j

Paper accppted for presentation at 2003 IEEE Bologna PowerTecb Conference, June 23-26, Bologna, Italy

Impact of Distributed Generation Allocation and

I

Sizing on Reliability, Losses and Voltage Profile

I
C k e n L. T. Borges, Member, IEEE, and Djalma M. Falcao, Senior Member, IEEE
I

electric losses, voltage profile, reliability, among other, needs

Abstracf- This ajticle presents a methodology for evaluating to be appropriately evaluated. The selection of the best places
the impact of DG d i t s installation on electric losses, reliability for installation and the size of the DG units in large
and voltage profile 'of distribution network. The losses and distribution systems is a complex combinatorial optimization
voltage profile evaludtion is based on a power flow method with
problem.
the representation of generators as PV buses. The reliability
indices evaluation is based on analytic methods modified to paper presents a methodology to the
handle multiple gendrations. The methodology may be used to of DG units h d l a t i o n on electric losses, reliability and
evaluate the influenci of the local of installation and the capacity voltage profile of distribution networks. The influence of the
of DG on these system performance characteristics for different local of installation and the caoacitv of DG on these svstem
. I
~~
-2- - - ~ ~ ~
generation expansion' planning alternatives The results obtained performance may be for different
with the proposed methodology for systems extracted from the generation expansion planning alternatives using the proposed
literature demonstrates its applicability.
I methodology.

Reliability, Electrical Losses, Voltage Regulation. 11. IMPACT OF DISTRIBUTED GENERATION

I. INTRODUCTION A. Reliability
If the DG units are correctly coordinated, they can have a
positive impact on distribution system reliability. A simple
example of DG use is as generation backup, in which the unit
is operated in the case of local utility supply interruptions. In
an online system that includes DG, load transfer to other
feeders via switches operation can be performed in
interruption situation in order to keep customers supply. With
the electrical sector restructuring taking place in the world, it
is of great importance to maximize the positive impact of the
presence of DG [I].
Another DG application that is gaining popularity is the
injection of the excess power of a DG unit generation in the
distribution network that may occur when the DG capacity is
higher than the necessary to attend the local loads. The energy
that is injected in the network is measured and the consumer
bas to pay only for the difference between the energy
consumed Corn the distribution utility and the amount injected
in the network.
When the DG is operating in parallel with the system, new
considerationsare introduced. For radial analysis, the simplest
alternative is to model DG as negative active and reactive
power injections, independent of the system voltage at the
terminal bus. To model DG units as negative loads can have a
positive impact in reliability whenever the model of reliability
evaluation considers capacity constraints during system
reconfiguration after a fault.

0-7803-7967-5/03/$17.00 02003 E E E
 If the method of reliability evaluation can deal with 111. ALLOCATION AND SIZrNG OF DG
systems with multiple DG sources operating in parallel, the Great attention should he rendered to the problem of
units of DG can be modeled as constant voltage sources. A allocation and sizing of DG. The installation of DG units at
preoccupation when using DG in those circumstances is to non optimal places can result in an increase in system losses,
avoid treating all DG sources as available for dispatch by the implying in an increase in costs and, therefore, having an
utility and using them in network reconfiguration situations. effect opposite to the desired. For that reason, the use of a
This may not always be carried out, since the DG units are not methodology capable of analyzing the influence on some
necessarily property of the energy utility. This problem can be system characteristics of DG allocation and sizing can be very
solved using a radial configuration and allocating transfer useful for the system planning engineer when dealing with the
switches whenever necessq. Reported results [I] indicate increase of DG penetration that is happening nowadays.
that when DG is used in systems with high power flows, it Given a set of possible expansion alternatives, the
results in the feeder’s where DG was allocated and also in the evaluation of a DG allocation and sizing strategy should be
adjacent feeders’ reliability improvement. made through a power flow program for distribution networks
B. Losses and Voltage Profile with the representation of generators. A way of modeling the
DG units is by constant power injection sources connected to
The distribution systems are usually regulated through tap
voltage controlled buses (PV buses). The representation of PV
changing at substation transformers and by the use of voltage
buses in radial systems, if a forward-hackward sweep power
regulators and capacitors on the feeders. This form of voltage
flow method is used, implies in the creation of network
regulation assumes power flows circulating !?om the
breakpoints, where the voltages of the two buses (terminal and
substation to the loads. DG introduces meshed power flows
fictitious) should he maintained at the same specified module
that may interfere with the traditionally used regulation
through reactivate power injection at the buses [4,5].
practices [2].
The system reliability indices in the presence of DG may
Since the control of voltage regulation is usually based on
be evaluated using analytic methods for reliability indices
radial power flows, the inappropriate DG allocation can cause
calculation in radial networks [6,7], adapted to handle
low or over-voltages in the network. On the other hand, the
multiple generation sources at distribution level. The
installation of DG can have positive impacts in the distribution
protection system selectivity needs also to be guaranteed in
system by enabling reactive compensation for voltage control,
order not to harm the reliability of the system.
reducing the losses, contributing for fiequency regulation and
acting as spinning reserve in main system fault cases.
IV. PROPOSED METHOWLOGY
Two situations have been reported 121 in examples: one of
increase and another of decrease in voltage that can happen The proposed methodology aims to evaluate the impact of
given the incompatibility of DG with the voltage regulation in DG units allocation and sizing on electric losses, reliability
radial power flows. A precise way of analyzing the voltage and voltage profile of distribution networks. Conceptually,
regulation of a system with embedded DG is by making a the methodology is based on the following methods:
simulation using power flow algorithms capable to analyze The electric losses and voltage profile evaluation is based
multiple sources of DG together with the operation of voltage on a power flow method with the representation of
regulators. In this analysis, it is important to recognize that the generators (PV buses).
power injected by the DG unit can result in voltages within The reliability indices evaluation is based on analytic
the allowed limits at the DG installation site, but it could methods modified to handle multiple generations.
result in undesired values at other parts of the feeder. The proposed methodology can be used as standalone by a
DG also causes an impact in electric losses due to its specialist to evaluate different DG installation alternatives or
proximity to the load centers. DG units should be allocated in it can be used as integral part of an automatic optimization
places where they provide a higher reduction of losses. This method.
process of DG allocation is similar to capacitor allocation to
A. Distribution Power Flow with Generator Representation
minimize losses. The main difference is that the DG units
The Newton-Raphson and fast decoupled methods for
cause impact on both the active and reactive power, while the
capacitor banks only have impact in the reactive power flow. power flow solution have efficiently solved the problem for
In feeders with high losses, a small amount of DG well conditioned power systems. However, these methods
strategically allocated (IO-20% of the feeder load) could cause may fail or have problems when dealing with distribution
a significant reduction of losses [3]. Unfortunately, the electric networks, mainly because of the following characteristics [SI:
energy utility doesn’t have absolute control of the installation Radial or almost radial topology (weakly meshed);
places, since DG is usually of the consumer‘s property. In High R K relation of the cables;
spite of that, it is of great interest for the utility to have a Mix of low impedance (switches, voltage regulators,
methodology for proper allocation of DG units in order to etc) with high impedance elements;
have an indication of the efFects caused in the system by the Unbalanced load operation.
location suggested by the independent producer.
 soww mkw

Fig. 2 -Reliability benefit due to ffi

V. RESULTS

A . Test System I
Fig. 3 shows an example system extracted from [9]. The
arrows show two locals for DG installation while the boxes
show the protective devices available in the network

A B C D
5mokW 4WOkW 30MkW ZWOkW
Fig 3 -Test System 1

Table 1 shows the main system reliability indices prior to

DG installation. The maximum voltage drop is 0.018503 pu
and the losses are 209.9 KW.

TABLE1-RELIABILITY INDICES WITHOUT DG

Table 2 shows the DG units available for installation on

any of the two locals. The DG breaker switching time was
considered 0.5 hour.

TABLE2 -AVAILABLEffi UNITS

Suppose that one alternative is to allocate DG units as

show in Table 3. In that case, the system reliability indices
after the DG allocation are shown in Table 4.

Local DG Type Capacity

#I DG #3 4000 kW
section 2 isolated by the opening the isolator on section 3. #2 DG #2 3000 kW
 SAlFI I SAID1
[/year] I [hourlyearl
1.153 2.262

It 'can he observed that the duration related indices

improves with DG installation, since some loads interruption
duration reduces when the main supply is unavailable.
However, the 6equency related indices remain the same, since
the DG breaker is also trigged when a fault occurs in the
system. In terms of power flow, the maximum voltage drop is
reduced to 0.007422 pu and the losses become 74.26 kW
after DG allocation. Table 5 shows a comparison of losses and
voltage drop prior and after DG installation, where a 64.62%
reduction in system losses is obtained. The voltage profile is
also improved by DG installation, becoming almost flat along
the feeder.

Comparison Without DG With DG

Losses (KW) 209.90 74.26
Max Voltage Drop 0.018503 0.007422
I I
Suppose now that the DG penetration must he limited to 1.805 I 3.492
5000 KW and another alternative is to install DG units as
shown in Table 6. Table 9 shows the DG units available for installation on
any of the two locals. The DG breaker switching time was
TABLE~-ALTERNATIVE~
considered 0.5 hour, as in the previous test system.
Local
TABLE9 - AVMLABLE
DG UNITS
2000 kW
13.8 kV
Table 7 shows a comparison of losses and voltage drop 13.8 kV
prior and after the installation o f the required amount of DG 15000 kW 13.8 kV
(5000 KW). Since the DG penetration is smaller than the
previous case (7000KW), the system losses are less reduced Suppose that one alternative is to allocate only one DG unit
(reduction of 38.36%) and the voltage profile improvement is as shown in Table IO. In that case, the system reliability
slightly smaller. indices after the DG allocation are shown in Table 11.

TABLEIO-ALTERNATIVE I
Comparisou Without DG With DG Local I DGType I Capacity
Losses (KW) 209.90 129.37 #I I DG#2 1 7500 kW
Max Voltage Drop 0.018503 0.0074 17

SUFI SAID1
B. Test System 2
[/year] Ibour/yearl
1.805 3.219
Fig. 4 shows an example system extracted 6om [lo], where
the filled circles represent isolators and the dashed circles It can be observed that the duration related indices
represent two locals for DG installation. improve with DG installation. In terms of power flow, the
Table 8 shows the main system reliability indices prior to maximum voltage drop is reduced to 0.029779 pu and the
DG installation. The maximum voltage drop is 0.141987 pu losses become 788.9 kW after DG allocation. Table 12 shows
and the losses are 2152.0 KW. a comparison of losses and voltage drop prior and after DG
installation, where a 63.34% reduction in system losses is
obtained. The voltage profile along the feeder is also
improved by DG installation.
 Comparison I Without DG With DG
This paper presented a methodology to evaluate the impact
Losses (KW) I 2152.02 788.87
of the local and capacity of DG units on electric losses,
Max Voltage Drop 0.141987 0.029779
_ .
I for two test systems and different allocation alternatives are
Suppose now that the SAIDI index obtained with
presented demonstratingthe applicability of the method.
Alternative 1 (3.219’hour/year)is considered elevated and DG
As continuation of the present work, it is being carried out
allocation should bd done in a way to reduce even more the
the integration between the method for evaluating the impact
SAID1 index. A DG unit with higher capacity can attend a
of DG on reliability, losses and voltage profile, described in
higher number of donsumers in the case of a main supply
this paper, and a Genetic Algorithm. The resulting method
interruption and theiefore, the overall system SAIDI index can
will be able to provide the optimal DG allocation and sizing
be reduced. For thatipurpose, consider as Alternative 2 the DG
for minimal losses and adequate voltage and reliability levels.
she*
installation as in Table 13.
Preliminary results indicate the suitability of the proposed
I
TABLE13 ALTERNATIVE^ methodolo~gyand will be reported in a following publication.
Local I I DGType I Capacity
#1 I I DG#3 I 15000kW VI[. REFERENCES
[I] R. E. Brown and L. A. Freeman, “Analyzing the Reliability Impact of
i
Table 14 shows the reliability indices calculated after the
I
Distributed Generation”, Proceedings of the IEEE Summer Meeting, pp.
l013-lO18, July 2001.
DG allocation of qlternative 2. It can be observed that the 121 p. p. ~ k- ~~ ~ ~ he , ~ ~of Dishbuted
~ i ~~~~~~i~~
~ i on ~ ~
SAID1 index was reduced i?om 3.219 hodyear to 1.563 Power System: Part 1 - Radial Distribution Systems”, Proceedings of
how/yea, what may be considered satisfactoly if the index E E E PES S W e r M = t i n g , vol. 3, PP. 1645 -1656 S e d e , 2000.
[3] F. L. Alvarado, “Locational Aspects of Distributed Generation”,
should be less than 2.0 hodyear, for example. Fmceedings of IEEE PES Winter Meeting, Volume 1, pp. 140, Ohio,
I mi
141 G.X. Luo and A. S e m l y e ~“Efficient Load Flow for Large Weakly
Meshed Networks”, IEEE Trans on Power Systems, Vol. 5, No. 4,
November 1990.
[5] D. Shirmohammadi, H.W. Hong, A. Semlyen and G.X. Lno, “A
Compensation-Based Power Flow Method for Weakly Meshed
! Disuibution and Transmission NehuorW, IEEE Trans an Power
Table 15 show; a comparison of the minimum voltage Systems, Vo1.3, N0.2, May 1988.
I
calculated along the, feeder prior and after the installation of
[6] R. Billinton and R Allan, “Reliability Evaluation of Engineering
System: Concepts and Techniques”, Plenum Press - New York Second
the DG of 15000 KW, here expressed in KV. EditioR 1992.
171 R. Billinton and R. Allan, “Reliability Evaluation of Power System”,
I
TABLE15 -MINIMUMVOLTAGE
COMPARISON
FOR ALTERNATIVE
2
Plenum Press-New York, Second Edition, 1996.
[8] M.S. Srinivas, “Distribution loads flows: A briefreview”, IEEE, 2000.
Comparison I WithoutDG I With DG [9l R. Allan, R. Billinton, “Probabilistic Assessment of Power Systems”,
Minimum Voltaie (kV) I 11.841 I 13.392 Proceedings o f r k IEEE, Vo1.88, N0.2, February 2000.
I [IO] Y.He, G. Anderson and R. N. Allan, “Determining Optimum Location
and Number of Automatic Switching Devices in Distribution System”,
Suppose now thit the minimum voltage along the feeder Proceedings of IEEE Power Tech’99 Conference, Budapest, Hungary,
should be 13.5 KV Since the minimum voltage shown in 1999.
I’
Table 15 happens plose to local # 2, one alternative is to
install another DG unit in local #2. Table 16 shows
Alternative 3 for 2 DG units installahon in order to improve
the voltage profile. J
I

1- TABLE ALTERNATIVE^

I
I :
Table 17 shows a companson of the minimum voltage
calculated along the feeder pnor and after the DG installation
of Alternative 3. It ‘can be observed that after DG installation
I
the minimum voltage obtained is higher that the desired
minimum limit.

TABLE17 - MNIMlh VOLTAOE COMPARISON FOR ALTERNATIVE 3

Comparison I WithoutDG I WithDG
Minimum Voltaie (kV) [ 11.841 [ 13.545