A New Method to Calculate Power Distribution Losses in an Environment of High

Unregistered Loads

Mentor Poveda, Member IEEE

Power Consultant
Abstract: The paper presents a new method to break down, by The competitive environment of today’s electric power
subsystem and cause, power distribution losses in electrical industry requires a better knowledge of costs; the method pre-
systems where the share of nontechnical losses is important. In sented herein enables a better estimate of losses. Since losses
this type of system the traditional method has given rise to involve an operating cost for transmission and distribution
inaccuracies due to high unregistered load. The new method is systems [ 11, it is necessary to calculate them more realistical-
based on the nature of the losses: resistive losses obtained from ly and accurately.
the squared ratio between hourly demand and maximum
demand, in each subsystem; transformer core losses not relat- II.BASES OF ESTIMATES
ed to the demand variations at the substation and at the primary
distribution feeders; and nontechnical losses with a demand If energy measured in the substation at the beginning of
profile similar to that of the load since they involve unregis- the primary feeder is referred to as “supplied energy” a)
tered loads. The method uses load profiles registered in today’s and, likewise, the sum of billed energy, energy for street light-
generalized electronic meters. The paper demonstrates the dif- ing, and consumption of the power utility’s own facilities is
ferences in calculated losses obtained by the new method referred to as “registered energy” (ER), then “total losses”
compared to the use of the traditional loss factor. (En) is the difference between “supplied energy” and “regis-
tered energy”. These losses include the set of technical loss-
Keywords: losses, power distribution. es and nontechnical losses, whose dimension will be the ini-
tial information for differentiating losses in the feeder.
I. INTRODUCTION

In various electric power utilities of Latin America, a high ETL= E, - E R

percentage of the energy entering the system is aimed at cov-
ering losses, since it is not rare to have losses accounting for The digital registers that are available at present provide
between 20 and 25% of supplied energy. Naturally, this per- the demand profile in addition to the supplied energy at the
centage cannot correspond to technical losses, that is, losses metering point. The demand profile appears as a matrix with
stemming from the operation of network components. Rather the following column-vectors: date (d), time (t), demand for
most losses come from theft, irregularities in the metering active power (DpRWI), and demand for reactive power
systems for billing purposes, and errors in billing information (DQ[kvar]).The demand interval can normally be selected at
processing. In other words, these losses involve an unregis- the operator’s will, but the most commonly used interval is 15
tered load and are called nontechnical losses or black losses. minutes. The register matrix is shown in (2). The register
The determining presence of the latter component of losses period can be one complete year, with column vectors of
makes traditional loss calculation methods obtain estimates 35,040 elements in all. Nevertheless, it is considered that a
that are far removed from the reality of the situation. Because minimum register of 672 data be used in one complete sample
of this, the present paper is proposing a new methodology to week of each season of the year.
break down losses into their components in distribution sys-
tems which is where the majority of these losses can be found.
Additionally, demonstration of the differences are provided.
12 I 1 59B 0 0 :15 2 , 3 6 0 1,325 -
Register = , I
I
.... ....
1212;;98 2 3 :45 2 , 4 5 0 1,345

= [d t DP

The sum of the elements of vector D, multiplied by the

time of the demand interval used is the energy of the entire

0-7803-5515-6/99/$10.00 0 1999 IEEE 609

Authorized licensed use limited to: Mapua University. Downloaded on February 14,2022 at 08:59:25 UTC from IEEE Xplore. Restrictions apply.
period considered in the register, as in (3). This will occur From this it can be deduced that the loss at demand i can be
only if dealing with demands, that is, integrated average calculated using equation (6).
power in each interval, since these come from the integration
’pi 2 COsQmax
of instantaneous power capacities, that is, from the same ener- D ~ =~
(-) - ~ - DRL-max (6)
gy PI. Dpmax ‘OSQi
n
kWh = Dpi x (demand - interval) (3) Equation (6) is applied for each one of the load demands
i=l and the demand profile of the resistive losses is obtained. The
The methodology being proposed is supported by the procedure turns out to be correct because demand is a step
nature of each loss component and in its ratio with respect to function in the interval; therefore, the ratio takes place during
demand in the corresponding register interval. As the losses the complete interval and yields as a result the loss demand,
in the maximum demand interval are calculated with models which is another step function.
for load flows or through the modeling of the primary distri- When you have demands of resistive losses, what is
bution feeder; on the basis of this value, the magnitude of each expressed in (3) is obtained, that is, the sum of demands mul-
component is calculated for the rest of the demand intervals of tiplied by the demand interval time is the energy of the resis-
the entire period under consideration. That is, the demand tive loss in the register period.
profile of each loss component is calculated and then on the
basis of that estimate, by means of integration, the energy of Transformercore losses. Since core losses (Dn) depend on
each component is calculated. the magnetic flux density in the core, which in turn depends
The technical losses in the distribution system of the elec- on the voltage feeding the transformer, it can be considered
tric power utility include resistive losses of the primary feed- independent from demand variation, that is, it remains con-
ers, the distribution transformer losses (resistive losses in the stant during the period of analysis.
windings and core losses), resistive losses in secondary net-
works, resistive losses in service drops, and the losses in kwh (7)
meters.
Once the power of the loss in the core of the transformers
Resistive losses. The resistive losses are calculated on the is calculated, one has the demand profile corresponding to that
basis of the expression IZR. In each subsystem, the resistive loss, which will be completely flat during the entire register
losses are calculated for the peak load demand by the usual period.
methods and on this basis resistive losses are obtained in a
given interval through the squared ratio between demand in Secondant network losses. By sampling the secondary net-
this interval and peak demand, as indicated below. The pro- works of the feeder, it is possible to calculate resistive losses
cedure that is described can be applied to all resistive losses, in the group of secondaries (DS3 belonging to this primary.
that is, in transmission and distribution system lines, for resis- Calculation of losses for the secondaries of the sampling is
tive losses of substation and distribution transformers, for the made on the basis of the demand profile registered in the low-
secondary lines, and for service drop conductors. voltage terminals of the distribution transformers which feed
In the case of resistive losses of the distribution transform them for at least one representative week of each season of the
en, the same demand profile registered at the beginning of the year. Afterwards, a typical curve of each season is obtained
primary feeder is considered to be a good estimate. and the general procedure for resistive losses described above
is applied to it, taking as a basis the losses at peak demand
The resistive loss demand (DRL-i)
at any load demand (DPi)is obtained from the modeling of the secondary networks of the
calculated by (4). sample.
DRL. i = li 2 R = ( DPi )2R
(4) The service drops are grouped in the
J?v cos$ most usual sizes and the most frequently used lengths are con-
The ratio between the peak loss and the loss at any load sidered to define the power of the resistive losses at peak
demand will be calculated by (5). demand of the corresponding client. On the basis of the power
of losses at peak demand, the general procedure to determine
D losses in the service drops is used with the demand profile reg-
( ax I2R istered in the terminals of the nearest distribution transformer
D ~ ~ - m a-x ‘ma, 2R - &Vcos$,,
- (5) to the corresponding load but considering individual peak
li2 R DPi I2R demand instead of the maximum diversified demand.
DRL-i ( -
J ~ cos+i
v
- M Energy meters have a loss ( D d that is essen-
tially due to the excitation of the voltage coils, that is, it can
be considered that it does not depend on the load. It can be

610

Authorized licensed use limited to: Mapua University. Downloaded on February 14,2022 at 08:59:25 UTC from IEEE Xplore. Restrictions apply.
calculated from the brands of meters actually existing in the from the sum of the demands of each column vector multi-
system. plied by the demand interval time, that is, applying (2) to each
component. E, is obtained from (9).
Nontechnical losses. As mentioned above, nontechnical loss-
es correspond to the nonregistered load, that is, the load that,
for one reason or another, is not billed. From the viewpoint of
E , = E, - E , - ErL
(9)
loss calculation, the demand profile of this loss component
should have a shape similar to the load demand profiles, that In addition, the load factor for the load demand profile
is, it can be obtained through a constant ratio with the load represented by the column vector DLis obtained from the cor-
demand. The load demand profile is obtained by subtracting responding energy and its peak demand. The load factor of
all the components of the technical losses from the demand this curve is the same for the “registered load” as for the “non-
profile registered at the beginning of the primary feeder. registered load”. That is, if the energy of nontechnical losses
and its load factor are known, then peak demand is defined.
III. ESTIMATE METHODOLOGY With the description of the last step, the calculation of all
components of losses, both in demand and energy, is com-
With each one of the procedures described in the previous plete.
section, the demand profiles of each component of the techni-
cal losses is obtained, that is, the value of demand by compo- IV APPLICATION OF THE METHODOLOGY
nent for each interval of the period being studied.
In each interval of the curve registered for supply in the In Latin America and the Caribbean, losses in electric
feeder, the demand of resistive losses in the feeder line and in power systems are very high. In 1991, the World Bank and
the distribution transformer winding, the demand of losses in the Latin American Energy Organization (OLADE) published
the core of the distribution transformers, and the losses in the a detailed study of the situation of the electric power sector in
energy meters are subtracted from the value of demand at the the region[3] and determined that the regional average power
beginning of the primary feeder. Likewise, the loss demands losses in 1988 amounted to 17%, with some countries record-
for the secondary networks and for the service drops are sub- ing losses of over 30%. Although this situation has now
tracted taking into account that these values take as a base the improved in some countries of the region, in others the
demand profile registered at the terminals of the secondary panorama described in the above-mentioned study has
voltage of the distribution transformers and typified for the remained unchanged.
sample group of primary feeder secondaries. On the basis of the above, samples of some primary feed-
Since all technical losses have been subtracted from the ers from various countries were taken in order to test the
demand register at the start of the feeder, the result obtained is methodology presented in this paper. The primaries that were
the demand profile for the real load. Nevertheless, this load selected were not typical cases; they were actually chosen to
includes “registered load” and “nonregistered load”. The for: apply the method to various shapes of load curves and from
mer corresponds to the definition made in the previous section different consumption sectors in order to observe the differ-
for “registered energy” whereas the latter is equivalent to non- ences that the method that was developed produces compared
technical losses. to the equation normally used to estimate the loss factor.
Equation (10) reproduces the usual expression as presented by
In the matrix obtained from the register at the start of the [2] on page 38, to determine the loss factor (FU) in function
primary feeder and defined in (2), you add the column vectors of the load factor &,).
corresponding to each loss component and the column of
demand of the real load (Q) as presented in (8), which can be
handled on a spreadsheet to conduct all the calculations
implicit in the equations needed to obtain D
,.
Where the constant A adopts values between 0.2 and 0.3 for
[d DP DRL DCL DSL DsdLDmL D L ] (8) some load cycles.
On the other hand, the samples included correspond to
In order to close the estimate, it is necessary to recall that countries where there is virtually no seasonality in the load
the differentiation between nontechnical losses and “regis- because they are located at a geographical latitude where the
tered load” is made considering that both components have climate is the same throughout the year; because of this, it is
the same nature and therefore have the same demand profile, enough to analyze one characteristic week to extrapolate the
that is, the same load factor. conclusions for yearly results.
The energy corresponding to nontechnical losses %) is In order to give an idea of the demand profile in each sam-
perfectly defined after calculating the energy corresponding to ple and at the same time present the demand profiles of the
“technical losses” (EtL),which is obtained, as can be recalled, loss components, three different ways were chosen to show
these profiles.

61 1

Authorized licensed use limited to: Mapua University. Downloaded on February 14,2022 at 08:59:25 UTC from IEEE Xplore. Restrictions apply.
a. 22.9 kV primaryjeeder The results of the analysis of the loss components, as a per-
centage of energy supplied to the feeder, is summarized in
This refers to a feeder providing service to a light-industry Table 2.
zone. It has a peak demand of 8.79 MW and a supplied ener-
gy of 715.2 MWh. The load factor of the demand profile of
supply is 48.4%. Demand profiles for the principal compo-
nents of losses are shown in Fig. 1. I
46%
45%
The results of the analysis of the loss components, as a per- 44%
43%
centage of energy supplied to the feeder, is summarized in Z 42%
Table 1. 41%
4 40%
39%
38%
37%
g 5 f f i f G g L f f G
Losses (%) Non technical losses
Component Demand Enerm
Total 6.1 5.5
Resistive: Feeder lines and Fig. 2. Loss factor as a function of nontechnical losses
distr. transf. windings 0.9 0.5
Distr. transf. cores and 0.3 0.5 Table 2
meters LOSS COMPONENTS FOR 13.2 kV FEEDER - NO.1
Secondariesand service 1.8 1.4 Losses %
Com onent
I Nontechnical I 3.1 I 3.0 I Total
Resistive: Feeder lines and
distr. transf. windin s
By keeping technical losses unchanged, the loss factor cor- Distr. transf. cores and meters 1.9 3.3
responding to a share of nontechnical losses is calculated to be Secondaries and service 2.4 1.7
between 0% and 10% of supplied energy. The loss factor flu^
tuates as presented in Fig. 2. I Nontechnical I 1.0 I 1.o I
b. 13.2 kVprimary feeder No. I C. 13.2 kVprimary feeder No. 2

This is a feeder that serves a residential sector. It has a The 13.2 kV feeder selected as an example is serving a
peak demand of 9.12 MW and a supplied energy of 872.1 commercial and residential load. It has a peak demand of 10.7
MWh. The load factor of demand profile of supply is 56.3%. MW and a supplied energy of 1,238.8 MWh. The load factor
The composition of the main losses produces the demand pro- of the demand profile of supply is 70.1%. The demand profile
file of Fig. 3.

200.0

E 150.0
:
lw.o

50,O

0.0

time (h)
-_ -CMe&neterd -Nontechnical - - - - Reslst:Fe&er(Ltransf
Fig. 1. Demand profiles of the loss components

612

Authorized licensed use limited to: Mapua University. Downloaded on February 14,2022 at 08:59:25 UTC from IEEE Xplore. Restrictions apply.
 600 T

I mCore&meters U Nontechnical E! Resisl:Feeder&transf I

Fig. 3. Compositionof the main losses for the 13.2 kV feeder No. 1

registered at the beginning of the feeder with the participation d. Loss factor and constant A variation
of the loss components appears in Fig. 5.
The results of the analysis of the loss components, as a In the above samples, the loss factor and A constant of
percentage of energy supplied to the feeder, is summarized in equation (10) adopt the values shown in Tables 4 and 5,
Table 3. respectively.
Table 4
Table 3 LOSS FACTOR RELATED TO NONTECHNICAL LOSSES
LOSS COMPONENTSFOR 13.2 kV FEEDER - NO. 2
Losses 'YO
Com onent Demand
Total
Resistive: Feeder lines and 62.9%
distr. transf. windin s
Distr. transf. cores and meters Table 5
Secondaries and service A CONSTANT RELATED TO NONTECHNICAL LOSSES

1 Nontechnical I 5.2 I 5.0 I

0.5 15 0.528

12000.0 T

1oooo.o

8000.0

6000.0

:
d 4000.0

2000.0 t

Fig. 4. Loss componentsin the demand profile of the 13.2 kV feeder No. 2

613

Authorized licensed use limited to: Mapua University. Downloaded on February 14,2022 at 08:59:25 UTC from IEEE Xplore. Restrictions apply.
 V. CONCLUSIONS VI. REFERENCES

The method described in the present paper shows the inss [I] H. Lee willis, Power Distribution Planning:
curity that the loss factor estimate introduces in the determi- Reference Book, Marcel Dekker, U.S.A., 1997.
[2] Westinghouse, Distribution Systems: Electric Utility
nation of the energy corresponding to power loss calculated at
Engineering Reference Book, Westinghouse Electric
peak demand owing to the particular characteristics of total Corporation, Pittsburgh, Pa., U.S.A., 1965.
loss components. [3] World Bank and Latin American Energy
Resistive losses, as a squared function of demand, empha- Organization, Evolution, Situation, and Prospects of
size load Deaks. Constant losses in the core tend to flatten out the Electric Power Sector in the Countries of Latin
the loss curve. And the nontechnical losses, with a demand America and the Caribbean, World Bank and
profile that is similar to the load curve, introduce another addi- OLADE, 1991 (ISBN 9978-70-012-9).
tional variation. It is clear that, depending on which compo-
Mentor E. Poveda (M’1981) was
nent is the most important, loss the demand profile will have born in Quito, Ecuador, on June 26,
a different shape and, because of this, the loss factor should be 1945. He graduated as an electrical
different. engineer from the Ecuadorian
The estimate that is usually made by adopting a value for National Polytechnic School on 1972
the A constant of equation (10) produces erratic results con- and obtained his Master of Science
firmed in Tables 4 and 5, and should be replaced by the esti- from Purdue University, USA on
mate of energy of losses based on demand profiles registered 1976. Part-time EE Professor at
Ecuadorian National Polytechnic
for the electric power system.
School and International Power
Finally, knowledge of the demand profiles of each loss Consultant with the Latin American
component makes it possible to better understand individual Energy Organization (OLADE) and
participation in both demand and energy and grant due impor- World Bank. His special fields of interest are power distribu-
tance to each circumstance that is analyzed. tion systems and rotating machines simulation.

614

Authorized licensed use limited to: Mapua University. Downloaded on February 14,2022 at 08:59:25 UTC from IEEE Xplore. Restrictions apply.