𝜁𝑚𝑎𝑥𝑖 𝑚𝑎𝑥𝑗 𝑑˜𝑖𝑗 point care utilities storage
Residential Commercial
𝜏˜0,𝑖 = 𝑤
˜𝑗 (.) (8)
𝑗=1
(𝑑˜𝑖𝑗 + 𝜁𝑚𝑎𝑥𝑖 𝑚𝑎𝑥𝑗 𝑑˜𝑖𝑗 ) LP1 10 35 15 10 30
LP2 10 10 10 60 10
here, 𝜁 is the resolving coefficient and 𝜁 ∈ [0, 1].
LP3 0 10 30 40 20
Thereupon, the preference relationships between load points
LP4 50 5 10 15 20
(𝑍˜𝑖𝑗 ) are obtained by fuzzy difference between 𝜏˜0,𝑖 and 𝜏˜0,𝑗
LP5 20 5 20 40 15
and is calculated by
LP6 10 5 5 20 60
𝑍˜𝑖𝑗 = 𝜏˜0,𝑖 (−)˜
𝜏0,𝑗 (9)
𝑍˜𝑖𝑗 is also a fuzzy number and an 𝛼 cut is written as 𝑍˜𝑖𝑗
𝛼
= Firstly, weight of each load type is to be determined. For
𝛼 𝛼
[𝑍𝑖𝑗𝑙 , 𝑍𝑖𝑗𝑟 ]. this purpose, subjective opinions are collected for importance
of each load type over other load types. The importance of a
If 𝑍˜𝑖𝑗𝑟
𝛼
> 0 and 𝑍˜𝑖𝑗𝑙
𝛼
< 0 with some 𝛼 values, then fuzzy
load type varies with different situations, here in this paper,
preference relation matrix (𝐸) is obtained as follows:
the work considers a natural disaster situation. The subjective
𝑆1 opinions collected from three experts and are given Table IV
𝑒𝑖𝑗 = ; 𝑆1 + 𝑆2 > 0
𝑆1 + 𝑆2 (10)
𝐸 = [𝑒𝑖𝑗 ]𝑙×𝑙 TABLE IV
∫ ∫ S UBJECTIVE OPINIONS AND RELATED POSITIVE NUMBERS FOR LOAD
where, 𝑆1 = 𝑥>0 𝜇𝑍𝑖𝑗 (𝑥)𝑑𝑥, and 𝑆2 = 𝑥<0 𝜇𝑍𝑖𝑗 (𝑥)𝑑𝑥. TYPE WEIGHTS
And the fuzzy strict preference relation matrix (𝐸 𝑠 ) is
load type Expert LT1 LT2 LT3 LT4 LT5
obtained, for strict comparison of load points, 𝑒𝑠𝑖𝑗 is calculated
E1 Ep Hp Hp Hp Hp
as {
LT1 E2 Ep Hp Mp Ep Mp
𝑠 𝑒𝑖𝑗 − 𝑒𝑗𝑖 ; 𝑒𝑖𝑗 ≥ 𝑒𝑗𝑖
𝑒𝑖𝑗 = (11) E3 Ep Mp Mp Hp Hp
0 ; otherwise E1 Np Ep Mp Ep Mp
LT2 E2 Np Ep Ep Mp Hp
𝐸 𝑠 = [𝑒𝑠𝑖𝑗 ]𝑙×𝑙 (12) E3 Lp Ep Lp Ep Mp
A non dominated degree is calculated to obtain a weight of E1 Np Lp Ep Ep Mp
load point without comparison with other load points, LT3 E2 Lp EP Ep Mp Mp
E3 Lp Mp Ep Mp Hp
𝜇𝑁 𝐷 (𝐿𝑃𝑖 ) = 𝑚𝑖𝑛{1 − 𝑒𝑠𝑗𝑖 }for 𝑗𝜖(1, 2, .., 𝑘) and 𝑗 ∕= 𝑖 E1 Np Ep Ep Ep Mp
(13) LT4 E2 Ep Lp Lp Ep Hp
where, 𝐿𝑃𝑖 is the 𝑖𝑡ℎ load point E3 Np Ep Lp Ep Mp
A non dominated degree states an absolute ranking of load E1 Np Lp Lp Lp Ep
points without getting into a comparison with other load LT5 E2 Lp Np Lp Np Ep
points. E3 Np Lp Np Lp Ep
In Table IV, the terms LT1, LT2, LT3, LT4 and LT5 experts and are shown in Table VI. The experts should be
denotes the load types i.e. healthe care, public utilities, grocery, selected carefully, so that the collected information reflects
residential and commercial respectively. the opinions of all consumers. However, the data used in this
The fuzzy reciprocal matrix is formulated using eq. (2) paper, is for illustration purpose only and varies accordingly
through the information provided in Table IV and Table I.The with the system is location and consumers behavior.
matrix 𝐵 is given below
TABLE VI
⎡ ⎤ S UBJECTIVE OPINIONS AND RELATED POSITIVE NUMBERS FOR LOAD
1 4.17 3.33 3.67 4.17 TYPE WEIGHTS
⎢0.24 1 1.3 1.5 3.337⎥
⎢ ⎥ load point Expert LT1 LT2 LT3 LT4 LT5
𝐵=⎢
⎢ 0.3 0.77 1 2 3.337⎥⎥
⎣0.27 E1 L H F L H
0.67 0.5 1 3.33 ⎦
LP1 E2 F VH H L F
0.24 0.3 0.3 0.3 1
E3 L H F F H
E1 L L F VH L
The weights of all the load types are determined using LP2 E2 F L F VH L
eq. (3) and these weights are converted into triangular fuzzy E3 L F L VH VL
numbers using eq. (4), where, 𝜀𝑟 and 𝜀𝑙 are taken as 10% of E1 NA L H H H
𝑤𝑗 . Load type, weight and related triangular fuzzy number are LP3 E2 NA L VH VH F
given in Table V. E3 NA F H H H
E1 VH VL F F H
TABLE V LP4 E2 VH L F F F
W EIGHTS AND FUZZY NUMBERS OF LOAD TYPES E3 VH L F L H
E1 F VL F H L
load type Weight Fuzzy number
LP5 E2 H L H VH VL
Health care 0.4668 (0.42,0.4668,0.5134)
E3 F L F H L
Public utilities 0.1747 (0.1572,0.1747,0.1922)
E1 L VL L F VH
Grocery 0.1742 (0.1568,0.1742,0.1916)
LP6 E2 F L L F VH
Residential 0.1259 (0.1133,0.1259,0.1384) E3 L L L H VH
Commercial 0.0584 (0.0525,0.0583,0.0641)

The fuzzy decision matrix 𝐷 is formulated using the


The weight values given Table V reflects the practical information in Table VI and eq. (5). The matrix 𝐷 is then
situation and reasons are as follows normalized using eq. (6). Now, hamming distance is evaluated
1) After a natural disaster, the health care should be a using eq. (7) and hamming distance matrix 𝐻 is formulated
high priority to treat the injured people and control the using these values. The fuzzy grey relation coefficients are
spreading of diseases. So, these loads got very high evaluated using weights of load types, hamming distance
priority for restoration purpose. matrix and eq. (8). Fuzzy preference relationship between
2) The type of loads included in public utilities are helpful load points is evaluated using eq. (9). These relationships are
for bringing back the normal conditions and the routine used to find fuzzy preference relation matrix 𝐸 and elements
life of people. On the other hand, grocery loads are very are evaluated using eq. (10) and this matrix is given below
important to supply food to people. As such both public ⎡ ⎤
utilities and grocery increases the complexity in deciding 0.5000 0.5467 0.5985 0.3494 0.4892 0.5981
the priority level that should be given to them. However, ⎢0.4533 0.5000 0.5525 0.3065 0.4425 0.5543⎥
⎢ ⎥
the proposed method gives almost equal priority to both ⎢0.4015 0.4475 0.5000 0.2591 0.3905 0.5048⎥
𝐸=⎢ ⎢0.6506 0.6935 0.7409 0.5000 0.6405 0.7343⎥

the type of loads. But, a bit skewed towards public ⎢ ⎥
utilities. ⎣0.5108 0.5575 0.6095 0.3595 0.5000 0.6086⎦
3) In comparison to residential loads, commercial loads are 0.4019 0.4457 0.4952 0.2657 0.3914 0.5000
given less priority, because of the moral obligation of For a strict comparison of load points, fuzzy strict
system operators to fulfill the basic needs of consumers preference relation matrix 𝐸 𝑠 is formulated using eq. (11)
rather than providing entertainment and/or other com- and matrix is as follows
mercial activities. ⎡ ⎤
The conversion of weights into triangular fuzzy numbers is 0 0.0934 0.1970 0 0 0.1961
⎢ 0 0 0.1051 0 0 0.1086⎥
done to accommodate the uncertainty in the experts’ opinion ⎢ ⎥
⎢ 0 0 0 0 0 0.0097⎥
and make it more realistic. 𝐸𝑠 = ⎢ ⎢0.3011 0.3871 0.4818 0 0.2810 0.4686⎥

In second stage, subjective ratings for a load point from ⎢ ⎥
⎣0.0216 0.1151 0.2190 0 0 0.2171⎦
the view of each load type is required to prioritize the load
points. Here also subjective ratings are collected from three 0 0 0 0 0 0
⎡ ⎤
(1.67, 3.67, 5.67) (5.67, 8, 9.33) (3.67, 5.67, 7.67) (1.67, 3.67, 5.67) (4.33, 6.33, 8.33)
⎢(1.67, 3.67, 5.67) (1.67, 3.67, 5.67) (2.33, 4.33, 6.33) (7, 10, 10) (0.67, 2, 4.33) ⎥
⎢ ⎥
⎢ (0, 0, 0) (1.67, 3.67, 5.67) (5.67, 8, 9.33) (5.67, 8, 9.33) (4.33, 6.33, 8.33)⎥
𝐷=⎢
⎢ (7, 10, 10)

⎢ (0.67, 2, 4.33) (3, 5, 7) (2.33, 4.33, 6.33) (4.33, 6.33, 8.33)⎥⎥
⎣(3.67, 5.67, 7.67) (0.67, 2, 4.33) (3.67, 5.67, 7.67) (5.67, 8, 9.33) (0.67, 2, 4.33) ⎦
(1.67, 3.67, 5.67) (0.67, 2, 4.33) (1, 3, 5) (3.67, 5.67, 7.67) (7, 10, 10)

⎡ ⎤
(0.17, 0.37, 0.57) (0.61, 0.86, 1) (0.39, 0.61, 0.82) (0.17, 0.37, 0.57) (0.43, 0.63, 0.83)
⎢(0.17, 0.37, 0.57) (0.18, 0.39, 0.61) (0.25, 0.46, 0.68) (0.7, 1, 1) (0.07, 0.2, 0.43) ⎥
⎢ ⎥
⎢ (0, 0, 0) (0.18, 0.39, 0.61) (0.61, 0.86, 1) (0.57, 0.8, 0.93) (0.43, 0.63, 0.83)⎥
𝑅=⎢


⎢ (0.7, 1, 1) (0.07, 0.21, 0.46) (0.32, 0.54, 0.75) (0.23, 0.43, 0.63) (0.43, 0.63, 0.83)⎥⎥
⎣(0.37, 0.57, 0.77) (0.07, 0.21, 0.46) (0.39, 0.61, 0.82) (0.57, 0.8, 0.93) (0.07, 0.2, 0.43) ⎦
(0.17, 0.37, 0.57) (0.07, 0.21, 0.46) (0.11, 0.32, 0.54) (0.37, 0.57, 0.77) (0.7, 1, 1)

⎡ ⎤
(0.43, 0.63, 0.83) (0, 0.14, 0.39) (0.18, 0.39, 0.61) (0.43, 0.63, 0.83) (0.17, 0.37, 0.57)
⎢(0.43, 0.63, 0.83) (0.39, 0.61, 0.82) (0.32, 0.54, 0.75) (0, 0, 0.3) (0.57, 0.8, 0.93) ⎥
⎢ ⎥
⎢ (1, 1, 1) (0.39, 0.61, 0.82) (0, 0.14, 0.39) (0.07, 0.2, 0.43) (0.17, 0.37, 0.57)⎥
𝐻=⎢


⎢ (0, 0, 0.3) (0.54, 0.79, 0.93) (0.25, 0.46, 0.68) (0.370.570.77) (0.170.370.57) ⎥ ⎥
⎣(0.23, 0.43, 0.63) (0.54, 0.79, 0.93) (0.18, 0.39, 0.61) (0.07, 0.2, 0.43) (0.57, 0.8, 0.93) ⎦
(0.43, 0.63, 0.83) (0.54, 0.79, 0.93) (0.46, 0.68, 0.89) (0.23, 0.43, 0.63) (0, 0, 0.3)

⎡ ⎤
(0.55, 0.61, 0.91) (0.72, 0.88, 1.3) (0.62, 0.72, 1.1) (0.55, 0.61, 0.91) (0.64, 0.73, 1.11)
⎢(0.55, 0.61, 0.91) (0.55, 0.62, 0.93) (0.57, 0.65, 0.98) (0.77, 1, 1.3) (0.52, 0.56, 0.83)⎥
⎢ ⎥
⎢ (0.5, 0.5, 0.65) (0.55, 0.62, 0.93) (0.72, 0.88, 1.3) (0.69, 0.83, 1.22) (0.64, 0.73, 1.11)⎥
𝐹 𝐺𝑅𝐶 = ⎢
⎢ (0.77, 1, 1.3)

⎢ (0.52, 0.56, 0.85) (0.59, 0.68, 1.04) (0.57, 0.64, 0.95) (0.64, 0.73, 1.11)⎥⎥
⎣(0.61, 0.69, 1.05) (0.52, 0.56, 0.85) (0.62, 0.72, 1.1) (0.69, 0.83, 1.22) (0.52, 0.56, 0.83)⎦
(0.55, 0.61, 0.91) (0.52, 0.56, 0.85) (0.53, 0.6, 0.89) (0.61, 0.69, 1.05) (0.77, 1, 1.3)

The non-dominated weights are determined using eq. (13) ∙ As health care loads are highly preferred, LP4 has got
and are used to rank load points without any further compar- high amount of load and has got more weight.
isons. The weight value represents the importance of a load ∙ LP5 is next to the LP4 as it has considerable amount of
point, where the higher weight value denotes higher priority health care loads.
and vice versa. The weights and ranks of load points are given ∙ Since LP1, LP2 and LP6 have the same amount of health
in Table VII. care loads, the ranking would done depend on the other
load types. As LP1 has more loads in public utility and
TABLE VII
grocery loads than remaining. The LP1 is given first
S UBJECTIVE OPINIONS AND RELATED POSITIVE NUMBERS FOR LOAD priority followed by LP2 and LP6 respectively.
TYPE WEIGHTS ∙ LP3 has got the least priority, because it doesn’t have any
health care loads.
Load point weight Rank
LP1 0.6989 3 V. C ONCLUSION
LP2 0.6129 4
LP3 0.5182 6 This paper proposed a new approach to evaluate the weights
LP4 1 1 of load points in the distribution system based on the fuzzy
LP5 0.7190 2 theory. Different load types are taken into consideration rather
LP6 0.5314 5 than the volume of total load in the load point for evaluation of
weights and is done using fuzzy multi criteria decision making,
where, each load type is a criterion for deciding the importance
The weights and ranks are analyzed by considering the load of a load point. The fuzzy approach effectively handles the
details given in Table III and weights of load types given in subjective opinions and ratings given by experts towards the
Table V. The observations are as follows prioritization of load points. The result section witnesses that
the proposed method determines the weights by respecting the [7] B. Moran, “Microgrid load management and control strategies,” in 2016
ground reality i.e. experts opinions on different load type and IEEE/PES Transmission and Distribution Conference and Exposition (T
D), May 2016, pp. 1–4.
the volume of different load types in a load point. [8] Y. Kumar, B. Das, and J. Sharma, “Multiobjective, multiconstraint
service restoration of electric power distribution system with priority
R EFERENCES customers,” IEEE Transactions on Power Delivery, vol. 23, no. 1, pp.
[1] X. Huang, Y. Yang, and G. A. Taylor, “Service restoration of distribution 261–270, Jan 2008.
systems under distributed generation scenarios,” CSEE Journal of Power [9] M. R. Kleinberg, K. Miu, and H. D. Chiang, “Improving service
and Energy Systems, vol. 2, no. 3, pp. 43–50, Sept 2016. restoration of power distribution systems through load curtailment of
[2] A. Mondal, M. S. Illindala, A. S. Khalsa, D. A. Klapp, and J. H. Eto, in-service customers,” IEEE Transactions on Power Systems, vol. 26,
“Design and operation of smart loads to prevent stalling in a microgrid,” no. 3, pp. 1110–1117, Aug 2011.
IEEE Transactions on Industry Applications, vol. 52, no. 2, pp. 1184– [10] J. M. Solanki, S. Khushalani, and N. N. Schulz, “A multi-agent solu-
1192, March 2016. tion to distribution systems restoration,” IEEE Transactions on Power
[3] H. Gao, Y. Chen, Y. Xu, and C. C. Liu, “Resilience-oriented critical load Systems, vol. 22, no. 3, pp. 1026–1034, Aug 2007.
restoration using microgrids in distribution systems,” IEEE Transactions [11] K. L. Butler-Purry and N. D. R. Sarma, “Self-healing reconfiguration
on Smart Grid, vol. 7, no. 6, pp. 2837–2848, Nov 2016. for restoration of naval shipboard power systems,” IEEE Transactions
[4] C. S. Chen, C. H. Lin, T. T. Ku, M. S. Kang, C. Y. Ho, and C. F. on Power Systems, vol. 19, no. 2, pp. 754–762, May 2004.
Chen, “Rule-based expert system for service restoration in distribution [12] A. Arif, Z. Wang, J. Wang, and C. Chen, “Power distribution system
automation systems,” in 2012 IEEE International Conference on Power outage management with co-optimization of repairs, reconfiguration, and
System Technology (POWERCON), Oct 2012, pp. 1–6. dg dispatch,” IEEE Transactions on Smart Grid, vol. PP, no. 99, pp. 1–1,
[5] C.-S. Chen, C.-H. Lin, and H.-Y. Tsai, “A rule-based expert system 2017.
with colored petri net models for distribution system service restoration,” [13] J. Buckley, “Ranking alternatives using fuzzy numbers,” Fuzzy Sets and
IEEE Transactions on Power Systems, vol. 17, no. 4, pp. 1073–1080, Systems, vol. 15, no. 1, pp. 21–31, February 1985.
Nov 2002. [14] M.-S. Kuo, G.-S. Liang, and W.-C. Huang, “Extensions of the multi-
[6] B. Maleki, M. Gandomkar, T. Maleki, and F. H. Gandoman, “Method criteria analysis with pairwise comparison under a fuzzy environment,”
of evaluating reliability of microgrids in island mode by using load International Journal of Approximate Reasoning, vol. 43, no. 3, pp. 268
prioritization,” in 2014 19th Conference on Electrical Power Distribution – 285, 2006.
Networks (EPDC), May 2014, pp. 76–81.

You might also like