Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

.

1989

EVALUATION OF TECHNICAL LOSSES ON ELECTRIC DISTRIBUTION SYSTEMS

J. J. Grainger T. J. Kendrew

North Carolina State University, USA Electric Power Research Institute, USA

In Eqn. ( l ) , the quantity Fab is the loss adjustment factor for


l" segment A corresponding to the 1 kW loss in segment B.
The bracketed term in Eqn. (1) represents the per-unit
Intrinsically associated with the delivery of customer load
demand losses in segment (or component) A. The loss
there are demand losses which, over a period of time, are adjustment factor can be made to reflect the per-unit
reflected as energy losses. From the system viewpoint,
capacity required in any upstream segment (or component)
such losses cannot be distinguished or separated from in order to accommodate one kilowatt of loss in another
system load. The capacity required to meet demand losses
downstream segment. Therefore, it follows that the loss
necessitates capital expenditures while production costs
are incurred in satisfying the energy losses. Each electric adjustment factor Fac between segments A and C depicted
in Fig. 2 is given by:
utility can benefit from studies aimed at establishing an
appropriate range of values for demand and energy losses.
This paper addresses the overall problem of calculating Fac = Fab * Ftx (2)
such losses within the different segments of the distribution
system. The methodology used by Electric Power Research Generalizing the relationship set forth in Eqn. (2), a table of
Institute (1) to calculate such losses on a physically - loss adjustment factors may be generated for the cascade
existing and operating electric utility is described. The connection of components A,B,C,D,E. This is illustrated in
losses for each major element of the electric distribution Table 1 for components of a radial distribution system.
system from the high voltage side of the substation
transformer to the customer's service entrance have been Table 1. Loss Adjustment Factors
identified. Of particular interest is the finding that even Distribution Demand Losses in
though the basic structure of individual distribution systems Segment B C D E
(including substations) affects the division of system losses A. Substation Fab Fac Fad Fae
among the various components, the combination of primary B. Primarv 1 _ _ Fbd
Fbc __ Fbe
_.
feeder and distribution transformer losses accounts for C. Transfirmer 1 Fcd Fce
approximately two-thirds of all energy and demand losses D. Secondary 1 Fde
on the distribution system. The results shown are based on E. Services 0 . 1
physical measurements and very comprehensive modeling
and thus offer practical guidelines for industry efforts to
evaluate losses on the distribution system.
Loss Adiustment Factors: With Diversity
DISTRIBUTION SYSTEM ~ J E R A R C H Y
The loss adjustment factors need to be modified to account
The distribution system constitutes a natural hierarchical for the diversity in time of peak load occurrence. Peak
structure within the overall power supply system. In what loads do not occur simultaneously on all components of the
follows the distribution system is considered to consist of system and so for the time of system or segment peak, it is
five segments, viz., distribution substations, distribution appropriate to define a corresponding coincidence factor,
primary lines, distribution service transformers, distribution
secondary, and distribution services. Losses occur in the coincident maximum demand
Fco= sum of individual maximum demands (3)
different system segments due to load currents and the
losses in subsequent downstream parts of the system. If
diversity is neglected, a kilowatt (kW) of demand loss in one which is a group quantity defined as the maximum
part of the system, at the time of occurrence of system peak, simultaneous demand of the grouped loads (or
requires a capacity somewhat greater than one kilowatt in components) to the sum of the maximum demands of the
the components of the system upstream towards individual loads (or components) within the group at the
generation. However, the previous statement must be same point of supply. The coincidence factor has a useful
modified to account for the diversity of peak loads in the meaning when applied to a group of loads or components
different segments of the system since it is improbable that with similar load cycles and maximum demands. For
such peak loads will occur simultaneously. Consequently, example, Fig. 3 shows a plot of a distribution transformer
the incremental capacity needed in each system load coincidence curve typical of many such curves
component must be influenced by (i) the demand losses presently used in the industry, Nickel and Braunstein (2). It
associated with loads in other downstream segments and is clear that an average coincidence characteristic may
(ii) the diversity in peak load occurrence within these have accuracy limitations for very small numbers of loads
segments. but as the number of similar loads is increased the
contribution of each load to the group maximum demand
Loss Adiustment Factors: Without Diversity decreases, and reasonable accuracy should be obtained
(2). In loss studies it is usual to consider that the load-
Consider a system component (or segment) B connected to variable demand loss incurred by a component is
its immediately adjacent upstream component (or segment ) proportional to the square of the load level. For such
A as depicted in Fig. 1. It can be shown that the capacity studies, it is preferable to define the average coincidence
required in segment A to accommodate an incremental loss factor bofor a group of N components within a segment in
in segment B of 1 kW (viewed as an additional load) is the following manner:
given by the approximation N N
Fab = 1 + 2(per-unit demand IOSS)A (1 1 C
b o = . . {Fco,i (pic,i + (,x(Pnc,j)2) 11 (4)
in which Pnc,i is the non-coincident peak demand of The framework presented above for evaluating demand
component i, and Fc0,i = (Pc,i/Pnc,i) is the coincidence factor and energy losses on the electric distribution system could
for the component i and Pc,i is its coincident peak demand. be applied to an individual radial distribution subsystem or,
Use of Eqn. (4) ensures that, as far as losses are on a segment basis, to the overall distribution system. The
concerned, the coincidence factor of the components with data required for such application is considerable and
loads substantially larger than others will make a bigger extensive modeling of system components is involved. In
contribution to the average coincidence factor of the the next section, the results of a research project
associated group or system segment. undertaken by the Electric Power Research Institute (1) are
reported. The losses for each major element of the electric
The coincidence factors between the various segments of distribution system from the high voltage side of the
the distribution system have been used, in a straight- substation transformer down to the customer's service
forward manner, to account for the impact of load diversity entrance have been identified. The derivation of each
on the loss adjustment factors. For example, the loss element's losses is discussed along with the associated
adjustment factor Fab when multiplied by the corresponding methodology.
coincidence factor (Lco)ab yields the capacity adjustment
factor Cab between segment (or component) A and the CAlCUl ATION OF LOSSES IN SEGMENTS
downstream segment (or group of components) B i.e.
The purpose here is to describe, in summary manner, the
Cab = Fab ( k o h techniques which have been used to estimate losses within
the various segments of an actual, physically existing
It is noted that the coincidence factor is used directly in Eqn. system (1). The methodology involves use of measured
(5)since the quadratic nature of the loss variation with load quantities and other physical characteristics of the system to
is already taken into account in the determination of both establish statistical and physically-based models through
Fab and (Lo)&. A table of capacity adjustment factors can which demand/energy losses can be calculated for the
be generated for the cascade connection of segments (or different components of the overall system.
components) in a manner analogous to that used for Table
1 e.g. Substation Losses

Owing to the enormous volume of metered data available


on a daily basis from 130 substations, it was necessary (i) to
wherein it is understood that the upstream-to-downstream statistically group substation transformers with similar
sequence is A+B+C+D. loading characteristics and (ii) to identify significant day
types having statistically-similar daily load-cycles. The
W Y LOSSES objective was to use the Significant Substations and
Significant Days for detailed analysis and calculation of
Energy losses defined over a specified time-period are losses.
average in nature and impact the generation production
costs over that period. Demand and energy losses in some Sianificant Substations. This grouping was
components, such as distribution substation and service accomplished by initially examining the monthly loads and
transformers, involve no-load and load-variable quantities. then daily load cycles via the following procedures:
No-load losses are essentially constant over the life of the
unit and generally can be calculated from manufacturer's the peak load on each transformer for the average
data. However, load losses vary with the continuously temperature day of each month was identified,
changing load level throughout the daily and seasonal load
cycle which complicates the evaluation of their associated the twelve data entries for each transformer were
energy component. To assist in such calculations the normalized to facilitate comparison of all substation
concept of loss factor FlShas been defined as the ratio of loadings on a monthly-peak load-cycle basis,
the energy loss during time period T to the energy loss that
would result if the peak loss persisted throughout the period using a statistical analysis package, the substation
T i.e., transformers were segregated into different sets of
clusters and a measure of compactness of each cluster
kWh loss durin time T (hrs) was calculated for each set,
Fls = (Peak kW I k s ) * T (7)
within each substation cluster a physically existing
Energy losses cannot be separately measured but must be substation was identified as representative on the basis
indirectly calculated. From the load duration curve relative of a mean-square criterion,
to the time period T the load factor Fld can be determined
using the definition: the actual daily waveshapes for different seasons of the
year were then studied for refinement of the clustering
Avera e load over time T procedures,
Fld = ( P e g load during T) (8)
it was concluded that the 130 substations could be
Although, in general, the loss factor cannot be determined segregated into a set of 7 significant groups of various
from the load factor, under certain assumptions it is sizes.
permissible to use the empirical relationship
Sianificant Davs. The statistical identification of this
grouping was accomplished as follows:
(9)
the hourly system load for each day of an entire year
was prepared,
in which 0.15 5 a 5 0.30 is often chosen. One kilowatt of
demand loss occurring during any hour in a given the hourly load cycle for each day was normalized and
component induces a need to supply more than one statistically analyzed using the clustering routine
kilowatt over the same hour in each upstream component of indicated above,
the system. Consequently, the energy loss calculated for
each segment (or component) must be increased by using four significant day groups were determined.
the loss adjustment factors set forth in Table 1.
489
Fig. 4 graphically reflects the segregation of substations
and days on the basis of load-cycle data. A daily load cycle 22 = qcose + xpsine: lateral impedance factor (ohmslkm)
was developed for each of the four significant days using
actual data acquired from each of the substations V : line-to-linevoltage (kV)
representing each of the seven significant groups. The
normalized daily load cycles for each group were AV : percentage voltage drop
considered to represent each of the actual substations (and
their transformers) within that group. No-load and full-load D : load density (kVA per km2)
losses for each substation transformer were specified for
rated conditions on manufacturers' data-sheets. d : lateral spacing (km)
Accordingly, the losses on each substation transformer
were determined on a daily basis by multiplying the &1,k2 : constants, depending on type of lateral
normalized load-cycle curves by the square of the
transformer daily peak demands which were known for To use these generalized loss formulae there is need to
each of the 365 days of the year. The equations used are: know the feeder and lateral loads at the time of system
peak, and indeed, throughout the load cycle. With such
Demand: detailed load information available the preferred approach
load kW loss = (rated load-loss)*(per-unit peak load)2 to loss evaluation is via a system simulation program. In the
No-load kW loss = specified by data sheet study reported in (1) all of the 477 primary three-phase
feeders were represented by detailed three-phase line-
Energy: section models. These models were based on line lengths,
load kWh loss = wire sizes and construction geometry available from
8760 company records. Magnitudes and geographic distribution
[(ratedload-loss)*(per-unit peak load in hour of loads were reconstructed from output data from the
j=1 distribution transformer load management (DTLM) program
which includes information on transformer capacity, peak
No-load kWh loss = (No-load kW loss) * 8760. kVA load, customer rate-class, number of customers,
number of transformers, and transformer location. The
v Feeder -Lateral Losses natural power factors appropriate to residential (0.89) and
commercial (0.85) feeders were used to convert the DTLM
The primary system considered here includes all lines kVA loads to real and reactive power. The feeder demands
emanating from the distribution substations to the were determined with a detailed model and the DTLM loads
distribution transformers from which the customer load is with a time-of-day customer ratio resulting in the ability to
served. Accordingly, all three-phase main feeders and their match substation loads within 3%. Twenty-six physically
associated three-phase and single-phase laterals are existing feeders were chosen for further detailed study. The
included. Such lines supply a combination of concentrated chosen feeders included those served by each substation
and uniformly distributed loads throughout each section for representative of the seven significant clusters indicated
which the peak and average demand losses may be above. The assumption underlying this choice is that such
calculated, as shown by Chang (4), using the formulae feeders are also representative of the total (477)feeders in
the system. It is interesting to note that the feeders
consisted of
Peak kW loss = k r f! 112 (1 + h + h2) 10 -3 (10)
Ave kW loss = k FlSrP 122 (1 + p + p2) * 10 -3 (11) 177 km of single-phase (6 wire-sizeshypes)
where 45 km of two-phase (3 wire-sizes/types)
11: sending-end current (amps) in line section at time
of system peak, more than 160 km of three-phase (13 wire-sizes/types)
h: ratio af section receiving-endcurrent to I1 at time
of system peak, The data describing each of the twenty-six feeders (e.g.
12: sending-end current (amps) in line section at time load, length, area) was used to statistically group the
of feeder (or lateral) peak feeders into similar types representable by residential- and
p: ratio of section receiving-end current (amps) to 12 commercial-type feeders. Analysis of the five calculated
at time of feeder (or lateral) peak, demand quantities
FlS: loss factor
r: conductor resistance (ohms/km) single-phase and two-phase line losses
1: section length ( km) single-phase line losses
two-phase line losses
three-phase line losses
and total feeder losses
1 for three phase section
k={ was undertaken to investigate correlation with twenty-seven
213 for single-phase lateral (wye or delta) variables relating to area served, average load per
transformer, line length, connected kVA loads,
An alternative but more approximate method of calculating residentiakommercial kVA mix, and voltage drop, among
others. Not unexpectedly, a strong correlation was found to
primary line losses assumes each feeder to have a exist between the lengths of single-phase and two-phase
geometric service area of uniform load density. For a laterals and their respective losses. It was, therefore,
rectangular service area it has been shown by Schultz (3)
that determined that a calculated power loss of 210 Watts per
kilometer for single-phase lines and 440 Watts per
kilometer for two-phase lines could be validly used to
determine losses on laterals. It is noteworthy that the length
of the three-phase lines did not correlate very well with the
three-phase losses except for commercial feeders. On the
where: other hand, it is to be remarked that a good approximation
to three-phase feeder losses could be based on per-unit
21 = rlcose + XlSin8: main impedance factor (ohms/km)
voltage drop as was demonstrated by linear regression
490
analysis. It was further determined that energy losses on twenty locations per transformer size were sampled
the primary lines could be calculated using substation loss having loads approximating the corresponding RMS
factors provided that the latter are used with loads adjusted loads,
to match substation demand.
for each location the actual physical configuration of the
mer Lo- lines was used to calculate the secondary line losses
based on the load allocation given by
The host utility involved in the loss-evaluation project had
more than 70,000 distribution service transformers, the sect. kVA load = transf. kVA load Fco.l* “I
loadings on which are either directly monitored (for Fw,t nt (15)
commercial customers) or indirectly calculated (for
residential customers). Most commercial customers on the where Fc0,l and Fco,t are the coincidence factors
system have demand meters which provide actual peak corresponding to the number nl of customers on the line
load information and so effort was directed towards section and to the number nt of customers, on the
establishing the relationship between the peak kVA transformer, respectively.
demand on residential load transformers and kWh
consumption of the associated residential customers. The The curves for the coincidence factor Fa and loss factor F s
host utility’s practice is to monitor currents drawn by appropriate to the host utility were available and these
residential-load transformers at many locations during each permitted calculation of the annual energy losses as
peak-load season. Consequently, the desired kVNkWh follows:
relationship was statistically established, as exemplified in
Fig 5,and used with output from the distribution transformer
load management program to provide a summary of kWh losses =
transformer load data on an annual basis. Loss factors Fi,, sect. kVA)2(resistance 2 length)Fld
8760 (16)
derived from load surveys which yielded actual half-hour (kV)*lOOO
demands of the different customer classes for an entire
year, were calculated to be The calculations were then extended to encompass the
entire system.

I
0.07 : for residential
FI,= 0.09 : small commercial (c400kW)
0.1 7 : large commercial
Service Losses

The losses associated with overhead (OH) and


underground (UG) services were calculated for each
The transformer energy losses were then determined using
the standard transformer loss-equation i.e. customer class based upon the following simplifying
assumptions relating to average values:
Annual kWh losses = [ (per-unit load)2 FlS* (13)
(rated losses) + (No-load losses) ] 8760
Residential OH UG
Demand losses for each service transformer were based ---------__---- __--- -___
upon its system peak coincidence factor Fco, quantified wire size #2AL #If0 AL
according to the number of connected customers as follows: length 18.3111 16.8m
loss factor FI, = 0.052
Residential SmaII Comm. I arae C o m a
NO. F a NO. Fm NO. Fa
Sm. Commercial OH UG
1 0.307 1 0.263 1 0.423 ....................... -___- ___-
2 0.378 3 0.420 2 0.513 wire size 4lOAL 350MCM
6 0.508 5 0.434 5 0.642 length 12.2171 12.2m
9 0.533 10 0.539 10 0.714 loss factor Fls= 0.1 1
18 0.551 15 0.582 20 0.768

v Lg. Commercial
.......................
OH
-____
UG
--__
Since commercial customers are served directly from the wire size 5OOMCM 750MCM
transformer, losses on the secondary system were length 12.2111 12.2111
calculated for residential customers only. The difficulties loss factor FlS= 0.1 8
caused by the many different varieties of designs and
parameters were circumvented by the following direct The loss factors were calculated from independent load
approach: survey data. The coincidence factors used in the secondary
line calculations were also applied to the calculation of
energy and demand losses in the service drops.
the root-mean-square (RMS) load was found for each
major size of transformer using the formula
PISTRIBUTIO-FS: WMM4EY
RMS load = dx(load)2/no. of transformers (14)
The methodology described above permitted calculation of
to yield the following results (all in kVA): demand and energy losses in each segment of the
physically-existing distribution system with the following
Transformer RMS load RMS load results:
Size (overhead) (underground)
______-_____--__
--------------- ---------------
25 20.8 19.5
37 37.1 30.0
50 50.3 39.3
75 68.7 55.3
100 92.8 86.0
491
Energy Losses On the system tested distribution transformer no-load
Percent Percent demand losses exceeded load-variable demand losses
Segment of sales of calculated losses
___--_-__-_-- ___________ ____________________------- by a factor of three.
Substation:
No load
Load
::::} 0.66 17.1
The per-unit daily load profile of each day of the year
can be clustered with other similar profiles. A year can
be represented statistically by between four and six
typical days.
Primary:
The per-unit daily load profile shape of weekends was
3$ fdrs 0’70} 0.74 19.0 not unique and could be represented with statistical
l4&2$ 0.04 significance in losses analysis by one of the typical
wee k-days.
Distribution Transformer:
no-load
load
}:::A 2.14 55.1 In a manner similar to that used for the significant days,
certain substation transformers could be clustered by
considering the twelve monthly peak loadings.
Secondary: 0.1 3 3.4
In spite of considerable effort in modeling the system in
Services: extreme detail 13.22% of total distribution losses were
res. 0.085
sm. comm. 0.079
lg. comm. 0.046 i 0.21
-____------
3.88
5.4
_____--__
100
undetermined. This figure is thought to be a
combination of errors in modeling, elements not
calculated, measurement inaccuracies, typical input
data versus actual, or perhaps power theft.

Demand Losses REFERENCES


Percent of
Segment substation peak 1. Electric Power Research Institute, Palo Alto, U.S.A.,
--__--_-__-__ ________________-_--- 1983, “Improved methods for distribution loss
Substation: evaluation”, Research Report No. EPRl EL-3261.
No load 0.15
Load 0.61) 0.76 2. Nickel, D. L., and Braunstein, H. R., 1981,
Transact ions. PAS-1OQ, 798-811.
Primary feeders: 1.23
3. Schultz, N. R., 1978, IEEE Transactions, PAS-97
Distribution Transformer: 603-609.
no-load 0.86 4. Chang, N. E., 1968, JFFE T ransactions. PAS-8L
load 0.301 1.16
1991-1994.
Secondary: 0.16

Services: 0.23
Gf?p.eratio n
Log ...

-
Figure 1. Cascade connection of two components.

On the basis of the above developments the following GeFFration


observations and conclusions are presented: - A 6 - ...
c + Load
Transmission and distribution networks require
additional capacity to supply the demand losses of Figure 2. Three components or segments in
downstream components. cascade.
Capacity adjustment factors can be developed based
on the hierarchical structure which characterizes the
electric distribution system.
The capacity adjustment factors provide a means to
calculate the additional capacity requirements induced
by the losses of components within an individual radial
subsystem or by the losses within the segments of the
overall distribution system. 8
The demand loss data needed for the calculation of $5
C
0.4; $( ’+- 5
2N+3
,
capacity adjustment factors has been made available I .

for radial distribution subsystems in the results from a 0.2-


comprehensive research effort on an actual system.
Distribution transformers account for the majority of
0.0 . , . , . , . , . , . , , , , .
energy losses and the combination of primary feeder Number of Customers
and distribution transformer losses accounts for more
than half of all energy and demand losses. Figure 3. Coincidence factor curve; average for
utilities.
492
I I

Substation Groups

Figure 4. Significant substation and day groups.

r
1 I I I
-
Consumptlon kWh

Figure 5. Residential load kVA/kWh relationship.

493

You might also like