CIRED 24st International Conference on Electricity Distribution Stockholm, 10-13 June 2013

Paper 0176

REDUCTION OF ENERGY LOSSES IN ELECTRICAL

DISTRIBUTION SYSTEMS
Hamed EMARA KASSEM Mohammed BADR Salwa ALI AHMED
EX-Chairman of Rural Electrification Faculty of Engineering, Egyptian Electricity Holding
Authority- Egypt Ain Shams university- Egypt Company (EEHC) – Egypt
[email protected] [email protected] [email protected]

electrical distribution networks, the losses in power distribution

ABSTRACT systems can be classified as technical and non-technical losses.
As shown in Figure2.
This paper contains a strategy to minimize energy losses in
an electrical distribution network. In order to develop the
strategy, a model was constructed to simulate an electrical
distribution network, and different parameters were included
that help in estimation of technical losses in medium voltage
(MV) distribution network. The main objective of this paper
is to minimize technical and non-technical losses in power
systems. There are losses in power systems that cannot be
predicted or calculated beforehand, these are the non-
technical losses. The bulk of these losses are likely caused by
electricity theft, rather than the other possibilities such as
poor maintenance of meters, and calculation and accounting
mistakes, some power systems may suffer from both. The
strategy is applied on a part of real distribution network. The
network segments consist of different network busses
delivering electrical energy from substation to supply Fig.2 Technical operation and commercial management
different customers. In this network, the load supplied by one
bus is a varying load for different times especially at A. Technical losses
maximum load conditions. They are usually present due to the following reasons:-
• I2R Losses or heat loss,
Key Words: Technical losses, non-technical losses,
• Unoptimized location of transformers,
Maneuvers, PRAO software, Multi- objective genetic • Lengthy single-phase lines,
algorithm, Electrical distribution network. • Phase imbalance,
• Loose joints,
INTRODUCTION • Low power factor loads,
Every element in a power system (a line or a transformer etc) • Overloading of transmission lines,
offers resistance to power flow and thus consumes some energy • Low quality of insulators and conductors,
while performing the duty expected from it. In order to study • Low quality earthing at consumer premises.
power losses in a distribution network, which constitute a The above losses are considered "technical", reducing them
portion of the total losses in electrical power systems, the is an imperative requirement for any electric company, because
logical first step is to understand the complete picture of power if they exceed certain "allowable" levels, they can actually
systems losses. endanger the company's financial status [1].
B. Non-Technical Losses
There are so many types of non- technical losses. Among
them are the following:
• Electricity theft
• Non-payment by customers
• Errors in technical losses computation
• Errors in accounting and record keeping that distort technical
information.
Fig.1 Connection between generation and distribution systems Non-technical (or commercial) losses are associated with
Fig.1 explains General Connection between generation and inadequate or missing revenue metering, with problems with
distribution systems, Generation (Production), Transmission billing and/or collection systems, and/or with consumer
(Transportation) and Distribution (Retailing). The efficient use pilferage. The purpose of this paper is to develop a technique or
of electrical energy will slow the destruction of the the analysis of the losses reduction problem and for the
environments natural resources and will also reduce the cost of subsequent development of action plans to solve the overall
electricity for the consumers. These paper studies losses in problem, through a systemic approach. Generally, an

CIRED2013 Session 5 0167

CIRED 24st International Conference on Electricity Distribution Stockholm, 10-13 June 2013

Paper 0176

optimization system has smaller worst voltage drops, overloads, The technique used in this research was the mutation
and energy losses. Reduction of losses improves system quality, probability in which the algorithm flips certain number of bits
security, and reliability. It can improve utilization factors of (from zero to one or vice versa) in each of the selected
existing facilities and defer feeder expansion project. As a result, individuals to be mutated in the aim of increasing the
a utility can produce power with higher quality and lower individual fitness. The probability value used was 0.2 (i.e.
electricity price, which will attract customers to the utility and 20% of the individuals will be mutated). The value was low
enhance the utility competitive ability in the deregulated comparably with the crossover ratio in order to keep the
environments. The network configuration management exploitation/exploration balance. The GA will run for several
technique is considered on top of these solutions. At first, they preset time, or the difference in the fitness value does not
considered two objectives such as optimization of voltage drop
change for several preset number of generations. Fig. 1
and overload then, loss reduction in distribution systems. So, an
Shows a flow chart describes how GA works to reach the
attempt was made to solve the network reconfiguration problem
optimum solution.
having multiple objectives such as minimization of power losses,
load balancing among supply transformers, minimization of the
worst voltage drop and overload. Pareto front [2] had proposed
a GA based approach to solve the optimization problem of
technical losses in the electrical distribution networks. The
reconfiguration algorithm directly constructs an optimal solution
for voltage drop, overload and minimum losses accounting for
the major constraints on maneuvers such as the network must
not have any closed loops, and the limitations of line capacities
or voltage drops are satisfied.
THE PROPOSED GENETIC ALGORITHM
A genetic algorithm developed by J.H. Holland (1975) [3] is
a model of machine learning, which derives its behavior from
a metaphor of the processes of evolution in nature. GAs is
executed iteratively on a set of coded chromosomes, called a
population, with three basic genetic operators: selection,
crossover and mutation. Each member of the population is
called a chromosome (or individual) and is represented by a Fig. 1 Basic structure of genetic algorithm
string all the strings and applies three operators in order to Let us consider problem having "m" objectives:
optimize the function. [4-5], GAs uses only the objective
function information and probabilistic transition rules for
genetic operations. Any two solutions u(1) and u(2) (having ‘t’ decision variables
each) can have one of two possibilities one dominates the
a) Selection other or none dominates the other. A solution u(1) is said to
This operator is used by the algorithm to choose the best dominate the other solution u(2), if the following conditions
individuals or chromosomes based on their fitness values (as are true: The solution u(1) is no worse (say the operator
the fitness of the individual increase it becomes more likely denotes worse and denotes better) than u(2) in all objectives,
to be selected) to form the initial population. [4]. or
b) Crossover
This operator is used by the algorithm to enhance the fitness The solution u(1) is strictly better than u(2) in at least one
value of the individuals. In this operator, two randomly objective, or
selected individuals (called parents) exchange some of their
genes together in order to get two new two individuals are
called offspring. And the method that was used is crossover If any of the above condition is violated, the solution u(1)
probability with a probability value equals to 0.8 which does not dominate the solution u (2). If u(1) dominates the
means that crossover operator will be applied on 80% of the solution u(2), then we can also say that u(2) is dominated by
individuals in the current population. This technique was u(1), or u(1) is non-dominated by u(2) , or simply between
chosen due its superiority among the other crossover the two solutions, u(1) is the non-dominated solution, then
techniques as it ensures the transportation of the best traits in the resultant solution v dominate any member of that set then
the current generation to next one [4-6]. the set is called local Pareto optimal set. [7].
c) Mutation DISTRIBUTION NETWORK PLANNING CRITERIA
This operator prevents the algorithm from being trapped in
local minima as it helps in the recovering of the lost traits due PRAO software [9] enables analysis of the current state of
to the crossover operator. It also helps in the exploration of electricity supply regarding load flow and voltage profile in
the search space thus adding more diversity to the population. the normal condition and the reliability or quality analysis in
CIRED2013 Session 5 0167
CIRED 24st International Conference on Electricity Distribution Stockholm, 10-13 June 2013

Paper 0176

the backup condition based on evolutionary algorithms for SOLUTION METHODOLOGIES

multi-criterion optimization. [8]. Those analyses are a basis
for variant planning of the medium or long term (up to 10 Collection and analysis done of the existing data of the
years) distribution network development, during which a year electric network of the part of distribution network and main
feeding system (substations 66/11kV) and distribution panel
by year electricity supply quality verification and the
(11 kV). Network fed from 66/11KV substation; this is
economical evaluation of every investment and the whole
mainly formed of 11 bus bars. The network consists of 24
long term investment plan is performed. Further important nodes, including the main substation bus bar. With total
data is outage frequencies of the system elements (overhead installed capacity of 15.730MVA, and a maximum load of
lines, cables, transformers,). Clearly, it is necessary to have 14287.8MVA, this means a utilization factor of about 0.806at
organized system of data acquisition (measurement) and data system peak. Table II, Present the input data of nodes and
storage (database) to successfully overlook the network sections expressed in node numbering respectively. Fig. 3
performance and development. Important part of the presents the single line diagram for the network showing
distribution network planning is criteria and assumptions incoming and outgoing feeders for sample network.
used in the planning process. Below are the examples used in
the practice in Egypt, divided into three parts, according to
the type of the analyses they are used in. Load flow and
voltage profile and power loss are analyses in the normal
conditions are conducted according to two criteria: "Load of
the lines and transformers up to 80 % of thermal current
and admissible voltage drop in 11kV network up to 6 %.
The equivalent criteria are lower: load of the overhead lines
and transformers up to 100 % of thermal current and
admissible voltage drop in 11 kV network up to 6%"

NETWORK INITIAL STATE ANALYSES

The first step of the distribution network planning is analyses
of its initial state with intention to encounter possible defects.
The example given in this paper is based on the simple
network fed from 66/11 kV substation in Cairo region, with Fig. 3 Single line diagrams for the network showing incoming and
very long feeders. Overview of the relative loads of lines and outgoing feeders for sample network
voltage drops indicates that the major power supply problem From PRAO program, all sections are shown in Fig. 4 in the
in this network is high voltage drop. The consequence is low normal operation scheme. [9].
power delivery quality in the entire region, despite existing
connection lines between feeders. The picture shows
expected number of permanent outages and expected yearly
duration of outages per consumer and placement of
(manually operated) switches in the 11 kV network. Average
power delivery quality data for feeders and the entire network
are given in the Table I.
TABLE I
EXISTING SITUATION OF SAMPLE OF DISTRIBUTION NETWORK
Basic Data of EXISTING SITUATION:
24 node &(sources: 1 -
Number of nodes
feeders: 7 - transformers: 2)
Number of segments 61 (open: 10)
Overload 6857.4 kVA Fig. 4. Relative loads and voltage drops of existing situation-
Sum of losses 723.8kW expected yearly outage duration

Max. voltage drop 8.38 % CONFIGURATION NETWORK

Minimum transit margin -3000.3kVA For optimization and reconfiguration using PRAO software is
based on GA and the construction of an initial population of
Total length 38.639 km feasible individuals this technique using the system loops and
Total powered load 14287.8kVA applying specialized genetic operators as in the short
example, the Result after 12 iteration (maneuver) by open
and close the switches as follows in Table III:
CIRED2013 Session 5 0167
CIRED 24st International Conference on Electricity Distribution Stockholm, 10-13 June 2013

Paper 0176

TABLE III was calculated using multi criterion algorithm. Multi criterion
RESULT AFTER MANEUVER SAMPLE NETWORK FROM EGYPT algorithm has been used to reduce the system power losses. This
algorithm is based on reconfiguration of the system under study,
Impact index Existing situation Maneuver
while observing the over load and voltage drop combined with
Overload kVA 6857.4 0.0 system loss to reach the optimum configuration for the system.
Sum of losses kW 723.8 198.7 Results of the paper show that the multi criterion algorithm has
Max. voltage drop % 8.38 3.04 succeeded to reduce technical loss for all sample networks of the
Total powered load kVA 14287.8 14287.8 system to reach the acceptable limit.
Non-technical losses such as unanticipated increases in system
Openings Closings losses due to equipment deterioration over time, but are usually
SHOUBR_291 EL_ZER_040 ignored in any calculations. System miscalculation on the part of
EL_ZER_054 EL_400_013 the utilities, due to accounting errors, poor record keeping, or
EL_ZER_033 271196_048 other information errors may also contribute to NTL [12].
EL_ZER_039 EL_SHO_046 • Non-technical losses are calculated by the use of statistical
196107_032 793400_025 analytical methods for detecting electricity theft by
167400_017 EL_SHO_044 analyzing utility billing information.
It's obvious from Table III that the maneuver succeed to • Smart Grid can capture the identification and reduction of
reduction overload, voltage drop and power loss this mean non-technical losses through aggregation of load and
that, the multi criterion algorithm solution to develop the collecting voltage data at points throughout the network,
network according to the criteria as shown on fig.5. fig.6 shows those modern meters that can delete non-
technical losses.

Fig.6 modern metering is essential for avoiding nontechnical

Losses
REFERENCES
[1] M. Sedighizadeh, and A. Rezazadeh, "Using Genetic
Fig. 5 Relative loads and voltage drops after maneuver Algorithm for Distributed Generation allocation to Reduce
Losses and Improve Voltage Profile", PWASET VOLUME 27
CONCLUSION FEBRUARY 2008 ISSN 1307-6884.
[2] Víctor H. Méndez Quezada, Juan Rivier Abbad, Tomás Gómez
In this paper, approach based on Genetic Algorithm in order San Román, “Assessment of Energy Distribution Losses for
to Multi-objective optimization analysis, management of Increasing Penetration of Distributed Generation” IEEE Trans.
electricity distribution network planning includes, besides on Power Systems, vol. 21, no. 2, May 2006 , pp.533-540.
classic load flow and voltage profile analyses, power delivery [3] D. E. Goldberg, J. Richardson, "Genetic algorithms with
quality assessments, operation reliability analyses and above sharing for multi-modal function optimization", in Proc. The
2nd International Conference on Genetic Algorithms, New
all economic evaluation of all investments. This is especially
York, 1987, pp. 41-49.
important in medium networks characterized with low [4] S.N.Sivanandam, S.N.Deepa, "Introduction to Genetic
consumption and long small cross sectioned conductor, Algorithms" ISBN 978-3-540-73189-4 Springer Berlin
causing two major problems: unacceptably high voltage drop Heidelberg New York, 2008.
and unacceptably low power delivery quality. Classic [5] Randy L. Haupt, Sue Ellen Haupt, "Practical Genetic
solutions for those problems, new MV cable or major Algorithms" ISBN 0-471-45565-2 Wiley interscience,
reconstructions of the existing conductor, are just not Hoboken, New Jersey, second edition.
economically justified investments in such areas. This paper [6] Mitchell Melanie, "An Introduction to Genetic Algorithms" A
deals with the electrical losses in distribution system which can Bradford Book The MIT Press Cambridge, Massachusetts
London, England, Fifth printing, 1999, ISBN 0-262-63185-7.
be divided into technical and non-technical. Study of the factors
[7] IE Davidson and A Odubiyi, Michael O. Kachienga and Brian
affecting both types of losses has been presented. The magnitude Manhire, Technical loss computation and economic dispatch
of energy dissipation or the unavoidable component of technical model for T&D systems in deregulated ESI, Power
losses depends largely on the system configuration, pattern of Engineering Journal, p 55, April, 2002.
loading of distribution conductors, magnitude and types of loads, [8] A Pareto-frontier Differential Evolution Approach for Multi-
characteristics of equipment’s etc. There is a formulation of the objective Optimization Problems, , IEEE, Vol. 2, pp. 971--978,
feeder reconfiguration to reduce technical losses and balance the Service Center, Piscataway, New Jersey May 2001.
load of distribution network under normal operating conditions. [9] ARE/MEE/EEA Arab Republic of Egypt, Ministry of
The calculation of the technical losses in the sample network Electricity and Energy, Rural Electrification Authority (REA),
Internal Annual Report of PRAO program, 2002-2003.
CIRED2013 Session 5 0167