Académique Documents
Professionnel Documents
Culture Documents
3, JULY 2006
Abstract—Evaluating the technical impacts associated with could be used to enhance distribution network planning and op-
connecting distributed generation to distribution networks is a eration [7]–[16]. Consequently, it is critical to assess the tech-
complex activity requiring a wide range of network operational nical impacts of DG in power systems in order to apply gen-
and security effects to be qualified and quantified. One means of
dealing with such complexity is through the use of indices that erators in a manner that avoids causing degradation of power
indicate the benefit or otherwise of connections at a given location quality and reliability.
and which could be used to shape the nature of the contract In this work, the technical impacts on medium-voltage level
between the utility and distributed generator. This paper presents reliability and power quality will be assessed based on a steady-
a multiobjective performance index for distribution networks state analysis and the application of distribution network impact
with distributed generation which considers a wide range of
technical issues. Distributed generation is extensively located and indices. Then, in order to calculate the multiobjective perfor-
sized within the IEEE-34 test feeder, wherein the multiobjective mance index by relating the different technical issues, relevance
performance index is computed for each configuration. The results (weighting) factors are presented.
are presented and discussed. Although in practice, distribution engineers will present
Index Terms—Distributed generation (DG), distribution net- some limitations in determining DG location, the existence of
works, multiobjective analysis. an index based on technical impacts indicates where DG could
be more beneficial for the distribution network (i.e. for the elec-
tric utility, helping distribution engineers take decisions and
I. INTRODUCTION
even shape the nature of the contract that might be established
A. Real and Reactive Power Losses where are voltages at node for the th distribution
Although reliability is the primary concern for utilities, net- network configuration considering minimum demand.
work losses are a key consideration particularly given the twin
C. Current Capacity of Conductors
drivers for efficiency given environmental and economic con-
cerns. While DG may unload lines and reduce losses, the re- As a consequence of supplying power near to loads, current
verse power flows from larger DG can give rise to excessive flows may diminish in some sections of the network, thus re-
losses. Consequently, the first and second indices ( and ) leasing more capacity, but they could also increase to levels be-
express real and reactive line power losses, respectively. Thus, yond distribution line limits. The fifth index gives impor-
a beneficial DG location would decrease total network losses, tant information about the level of currents through the network
which means near unity values of and regarding the maximum capacity of conductors. Since recon-
ductoring is out of the scope of this work, only configurations
(1) with positive values (calculated currents values greater than
current capacity) will be analyzed. Within those configurations,
close to unity values for this index mean reserve capacity for
(2) demand growth
where is the total complex power losses for the th dis- (5)
tribution network configuration, and is the total com-
plex power losses for the distribution network without DG.
where and are the currents through branch for
the th distribution network configuration, and
B. Voltage
are the current capacity of conductors, and is the number of
One advantage of careful location and sizing of DG is the lines.
enhancement of the voltage profile. Therefore, the third index
is related to the maximum voltage drop which, in this D. Three-Phase and Single-Phase-to-Ground Short Circuit
three-phase approach, will consider the maximum drop between The sixth and seventh indices ( and ) are related to
each phase of each node and the root node. This index could be the protection and selectivity issues since they evaluate the max-
also used to find prohibitive locations for DG considering pre- imum short-circuit current variation between the scenarios with
established voltage drop limits. In this way, according to (3), the and without DG. These indices give the power engineer a no-
closer to unity index is, the better the network performance tion of how the DG impacts on the protection devices that were
planned for a network without such generation units. Hence,
(3) a low impact on this concern means close-to-unity values for
and indices
where are the phases , , and ; are the voltages at the
root node (equal in magnitude for the three phases); are the (6)
voltages at node for the th distribution network configuration;
and is the number of nodes.
This maximum voltage drop approach is suitable for (7)
single-feeder analysis. In multiple-feeder systems, DG relieves
the load demand and modifies the voltage profile in the feeder(s)
where is the three-phase fault current value in node
in which it is located. Therefore, must first consider the
for the th distribution network configuration, is the
maximum improvement of the voltage drop in the feeder(s)
with DG and then normalize those values with respect to the three-phase fault current value in node for the distribution net-
largest . In this way, a DG configuration that presents work without DG, and are the largest three-
phase fault current value in the network for the th distribution
the largest improvement will receive an equal to unity
network configuration and its correspondent for the distribution
whereas the others will have proportional values (Appendix A).
In order to ensure that network voltages will not be adversely network without DG, is the single-phase fault current
affected, the case of minimum demand during maximum power value in node for the th distribution network configuration,
is the single-phase fault current value in node for the
generation is also considered, since it represents a critical oper-
distribution network without DG, and are the largest
ating case [16]. Thus, the fourth index, related to voltage regula-
tion, shows the difference between nodal voltages during max- single-phase fault current value in the network for the th distri-
imum and minimum demand. It is desirable to have this vari- bution network configuration and its correspondent for the dis-
tribution network without DG.
ation as small as possible (i.e., close to unity values for index
The indices described above are signals of the network per-
)
formance, where values that are close to unity indicate better
network performance. However, these indices are not related in
such a way that would allow a unique index to indicate the extent
(4) of the DG impact, in a global manner, on a distribution network.
1454 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 3, JULY 2006
TABLE I
RELEVANCE FACTORS
Fig. 2. Impact indices for IEEE-34 test feeder with a 300-kW generator sited Fig. 3. Impact indices for the IEEE-34 test feeder with a 1200-kW generator
at each node of the circuit. (a) Indices ILp, ILq, IVD, IVR, and IC. (b) Indices sited at each node of the circuit. (a) Indices ILp, ILq, IVD, IVR, and IC. (b)
ISC3 and ISC1. Indices ISC3 and ISC1.
Fig. 4. Multiobjective performance index for the IEEE-34 test feeder with
different power generation outputs extensively located in the circuit. Fig. 5. Multiobjective performance index and total power losses for IEEE-34
test feeder varying the power generation output at node 19.
TABLE III
IMPACT INDICES COMPARISON FOR IEEE-34 TEST FEEDER CONSIDERING
DIFFERENT POWER FACTORS (POWER GENERATION 600 kW)
and almost the same values than the three-phase analysis. Nev-
ertheless, high unbalanced loads could make noticeable differ-
ences between values using single-phase and three-phase
approaches. Fig. 8. Multiobjective performance index (IMO), active power losses index
In general, distribution networks will not present the same (ILp), and voltage drop index (IVD) for nine-bus systems for three power
generation output scenarios (0.05, 0.10, and 0.15 p.u.).
impact indices tendencies, regarding power generation output
and operating power factor. It is not necessarily surprising that APPENDIX
a higher power generation output does not bring more benefits
An analysis of a nine-bus radial distribution test system
since it will depend mainly on the load distribution and location
(Fig. 7), presented in [17], is performed in this section, as well
of the DG.
as a comparison between results obtained in that reference and
The proposed methodology does not consider the time varia-
those with the proposed methodology. The following assump-
tion of loading levels and their associated DG benefits or oth-
tions were considered: 135-kV three-phase three-wire balanced
erwise. Extension of the approach with time-varying loading
system; line section impedance is ;
and power generation patterns could evaluate the time-depen-
system zero- and positive-sequence impedances at the HV/MV
dent benefits or drawbacks of DG. Other constraints or impact
substation are and ,
indices, reflecting economic [12]; environmental [17]; or relia-
respectively; the generator’s zero–, positive-, and neg-
bility [2], [7] considerations may be included depending on the
ative-sequence impedances are ,
availability of required data as well as the analysis of reverse
, and , respec-
power flows.
tively. Minimum demand level considered is 10% of the
maximum. Loads were modeled as constant power.
VI. CONCLUSION Eight configurations were analyzed by locating a DG unit in
nodes 2 to 9 considering three power generation scenarios: 18,
Various impact indices were addressed in this work, aimed at 36, and 54 MW (0.05, 0.10, and 0.15 p.u., respectively, on a
characterizing the benefits and negative impacts of DG in dis- 400-MVA base) with a power factor 0.9 lagging, whereas the
tribution networks. Furthermore, a multiobjective performance total network demand is 311.352 MW.
index that relates impact indices by strategically assigning a rel- Figs. 8 and 9 show calculated values and impact indices
evance factor to each index was proposed. , and for the three power generation scenarios.
Though the selection of values of relevance factors will It can be noticed that according to the obtained values
depend on engineering experience, the presented values solved, (using relevance factors from Table I), node 7 presents most
in a satisfactory and coherent fashion, the DG location problem, benefits for the three power generation scenarios. This is mainly
considering different power generation outputs for the IEEE-34 due to its high influence in reducing losses and improving the
test feeder. Nevertheless, the proposed relevance factors are voltage profile in its corresponding feeder. At the same time,
flexible since electric utilities have different concerns about node 7 presents a reduced number of line sections with reverse
losses, voltages, protection schemes, etc. This flexibility makes power flow (solely for 54-MW generation) and better short-cir-
the proposed methodology even more suitable as a tool for cuit index values than nodes 8 and 9 which are farther from the
finding the most beneficial places where DGs may be located, substation (Fig. 9).
as viewed from an electric utility technical perspective. Conse- Since this network presents three feeders, the calculation of
quently, these may have an economic influence, since technical was normalized considering the best voltage profile im-
impacts may be used to shape the nature of the contract that provement found. In this way, a DG unit located at node 9 pre-
might be established between the utility and the DG owner [15]. sented the largest improvement ( equal to unity, 0.011 p.u.
1458 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 3, JULY 2006
[10] E. Van Geert, “Toward new challenges for distribution system planners,”
in Proc. Int. Conf. Exhibit. Electricity Distribution. Part 1. Contribu-
tions, vol. 6, 1997, pp. I1/1–I1/3.
[11] A. Silvestri, A. Berizzi, and S. Buonanno, “Distributed generation plan-
ning using genetic algorithms,” in Proc. IEEE PowerTech, 1999, p. 257.
[12] G. Celli and F. Pilo, “MV network planning under uncertainties on dis-
tributed generation penetration,” in Proc. IEEE Power Eng. Soc. Summer
Meeting, vol. 1, 2001, pp. 485–490.
[13] R. E. Brown, J. Pan, X. Feng, and K. Koutlev, “Siting distributed gen-
eration to defer T&D expansion,” Proc. IEEE Power Engineering Soc.
Transmission Distribution Conf. Expo., vol. 2, pp. 622–627, 2001.
[14] G. W. Ault and J. R. McDonald, “Planning for distributed generation
within distribution networks in restructured electricity markets,” IEEE
Power Eng. Rev., vol. 20, no. 2, pp. 52–54, Feb. 2000.
[15] R. C. Dugan, T. E. McDermott, and G. J. Ball, “Planning for distributed
generation,” IEEE Ind. Appl. Mag., vol. 7, no. 2, pp. 80–88, Mar./Apr.
2001.
Fig. 9. Line-to-ground short-circuit index (ISC1) for nine-bus systems for [16] C. L. Masters, “Voltage rise—the big issue when connecting embedded
three power generation output scenarios (0.05, 0.10, and 0.15 p.u.). generation to long 11 kV overhead lines,” Power Eng. J., vol. 16, no. 1,
pp. 5–12, Feb. 2002.
[17] P. Chiradeja and R. Ramakumar, “An approach to quantify the technical
over the configuration without DG) with 0.05 p.u. of power gen- benefits of distributed generation,” IEEE Trans. Energy Convers., vol.
eration, whereas for 0.10 and 0.15 p.u. (0.023 and 0.013 p.u., 19, no. 4, pp. 764–773, Dec. 2004.
[18] W. H. Kersting, “The computation of neutral and dirt currents and power
respectively) it was for the connection at node 7. It can be ob- losses,” in Proc. IEEE Power Eng. Soc. Transmission Distribution Conf.
served that when locating DG at nodes 8 and 9, values Expo., 2003.
are the smallest, thus discouraging such locations (Fig. 9) when [19] W. H. Kersting, “Radial distribution test feeders,” IEEE Trans. Power
Syst., vol. 6, no. 3, pp. 975–985, Aug. 1991.
values are calculated. [20] R. M. Ciric, A. Padilha-Feltrin, and L. F. Ochoa, “Power flow in four-
Comparing nodes 8 and 9 according to the values, it can wire distribution networks-general approach,” IEEE Trans. Power Syst.,
be concluded that for power generation of 0.05 and 0.10 p.u., the vol. 18, no. 4, pp. 1283–1290, Nov. 2003.
connection at node 8 is more beneficial, whereas for 0.15 p.u., it
is at node 9. However, only the latter power generation scenario
result agrees with the analysis performed in [17] (Table A-4,
Luis F. Ochoa (S’01) received the M.Sc. degree
Cases 1 and 2, Set #1) where it was found that the connection in 2003 from the Universidade Estadual Paulista
at node 9 is better than node 8 for the same three scenarios. (UNESP), Ilha Solteira, Brazil, where he is currently
This occurs because the proposed methodology utilizes other pursuing the Ph.D. degree.
Currently, he is a Visiting Postgraduate Researcher
impact indices, such as , and , that affect the with the School of Engineering and Electronics,
global index. Consequently, it can be concluded that these University of Edinburgh, Edinburgh, U.K. His
impact indices should be taken into account in order to consider research interests are distribution system analysis
and distributed generation.
the drawbacks of each analyzed configuration.
REFERENCES
[1] A. Invernizzi, B. Buchholz, M. Stubbe, N. Jenkins, B. Dowd, and M.
Ceraolo, “Distribution Systems and Dispersed Generation: a New Focus
for CIGRE,” Electra, no. 213, pp. 17–21, Apr. 2004. Antonio Padilha-Feltrin (M’89) received the B.Sc.
[2] N. Jenkins, R. Allan, P. Crossley, D. Kirschen, and G. Strbac, Embedded degree from the Escola Federal de Engenharia de Ita-
Generation. London, U.K.: IEE Power and Energy Series 31, Inst. jubá (EFEI), Itajubá, Brazil, and the M.Sc. and Ph.D.
Elect. Eng., 2000. degrees from the Universidade Estadual de Campinas
[3] P. P. Barker and R. W. de Mello, “Determining the impact of distributed (UNICAMP), Campinas, Brazil.
generation on power systems: Part 1—Radial Distribution Systems,” in Currently, he is an Associate Professor with
Proc. IEEE Power Eng. Soc. Summer Meeting, 2000, pp. 1645–1656. the Universidade Estadual Paulista (UNESP), Ilha
[4] N. Hadjsaid, J.-F. Canard, and F. Dumas, “Dispersed generation impact Solteira, Brazil. From 1995 to 1997, he was a
on distribution networks,” IEEE Comput. Appl. Power, vol. 12, no. 2, Visiting Faculty Member with the Electrical and
pp. 22–28, Apr. 1999. Computer Engineering (ECE) Department, Univer-
[5] J. A. P. Lopes, “Integration of dispersed generation on distribution net- sity of Wisconsin-Madison. His main interests are
works-impact studies,” in Proc. IEEE Power Eng. Soc. Winter Meeting, analysis and control of power systems.
vol. 1, 2002, pp. 323–328.
[6] M. T. Doyle, “Reviewing the impacts of distributed generation on dis-
tribution system protection,” in Proc. IEEE Power Eng. Soc. Summer
Meeting, vol. 1, 2002, pp. 103–105.
[7] R. E. Brown, “Modeling the reliability impact of distributed genera- Gareth P. Harrison (M’02) is a Lecturer in Energy
tion,” in Proc. IEEE Power Eng. Soc. Summer Meeting, vol. 1, 2002, Systems with the School of Engineering and Elec-
pp. 442–446. tronics, University of Edinburgh, Edinburgh, U.K.
[8] G. Strbac and N. Jenkins, “Calculation of cost and benefits to the dis- His research interests include network integration of
tribution network of embedded generation,” in Proc. Inst. Elect. Eng. distributed generation and analysis of the impact of
Colloq. Econom. Embedded Gen., 1998, pp. 6/1–6/13. climate change in the electricity industry.
[9] P. Espie, G. W. Ault, G. M. Burt, and J. R. McDonald, “Multiple criteria Dr. Harrison is a Chartered Engineer and member
decision making techniques applied to electricity distribution system of the Institution of Electrical Engineers, U.K.
planning,” Proc. Inst. Elect. Eng., Gen., Transm. Distrib., vol. 150, no.
5, pp. 527–535, Sep. 2003.