MODEL SIMULATION TO REDUCE TECHNICAL AND NON-

TECHNICAL LOSSES OF ELECTRICITY DISTRIBUTION

SYSTEM
E. Suryani a, R. A. Hendrawan a, U. Salama a, L. P. Dewi b
a
Information Systems Department-Faculty of Information Technology, Institut Teknologi Sepuluh Nopember (ITS), Surabaya,
Indonesia, E mail: [email protected]
b
Informatics Department-Faculty of Industrial Technology-Petra Christian University, Surabaya, Indonesia

Keywords: electrical energy, simulation, system dynamics, meter reading and illegal usage. Thus, this research will
scenario model. develop a simulation model for to reduce the technical and
non-technical losses.
Abstract Electrical losses are mainly caused by technical and non-
technical factor. One of the possibilities that cause
The reliability of electricity distribution systems plays an distribution losses is network configuration, for instance,
important role to guarantee the sustainability of electrical when its length tends to keep increasing, or when the
power supply to consumer. The main reason why we should configuration is overload [3]. Electricity distribution flow can
sustain the electrical power supply is to reduce “blackout”. be seen in Figure 1. From this image, we can see that
Based on the above problems, a comprehensive study is technical losses are incurred in MVN, LVN, TRF, and in
required to help the electricity company to reach HHC. The technical losses are caused by error in meter
interoperability of their electricity distribution system. The reading and illegal usage.
method used in this research is System Dynamics Simulation,
based on consideration that this framework enables us to MVN Losses

analyse systems and to conduct performance trade-offs of

alternative “what-if?” options. The model output can be used Transformator LVN Losses HHC Losses
Losses
as a direction in making decisions. Research results show that
technical losses can be reduced through some scenarios such Transformator
as Medium Voltage Network’s cross sectional area
replacement, which will also affect the Low Voltage Network Medium Low Voltage Household
Network Connections
and Household Connection and increasing power factor (Cos Voltage
Network (LVN) (HHC)
). Meanwhile, non- technical losses can be reduced by (MVN)

inspection and Automatic Meter Reading (AMR) usage. Figure 1. Electricity distribution flow

Considering the above problems, System dynamics is utilized

1 Introduction to model the electricity distribution behaviour in consuming
electrical energy.
Recently, the reliability of a distribution system plays an
important role to ensure sustainability of electrical supply to 2 Literature Review
consumer. The electricity distribution system used by national
electricity company (PLN) is ‘electrical centralized system’ Electricity distribution losses defined as the difference
[1]. This centralized system has some negative impact such between the electricity that goes into the distribution network
as: and that leaves it towards the customer site [2]. These losses
• Electrical energy losses can be classified into technical loses and non-technical losses.
• Unstable electrical voltage Technical losses are resulting in: losses in drivers, corona
• Electrical blackout effect, iron of the transformers, eddy currents, connectors, and
In general, electricity distribution losses can be classified as ohmic losses [4]. We classified these losses to the segment of
technical and non-technical losses. Technical losses depend the electric system where it happens, being subdivided into
on the energy flowing through the network, the nature of losses in the Medium Voltage Networks (MVN), Low
transmission lines and transformers, meanwhile non-technical Voltage Networks (LVN), substation Power Transformers,
losses, are caused by incorporate measurement errors, and Home Connections (HHC). Non-Technical Losses are
recording errors, theft, and timing differences [2]. Technical losses caused by external actions to the power system or are
losses are divided into Medium Voltage Network (MVN) caused by loads and condition that the Technical Losses
losses, Low Voltage Network (LVN) losses, Household computation failed to take into account [5]. In general, Non-
Connections (HHC) losses, and Transformator (TRF) losses. Technical Losses occur as a result of error meter reading
The Non-Technical Losses are mainly caused by error in (administrative error) and illegal usage such as theft.
Technical losses are caused by flowing in the electrical assuming the Load Density Factor (LDF) in Amp/kms is
network and generate the following types of losses: Copper equal in all feeders. MVN losses are caused by current, length
losses because of the finite resistance of conductors, long of MVN, MVN resistance, correction factor (a factor that
single phase lines, unbalanced loading, overloading and low occurred by instability, resistance factor, and temperature:
voltage, poor standard of equipment. On the other hand, non- ranging between 0.689-1.870), loss factor, L equivalent, and
technical losses are caused by actions external to the power ignition time [10]. The MVV flow diagram can be seen in
system or are caused by theft, metering inaccuracies, and Figure 2. Simulation result shows that MVN losses are
errors in technical losses computation [5]. Technical losses ranging between 306.000 kWH – 374.000 kWH. Electricity
reduction can be done through changing several devices in distribution losses in 2011 reached its peak in August, while
transmission and distribution system or periodic maintenance. in 2012; the losses reached its highest in October.
Meanwhile, to reduce illegal usage, some steps need to be current squared length of MVN

taken such as accelerating privatization policy, developing

feeder current
regional price policy, increasing people’s awareness about losses
l equivalent
total trafo 25 MVN
economic and social dimension of illegal usage, and pushing power trafo 25 output voltage

people to use other source of energy [6]. Distribution losses total trafo 50
power trafo 50
correction factor
can be reduced by choosing the right transformer and feeder, power trafo 100
MVN
resistance loss factor
total trafo 100 total
distribution network reconfiguration, and placing shunt feeder
power
capacitor in the right [7]. Each network service unit need to total trafo 160 power trafo 160
power trafo 630
be improved either internal or external. Several factors can power trafo 200

affect the performance of electricity distribution system such total trafo 200 power trafo 250
power trafo 315
total trafo 630

as providing knowledge about the electricity distribution and total trafo 250
total trafo 315
socializing company’s program about energy saving to the
local community periodically to reduce operational cost and
Figure 2. MVN sub model flow diagram
increase company income [8].
In this research, we utilized system dynamics as a method for
the model development. The development of System 3.2 Low Voltage Network (LVN) and Household
Dynamics model, there are several steps needs to be done [9]: Connection (HHC) Sub model
1. Problem articulation: in this phase, modeller has to find the The nodes in LVN are the poles where household connections
problem and determine the significant variables to give are connected. Modeling LVN is similar to MVN, only with
understanding in designing policy to cope with the problem. different level of voltage. The LVN and HHC flow diagram
2. Dynamics hypothesis: in this phase, modeller needs to can be seen in Figure 3. HHC losses are calculated the same
develop a theory of how the problem arises. Modeller also way, assuming that consumers load current is on each of the
needs to build a causal diagram explaining causal household connection’s end. Simulation result shows that
relationships among variables, and convert them into level LVN losses in Pamekasan are ranging between 400.000-
and rate (stock and flow) diagram. 480.000 kWH, while the HHC losses are ranging between
3. Formulation: in this phase, modeller has to define equation 310.000 – 375.000 kWH.
describing relationships among system variables, estimates length of LVN <l equivalent>
parameter, and determine initial condition.
4. Testing: this phase is a testing process for the model, <feeder current> losses
<correction factor>

whether the behaviour of the model represents the actual LVN

system. <total
LVN
resistance <loss factor>
5. Policy formulation and evaluation: this phase is a suggested feeder
power>
length of HHC

policy formulation and evaluation process. If the structure and correction factor

behaviour of the model represents the actual system, it can be length of HHC
<feeder current>

used to design and evaluate policy by altering the parameter <output voltage>

or redesigning new structure to improve system performance. <loss factor>

losses
HHC

3 Base Model Development <total

feeder
HHC
resistance
<l equivalent>

power>
<correction factor>
When the significant variables of electricity distribution
system are obtained, the causal loop diagram can be
developed by using those variables. The system dynamics Figure 3. LVN and HHC Sub Model
model is developed with 24 month of timeframe. The sub
models developed in this research are: 3.3 Transformer Sub model
Transformer losses are classified into Ferro (Fe) and cuprum
3.1 Medium Voltage Network (MVN) Sub model (Cu) losses. Ferro losses are determined by voltage and have
The MVN is modelled to a main feeder, which the nodes are a constant value. Cuprum losses are equal to squared loading
distribution transformator placed in that main feeder, level. Transformer losses are caused by Fe losses, Cu losses,
utility factor, correction factor, and loss factor. Utility factor
is the ratio between the electricity power generated and feeder non-technical losses are between 870.723 kWH – 1.849.640
production capacity within the operation hour of generator. kWH as seen in Figure 6.
There are about 471 transformers in Pamekasan. Simulation <susut JTM>
<Time>

result shows that transformer losses obtained in this model is <susut SR>

about 340.882 kWH. <susut JTR>

total Total Susut
susut
teknis
susut non
3.4 Supply and Demand Sub model <susut Trafo 25>
teknis
susut administrasi

This sub model is developed to analyze electricity distribution <susut Trafo 50> total <susut Trafo 630>
susut susut pemakaian
losses viewed from electricity supply from PLN and Trafo ilegal
<susut Trafo 100>
electricity consumption (demand). The losses can be <susut Trafo 315>

determined by looking at the difference between electricity <susut Trafo 160> <susut Trafo 250>
supply and its consumption. This sub model can also <susut Trafo 200>

determine the Pamekasan Utility. Figure 5. Technical, non-technical losses, and total losses
Flow diagram of supply and demand sub model can be seen submodel
in Figure 4. Simulation result shows that electricity losses
viewed from supply and demand side is ranging between 2,5
– 3,8 million kWH.
initial supply initial demand
<Time>

supply
produced Demand
rate of supply kwh kwh309
rate of demand

Pamekasan
seasonal effect Utilization
electricity
supply losses
total based on
expense supply <Pamekasan Design
and Capacity>
demand
electricity price per kwh
percentage
of losses
expense
Figure 6. Total losses, technical losses, and non-technical
losses
expense losses
Figure 4. Supply and Demand sub model
4 Model Validation
3.5 Design Capacity Submodel Model validation is a testing process whether the developed
model represents the actual system. There are two ways of
This submodel is developed to determine the maximum
testing [11]: Means Comparison, stated by E1 as equation (1),
production capacity in Pamekasan. Based on the data from
and Variance Error, stated by E2 as equation (2). It stated that
Pamekasan PLN, the feeders in Pamekasan are: APG, ARL,
a model is validated if E1 ≤ 5% and E2 ≤ 30%.
BYA, CBR, GZL, PGT, PLG, PRP, and SMD. These entire
feeders are contributing to total installed capacity (Pamekasan
E1
S A (1)
Design Capacity).
A
Design capacity value is calculated from feeder power, days
in a month, power factor, and conversion factor from KVA to Ss  Sa
E 2 (2)
kWH. Simulation result shows that capacity in Pamekasan is
Sa
about 30 million kWH.
Whereas:
3.6 Total Losses Submodel S = average value of simulation
A = average data
Total technical losses submodel is the total of MVN, LVN,
and HHC losses as seen in Figure 5. Non-technical losses can Ss = standard deviation of simulation
be calculated by subtracting total losses from the measuring Sa = standard deviation of data
result with the technical losses obtained from the simulation.
Non-technical losses are mainly caused by error in 4.1 Total Technical Losses
administration, and illegal usage. The simulation result
Comparison between data and simulation result of total
shows that the technical losses can be reduced by increasing
technical losses can be seen in Figure 7.
power factor (Cos ) and cross sectional is addition.
The total losses value obtained from simulation results are E1 = |1.433.159,239 – 1.502.278,641| /1.433.159,239 = 0,04
ranging between 2.271.210 kWH – 3.832.100 kWH, technical E2 = |185.553,085 – 157.209,02| / 185.553,085 = 0,15
losses are between 1.361.640.5kWH – 1.982.460 kWH, while
 sectional area. In this scenario, the proposed cross sectional
From the E1 and E2 calculation we can see that E1 ≤ 5% and area is 150 mm2 PVC insulated copper cable.
E2 ≤ 30%, which means the model is valid.
5.2 Effect of MVN Cross Sectional Addition to LVN
and HHC
The cross sectional addition in MVN will eventually bring
effect to low voltage network (LVN) and household
connection (HHC). As a result of the cross sectional addition
in MVN, the resistance in LVN and HHC is decreased, which
means the losses will also be decreased. The simulation result
shows that the technical losses in LVN are decreased to
300,000 kWH per month, and technical losses in HHC are
decreased to 200,000 kWH.

5.3 Power Factor Increase (Cos )

Figure 7. Comparison between data and model for total
technical losses (kWH) This scenario is developed based under consideration that the
Cos  in Pamekasan is still below 0.8. It is expected that by
4.2 Total Non-Technical Losses increasing Cos  will reduce the instability in output voltage
and eventually will reduce voltage drop and technical losses.
Figure 8 shows the comparison between data and model for Based on interview with management of PLN, the Cos 
total non-technical losses. The calculated E1 and E2 are as value should be increased to 0.9 (using 38 kVAR capacitor).
follow: The value of needed capacitor formulation can be seen in
Euation. (3). The simulation result shows that by increasing
E1 = |1.337.144,63 – 1.317.639| / 1.337.144,63 = 0,014 Cos , the technical losses can be decreased to 310,298 kWH
E2 = |173.121,942 – 223.306| / 185.553,085 = 0,28 per month.

From the E1 and E2 calculation we can see that E1 ≤ 5% kVar capasitor = (3)
and E2 ≤ 30%, which means the developed model is (existing power pp*(TAN (ARCCOS(existing cos ))-TAN(ARCCOS(proposed cos ))))
validated. 1000

5.4 Inspection and AMR Usage

This scenario is developed based on consideration that
electricity losses occurred not only because of technical
losses, but also non-technical losses. The non-technical losses
are mainly caused by error in meter reading and illegal usage.
In this scenario, we can see the effect of inspection and AMR
usage to reduce non-technical losses. Inspection is conducted
to detect electricity theft, while the AMR usage is to reduce
error in meter reading. The effect of inspection can be
calculated by using Equation (4). The loss reduction in this
scenario is the difference between non-technical losses
without AMR usage and non-technical losses using AMR and
Figure 8. Comparison between data and model for total non- inspection, as seen in Equation (5).
technical losses (kWH)
Effect of inspection =
(inspection realization/stalled inspection) *100/9.47*NTL (4)

5 Scenario Losses reduction AMR Insp =

NTL - NTL W AMR and Inspection (5)
5.1 MVN Cross Sectional Replacement
The simulation result shows that by using AMR and
This scenario is developed by replacing cross sectional area in
inspection, the non-technical losses can be reduced to 77,820
MVN, which will result in decreasing MVN resistance and
kWH per month.
decreased MVN losses to 200,000 kWH per month. Currently
in Pamekasan, the cable used in MVN has 70 mm2 of cross
5 Conclusion
In medium voltage network (MVN) sub model, we assumed
that the load density factor is equal in all feeders. The value
of design capacity is determined by total feeder power,
number of days in a month, number of hours per day, and
conversion factor from kVA to kWH. Technical losses can be
reduced by adding cross sectional area to 150 mm2 in MVN,
and increasing Cos  by adding capacitor banks. Non-
technical losses can be reduced by conducting inspection to
detect electricity theft, and AMR usage to reduce error in
meter reading. From the scenario result, we can see that by
increasing Cos , it will reduce losses and increase revenue.
By replacing cross sectional area in MVN cable, the technical
losses can be reduced, so that the average revenue can
increase. Meanwhile, by conducting inspection and AMR
usage, the non-technical losses can be reduced.

References
[1] A. Saepudin. “Strategic steps to overcome energy crisis”,
Jakarta, 2012.
[2] Sohn Associates Limited, “Electricity distribution
system losses”, London, 2009.
[3] A. Handoyo, “Analysis techniques to determine
electricity distribution losses on PLN Company,
Semarang District,” 2006.
[4] Y. Al-Mahroqi, I. Metwally, A. Al-Hinai and A. Al-
Badi, ”Reduction of Power Losses in Distribution
Systems”, World Academy of Science, Engineering and
Technology, 6, pp. 498-505, 2012.
[5] J. Navani, N. Sharma and S. Sapra, “Technical and non-
technical losses in power system and its economic
consequence in Indian economy”, International Journal
of Electronics and Computer Science Engineering, 1,
(2), pp. 757-761, 2012.
[6] N. Onat, “Transmission and distribution losses of
Turkey's power system”, Advances in Energy Planning,
Environmental Education and Renewable Energy
Sources, pp. 170-175, (2010).
[7] K. Murthy and M. Raju, “Electrical energy loss in rural
distribution feeders-A case study”, Journal of
Engineering and Applied Sciences, Vol. 4 No. 2, pp. 33-
37, 2009.
[8] M. Singgih and E. Anggraini, “Analysis of electricity
distribution efficiency by considering the operational
risk”, 2008.
[9] J. D. Sterman, “Business Dynamics: Systems Thinking
and Modeling for a Complex World”, New York: Irwin
McGraw-Hill, 2000.
[10] R. Mulyadi, “Tips to decrease electricity distribution
losses in PT PLN-Jakarta Raya and Tangerang”, 2011.
[11] Y. Barlas, “Formal aspects of model validity and
validation in system dynamics”, System Dynamics
Review, Vol. 12, No. 3, pp. 186-210, 1996.