Beruflich Dokumente
Kultur Dokumente
org
Published in IET Generation, Transmission & Distribution
Received on 19th March 2010
Revised on 17th October 2010
doi: 10.1049/iet-gtd.2010.0199
ISSN 1751-8687
Abstract: Improper placement of distributed generation (DG) units in power systems would not only lead to an increased power
loss, but could also jeopardise the system operation. To avert these scenarios and tackle this optimisation problem, this study
proposes an effective method to guide electric utility distribution companies (DISCOs) in determining the optimal size and
best locations of DG sources on their power systems. The approach, taking into account the system constraints, maximises the
system loading margin as well as the profit of the DISCO over the planning period. These objective functions are fuzzified
into a single multi-objective function, and subsequently solved using genetic algorithm (GA). In the GA, a fuzzy controller is
used to dynamically adjust the crossover and mutation rates to maintain the proper population diversity (PD) during GA’s
operation. This effectively overcomes the premature convergence problem of the simple genetic algorithm (SGA). The results
obtained on IEEE 6-bus and 30-bus test systems with the proposed method are evaluated with the simulation results of the
classical grid search algorithm, which confirm its robustness and accuracy. This study also demonstrates DG’s economic
viability relative to upgrading substation and feeder facilities, when the incremental cost of serving additional load is considered.
276 IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276 –287
& The Institution of Engineering and Technology 2011 doi: 10.1049/iet-gtd.2010.0199
www.ietdl.org
1 Introduction size is better determined relative to the system power
demand rather than system generation, as adopted in the
The term distributed generation (DG), otherwise known as approach. Also, a DG whose size is equal to the capacity of
embedded generation (EG) or decentralised generation (DG) the total network load is proposed by Gozel et al. [6]. Our
in some quarters, is generally defined as any source of concern is that this size seems to be too large as it could
electric energy of limited capacity that is directly connected result in voltage and line loading MVA limits violations on
to the existing network on the customer site of the meter. the system. Acharya et al. [7] used the exact loss formula
The existing network could be a distribution network or a that involves the calculation of the bus admittance from its
sub-transmission network. At present, there is no consensus impedance to determine the optimal size and location of a
on what exactly the rating of DG as well as the voltage single DG unit. However, the large size and complexity of
level at the point of common coupling (PCC) to the grid distribution networks would render this approach to fall
should be. The definitions of these two parameters short of robustness requirements.
somewhat vary from country to country and from one In the same vein, Elnashar et al. [8] categorised the impacts
research group to another. For example, while Portugal pegs of DG on the distribution systems as positive, such as voltage
the capacity limit of DG to 10 MW that can be connected at profile improvement and negative, such as increased system
any voltage level, Austria defines DG as an electric energy loss and short-circuit level, and used these to formulate a
source of up to 10 MW to be connected at the MV level. In multi-objective function. Nevertheless, this problem
France, DG is defined in capacity as greater than 40 MW formulation is flawed, because DG would lead to increase
and connected at 225 kV voltage level. Spain allows the in loss only when it is sited in improper locations, and so
DG capacity to go up to 50% of the feeder capacity [1]. So its impact cannot be considered negative in all cases.
far, Australia has been able to practically install the largest Hedayati [9] used continuation power flow to determine the
size of 130 MW DG at 132 kV on its power grid, while most sensitive buses to voltage collapse and proposed DG
Denmark allows the utility to decide the voltage level of installation on three of the most sensitive buses. In [10] the
connection for DG units ranging from less than 2 MW to Tabu search approach was employed, and in [11 – 14] the
greater than 50 MW. On the other hand, the working group simple genetic algorithm (SGA) approach was adopted,
of CIGRE (International Council on Large Electric while Haghifam et al. [15] used the concept of Pareto
Systems) defines DG as a source of electric energy of optimality based on the non-dominant sorting genetic
maximum capacity of less than 50– 100 MW [2]. Although algorithm (NSGA-II) for the optimisation problem. Two
the capacity limit is not clearly stated for Germany, UK and analytical approaches for meshed and radial networks were
The Netherlands, the voltage levels at PCC are defined as proposed by Wang and Nehrir [16] to locate only one DG
up to 132 kV for UK, up to 110 kV for Germany and up to unit, even though DG size was not optimised. Jabr and Pal
150 kV for The Netherlands [1]. [17] proposed an ordinal optimisation approach to strike a
Based on these varying capacity limits, Ackermann et al. balance between loss minimisation and DG capacity
[3] suggested various classifications of DG based on their maximisation, while Celli et al. [18] combined the GA with
capacities as micro DG of 1 W to less than 5 kW, small DG the 1-constrained technique for the multi-objective problem.
of 5 kW to less than 5 MW, medium DG of 5 MW to less The selection of the optimal size and best locations for
than 50 MW, and large DG of 50 MW to less than installation of DG sources in large distribution systems is a
300 MW. This classification is a welcome idea since it complex combinatorial optimisation problem. In tackling
encompasses all various definitions given by different this problem, it is crucial to consider the market value of
groups/countries. the services provided by the distribution companies
Lots of factors are responsible for the recent renewed global (DISCO) as well as the possibility of deployment of DG on
interest in DG. Prominent among these include deregulation of the existing network in the face of many constraints. This
energy market, the environmental concerns of reducing the paper proposes an effective method for determining optimal
greenhouse gas emissions produced by the conventional allocation of DG in power systems. The objective functions
power plants and constraints on installing new or upgrading considered in this paper are maximisation of the system
existing transmission lines for long-distance power loading margin, as well as profit for the DISCO. These
transmission. Interestingly, recent research has shown that objective functions are first fuzzified to evaluate their
installation of DG units in the utility’s distribution system of imprecise nature, and then transformed into a single multi-
a power network would lead to the attainment of numerous objective function, before it is finally solved using the
potential benefits, categorised as economic, technical and genetic algorithm (GA). The main purpose of the fuzzy
environmental [4]. To maximise these benefits, it is crucial to expert systems introduced into the GA in this study is to
find the optimal number or size of DG units and their enable a better convergence than that obtainable in a simple
appropriate locations in power distribution systems. GA [11 –13]. The proposed method is tested on IEEE 6-bus
Otherwise, higher system power loss would not only be and 30-bus test systems. The numerical results obtained are
recorded when DG units are sited in improper locations, but evaluated with the simulation results of the classical grid
the overall system operation could also be put in jeopardy. search algorithm [19] using MATPOWER, which confirm
A number of approaches have been proposed by different its robustness and accuracy. The method proposed in this
researchers in determining the optimal size and location of paper has a benefit over most of the earlier works cited in
DG in power networks. For instance, a ‘2/3 rule’ method this literature, in the sense that it takes the annual load
often used in siting shunt capacitors in distribution systems, growth rate into account; this means that the algorithm
was adopted by Willis [5] to place a DG unit in a radial could be used in future years as the load growth increases.
distribution feeder. He proposed that 2/3 capacity of the The rest of this paper is structured as follows: Section 2
incoming generation be placed at approximately 2/3 of the presents formulation of the fuzzy multi-objective
length of the feeder. However, apart from the fact that this optimisation problem, while Section 3 gives a brief
method cannot be applied to a meshed network, it also overview of the GA as an optimisation and search
cannot guarantee optimal size and best location, since DG technique. The implementation algorithm for optimal
IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276– 287 277
doi: 10.1049/iet-gtd.2010.0199 & The Institution of Engineering and Technology 2011
www.ietdl.org
placement of DG units is carried out in Section 4. To evaluate Breaking the first term of the RHS of (3) yields
the proposed method, two IEEE test systems are used to
obtain numerical results and are presented in Section 5,
NG
NL
NG
NG
NL
while Section 6 discusses the results obtained in the PGj g + j
PDG k
= PG0 g + DPGj g + j
PDG k
(4)
preceding section. The conclusion to this paper is drawn in g=1 k=1 g=1 g=1 k=1
Section 7.
The total generation increase in the system, that is, the last two
terms of the RHS of (4), could be expressed in terms of
2 Problem formulation additional load demand and system total loss increase, as
presented in (5)
2.1 Objective functions
NG
NL
NB B −1
N
To determine the optimal size and location of DG in
DPGj g + j
PDG = DPDj i + j
DPLm (5)
distribution power systems, the basic objective functions k
g=1 k=1 i=1 m=1
considered in this study are maximisation of the network
loading margin (for operational efficiency) and the
However, the system total loss increase in terms of without
DISCO’s profit (by minimising investment). Each of these
and with DG connection is given in (6)
is treated next in the literature.
B −1
N B −1
N B −1
N
j j j
2.1.1 Loading margin maximisation: Loading margin, DPLm = PLm − PLm DG
(6)
one of the most popular voltage stability indices, could be m=1 m=1 m=1
defined as the amount of load increase necessary to bring a
power system operating at a particular operating point, to To maximise the loading margin of the system, (7) is derived
the verge of voltage collapse. Incidentally, economic and from (5) and (6)
environmental factors impose several restrictions on
network expansion in the electric power industry worldwide
NB
NG
NG N
B −1 N
B −1
today, even as the demand for power increases. Under these max DPDj i = DPGj g + j
PDG k
− j
PLm + j
PLm DG
conditions, power systems are operated close to their i=1 g=1 m=1 m=1 m=1
278 IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276 –287
& The Institution of Engineering and Technology 2011 doi: 10.1049/iet-gtd.2010.0199
www.ietdl.org
The investment cost of DG comprises of the cost of purchase, presented in (15)
installation, operation and maintenance and tax cost and the
cost of fuel for the planning period. The expression for this (F1 (z) − z∗1 )
is presented in (11) mF1 (z) = ∗
for z∗1 ≤ F1 (z) ≤ znad
1 (15)
1 − z1 )
(znad
Tp
Tp
iv
CDG pc
= CDG + CDG
in
+ O&M&T
gt Td CDG + gt CF (11) 2.2.2 Fuzzification of the second objective
t=0 t=0
function – profit maximisation: The second objective
is to maximise the profit for the DISCO as expressed in
Equation (12) is obtained from (9). (12). If mF2(z) is the satisfaction parameter of the second
Let objective function, then the expression in (16) holds
LN N
B −1
(F2 (z) − z∗2 )
for z∗2 ≤ F2 (z) ≤ znad
Tp
j mF2 (z) = ∗ 2 (16)
F2 (z) = min(−PR ) = CE gt Td PLm DG 2 − z2 )
(znad
t=0 j=1 m=1
The overall fuzzy multi-objective function at node k of the
LN
NL
j
Tp
LN
NL
j case-study system can therefore be expressed as
+ iv
CDG PDGk
− CE gt Td PDGk
j=1 k=1 t=0 j=1 k=1
min Fk (z) = W1 mF1 (z) + W2 mF2 (z) (17)
(12) z[S
Equations (8) and (12) are subject to (13) For the sake of simplicity, the two objective functions are
assumed to be equally weighted in this study, that is, W1 ¼
N
B −1 N
B −1
W2 ¼ 0.5. It is, however, worth mentioning that these two
j j
0≤ PLm DG
≤ PLm , weighting factors can be varied in accordance with the
m=1 m=1 preference of the decision maker – the operator in charge.
Since DG is not centrally dispatched, unit commitment
NL
j
NB
schedules are not considered in the formulation of (17)
0≤ PDG k
≤ 0.4 PDj i for j = 1 to LN (13) because they are out of the control of the system operator.
k=1 i=1
Maintaining schedules for lines and units, forced outages
have been taken into account during the off-time of 65 days
The DG considered in this study is a natural gas-driven
per year of the DG operation.
combustion turbine. It is assumed to be operated for 300
The following system-bound constraints are considered in
days per year. This is equivalent to 7200 h per year, for a
this study:
horizon planning period of 20 years. It is also pertinent to
add that the overall DG capacity limit considered in this
Bus voltage constraint
study is 40% of the system peak load demand, to avoid
some power quality and protection issues, which might
arise as the DG penetration level increases. The balance of Vimin ≤ Vi ≤ Vimax ∀i [ {1, 2, 3, . . . , NB } (18)
60% is provided by the central generators.
Feeder power flow constraint
2.2 Membership function for the objective
functions Siy ≤ Siymax , and Syi ≤ Syimax ∀i, y [ {1, 2, 3, . . . , NB }
(19)
Since power loading is measured in MW and the DISCO’s
profit in USD, (14) is presented to separately normalise the
objective functions so that their objective values will be
approximately of the same magnitude. 3 Brief overview of GA
The GA is a general-purpose optimisation and search
(F (z) − z∗x )
mFx (z) = xnad for z∗x ≤ Fx (z) ≤ znad (14) technique that is based on principles inspired by the genetic
(zx − z∗x ) x
evolutionary mechanism observed in the populations of
natural systems and living organisms [21]. Generally, GAs
where zx∗ , called the ideal objective vector, is start with an initial set of random solutions that lie in the
obtained by minimising each objective function feasible solution region, otherwise known as population.
individually, and znad
x , known as the nadir objective vector, Each solution in the population, called a chromosome,
is the upper bounds of the Pareto-optimal set that can be represents a possible solution to the optimisation problem.
estimated from a pay-off table [20]. Now the value for each If the chromosome has N variables given by V1 , V2 ,
new objective function ranges between 0 and 1. In other V3 , . . . , VN , it is written as an N element vector [V1 , V2 ,
words, mFx(z) ¼0 when Fx(z) ¼ zx∗ , and mFx(z) ¼ 1 when V3 , . . . , VN]. A position or set of positions in a
Fx(z) ¼ znad
x . chromosome, which is part of the solution, is called a gene.
The number of variables of the optimisation problem is
2.2.1 Fuzzification of the first objective function – always equal to the number of genes. Each iteration in a
loading margin maximisation: The first objective GA is called a generation. The flowchart shown in Fig. 1
function formulated in (7) is fuzzified. Defining mF1(z) as summarises the general procedure involved in the SGA.
the satisfaction parameter of power loading margin This generational process is repeated until a stopping
maximisation, the membership function for this objective is condition has been reached.
IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276– 287 279
doi: 10.1049/iet-gtd.2010.0199 & The Institution of Engineering and Technology 2011
www.ietdl.org
Fundamental to the GA structure is the encoding
mechanism for representing the variables of the
optimisation problem. A common method of encoding GAs
is binary encoding; other types include permutation, value
and tree encodings. In binary encoding, every chromosome
is a string of bits, 0 or 1. An example of a binary-encoded
chromosome that has N variables, each encoded with 9 bits
is presented as follows.
111110101 011001010 · · · 001011001
chromosome =
gene1 gene2 geneN
280 IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276 –287
& The Institution of Engineering and Technology 2011 doi: 10.1049/iet-gtd.2010.0199
www.ietdl.org
population shows a high level of diversity if the value is low, Step 4: Enter the maximum number of iterations m.
that is, close to 0. D represents the change in the best fitness Step 5: Randomly create binary strings (chromosomes) equal
since the last control action. The input and output quantities to the population size, Np¼ 30. Each chromosome is a
are defined as follows: ( fA)/( fB) [ [0, 1], ( fW)/( fA) [ concatenation of two genes, DG size and bus location of
[0, 1], D [ [0, 14], Pc [ [0.6, 0.9], Pm [ [0.01, 0.015] and eight binary bits each (i.e. p0 ¼ [PsDGBk]), where s ¼ 0 to
Np [ [30, 45] in this study. The linguistic value sets of the Pmax
DG (maximum DG size), and k ¼ 1 to NL . In other
inputs and outputs are all {low, medium, high} with respect words, p0 has 16 × 30 ¼ 480 bits initially.
to the lower- and upper-limit values, respectively. The Step 6: iter ¼ 1.
fuzzy IF – THEN rules in [26] are adopted to combine fuzzy Step 7: Apply each chromosome of p0 on the network to
sets in this paper as presented in Table 6 in the Appendix. calculate Fk(z). If no voltage and MVA line limits, accept
the chromosome for the next generation, else reject the
4 Proposed algorithm for DG placement size – location pair.
Step 8: Create next-generation population:
To solve the fuzzy multi-objective optimisation problem For pS ¼ 1 to Np × SR ( pS is the population of the members
formulated in (17), the FGA is used. The goal of the that survive for the next generation at 50% selection rate).
proposed algorithm is to find the optimal size and site for 1. Select a pair of best-fitted chromosomes, using the roulette
DG placement in the network, while keeping the voltage wheel approach to form P1 .
and line power flow within permissible limits. The 2. Apply crossover rate to breed population P2 .
algorithm is presented in steps as follows: 3. Apply the fuzzy controller using the rules in Table 6, by
adapting Pc to stabilise optimal populations.
Step 1: Input the network data and maximum DG size. 4. Mutate chromosomes in P2 to form P3 .
Step 2: Set the load level j. 5. Apply the fuzzy controller using the rules in Table 6 to
Step 3: Without DG, run a power flow to determine the total maintain the PD, by adapting Pm .
NB
power demand i=1 PDj i of the network. 6. Evaluate the individual fitnesses of P3 using (17).
IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276– 287 281
doi: 10.1049/iet-gtd.2010.0199 & The Institution of Engineering and Technology 2011
www.ietdl.org
Step 12: While total DG size in the network,
NB
T
PDG , 0.4 i=1 PDj i repeat Steps 5 – 12.
Step 13: Stop and end the programme.
Fig. 4 Objective function plot of IEEE 30-bus test system (basecase loading)
282 IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276 –287
& The Institution of Engineering and Technology 2011 doi: 10.1049/iet-gtd.2010.0199
www.ietdl.org
of the test system are 21.25 MW and 5.75 MVars,
respectively. On the other hand, the IEEE 30-bus test
system shown in Fig. 2 is considered as a sub-transmission
system. It consists of 30 buses at 135 kV, 6 generators and
41 lines, and its basecase active and reactive powers are
given as 189.20 MW and 107.20 MVar, respectively. The
line data of this system are provided in Table 5 in the
Appendix.
Fig. 7 Objective function plot of an IEEE 30-bus system for first DG (peak loading)
IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276– 287 283
doi: 10.1049/iet-gtd.2010.0199 & The Institution of Engineering and Technology 2011
www.ietdl.org
capacity limit, with respect to the total load demand, was not bus 10. Zero megawatt of DG simply means without DG
considered in [14]. It is worth mentioning here that higher connection to the system. The obtained results are plotted
penetrations of DG may provoke some power quality and as shown in Fig. 6. The minimum power loss is obtained
protection issues in the system. with the first DG capacity of 40 MW, which actually
corroborates the results obtained with the proposed
6.2 Case II – 30-bus test system algorithm. Any attempt to increase the DG capacity beyond
40 MW at that location would result in an increased system
Similarly, applying the proposed method on the 30-bus test total loss, as portrayed in the figure.
system, 40 and 35 MW of DG units are obtained as the Assuming a power demand growth rate of about 2.5% per
optimal sizes and are, respectively, located at buses 10 and year at a constant power factor, the total load demand of the
6 for the basecase loading. Placing a 40 MW DG alone at system under consideration would have grown by 62% in
different load buses, the objective function values using the horizon planning period of 20 years. The proposed
(17) are plotted and presented in Fig. 4. From the figure, method is applied to this system, having 302.72 MW and
the minimum objective function value of 0.317 is obtained 171.52 MVars as the load active and reactive powers,
at bus 10, and hence declared the first optimal site. This bus respectively. DG units 80 MW and of 40.4 MW rounded to
could be confirmed as the best site from Fig. 5, plotted with 40 MW are obtained as the optimal sizes, and are located at
the results of power flow studies carried out when the DG buses 8 and 19, respectively, this time around. A plot of the
size is placed at different load buses. It is found that the objective function values for the first DG placed on all load
least total real power loss is obtained at bus 10. This buses is illustrated in Fig. 7.
process is repeated for the second DG unit, whose size and Considering Fig. 5, the vitality of the best location for DG
location are, respectively, found to be 35 MW and bus 6. units in power systems is apparent. It is noted in the figure that
To verify the optimality of DG capacity obtained, several with the optimal DG size located at buses 26, 29 and 30 in
power flow simulations are carried out in MATPOWER each case, higher total system loss is recorded separately in
[32] with different DG sizes from 0 to 70 MW, located at each of these locations. It could be observed that each of
Load level
Wt DG No DG Wt DG No DG Wt DG No DG
Year Purchase and Operation and maintenance Fuel cost, $ Energy Annual PWF Annual discounted
installation cost, $ and tax cost, $ loss cost, $ cost, $ cost, $
0 172 200 000.00 4 320 000.00 46 033 920.00 1 372 026.60 223 925 946.60 1.000 223 925 946.60
1 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.916 47 389 781.58
2 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.839 43 417 115.52
3 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.769 39 777 476.42
4 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.705 36 442 946.79
5 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.645 33 387 949.42
6 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.591 30 589 051.23
7 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.542 28 024 783.54
8 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.496 25 675 477.36
9 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.455 23 523 112.56
10 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.417 21 551 179.62
11 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.382 19 744 553.02
12 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.350 18 089 375.19
13 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.320 16 572 950.25
14 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.294 15 183 646.58
15 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.269 13 910 807.68
16 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.246 12 744 670.34
17 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.226 11 676 289.82
18 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.207 10 697 471.21
19 0.00 4 320 000.00 46 033 920.00 1 372 026.60 51 725 946.60 0.189 9 800 706.56
total 1 206 718 932.00 682 125 291.31
284 IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276 –287
& The Institution of Engineering and Technology 2011 doi: 10.1049/iet-gtd.2010.0199
www.ietdl.org
Table 4 Twenty-year cost evaluation of substation and feeder expansion
Year Substation and Operation and Additional energy Energy Annual PWF Annual
feeder expansion maintenance and purchase cost, $ loss cost, $ cost, $ discounted
cost, $ tax cost, $ cost, $
0 190 080 000.00 2 592 000.00 103 680 000.00 3 331 944.00 299 683 944.00 1.000 299 683 944.00
1 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.916 100 415 890.06
2 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.839 91 998 066.93
3 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.769 84 285 906.49
4 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.705 77 220 253.31
5 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.645 70 746 910.96
6 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.591 64 816 226.26
7 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.542 59 382 708.44
8 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.496 54 404 680.20
9 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.455 49 843 958.04
10 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.417 45 665 559.36
11 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.382 41 837 434.13
12 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.350 38 330 219.09
13 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.320 35 117 012.45
14 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.294 32 173 167.61
15 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.269 29 476 104.09
16 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.246 27 005 134.30
17 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.226 24 741 304.90
18 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.207 22 667 251.40
19 0.00 2 592 000.00 103 680 000.00 3 331 944.00 109 603 944.00 0.189 20 767 064.95
total 2 382 158 880.00 1 270 578 796.96
IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276– 287 285
doi: 10.1049/iet-gtd.2010.0199 & The Institution of Engineering and Technology 2011
www.ietdl.org
2 CIGRE WG 37-23: ‘Impact of increasing contribution of dispersed 29 Dondi, P., Bayoumi, D., Haederli, C., Julian, D., Suter, M.: ‘Network
generation on the power system’. Final Report, 1998 integration of distributed power generation’, J. Power Sources, 2002,
3 Ackermann, T., Andersson, G., Söder, L.: ‘Distributed generation: 106, (1– 2), pp. 1– 9
a definition’, Electr. Power Syst. Res., 2001, 57, (3), pp. 195– 204 30 Willis, H., Scott, W.: ‘Distributed power generation: planning and
4 Chiradeja, P., Ramakumar, R.: ‘An approach to quantify the technical evaluation’ (CRC, 2000)
benefits of distributed generation’, IEEE Trans. Energy Convers., 19, 31 U.S. Energy Information Administration: ‘Natural Gas Weekly Update’,
(4), pp. 764– 773 15 March 2010, http://tonto.eia.doe.gov/oog/info/ngw/ngupdate.asp
5 Willis, H.L., Inc, A., Raleigh, N.C.: ‘Analytical methods and rules of 32 Zimmerman, R., Gan, D.: ‘Matpower’ (Power Systems Engineering
thumb for modeling DG-distribution interaction’. IEEE Power Research Center, School of Electrical Engineering, Cornell University,
Engineering Society Summer Meeting, 2000, pp. 1643–1644 Ithaca, NY, 1997)
6 Gozel, T., Hocaoglu, M.H., Eminoglu, U., Balikci, A.: ‘Optimal placement
and sizing of distributed generation on radial feeder with different static
load models’. Future Power Systems, Int. Conf., 2005, pp. 1–6 10 Appendix
7 Acharya, N., Mahat, P., Mithulananthan, N.: ‘An analytical approach for
DG allocation in primary distribution network’, Int. J. Electr. Power IEEE 30 bus test system line data and fuzzy control rules are
Energy Syst., 2006, 28, (10), pp. 669– 678
8 Elnashar, M.M., El Shatshat, R., Salama, M.M.A.: ‘Optimum siting and
given below.
sizing of a large distributed generator in a mesh connected system’,
Electr. Power Syst. Res., 2010, 80, (6), pp. 690–697
9 Hedayati, H., Nabaviniaki, S.A., Akbarimajd, A.: ‘A method for
placement of DG units in distribution networks’, IEEE Trans. Power Table 5 IEEE 30-bus test system line data
Delivery, 2008, 23, (3), pp. 1620–1628
10 Golshan, M.E.H., Arefifar, S.A.: ‘Optimal allocation of distributed Branch From To R, X, B, MVA
generation and reactive sources considering tap positions of voltage no bus bus pu pu pu limit
regulators as control variables’, Eur. Trans. Electr. Power, 2007, 17,
(3), pp. 219– 239 1 1 2 0.02 0.06 0.03 130
11 Kumar, V., Kumar, H.C.R., Gupta, I., Gupta, H.O.: ‘DG integrated 2 1 3 0.05 0.19 0.02 130
approach for service restoration under cold load pickup’, IEEE Trans. 3 2 4 0.06 0.17 0.02 65
Power Deliv., 2010, 25, (1), pp. 398 –406 4 2 5 0.05 0.20 0.02 130
12 Singh, R.K., Goswami, S.K.: ‘Optimum allocation of distributed
5 2 6 0.06 0.18 0.02 65
generations based on nodal pricing for profit, loss reduction, and
voltage improvement including voltage rise issue’, Int. J. Electr. 6 3 4 0.01 0.04 0.00 130
Power Energy Syst., 2010, 32, (6), pp. 637– 644 7 4 6 0.01 0.04 0.00 90
13 El-ela, A., Allam, S., Shatla, M.: ‘Maximal optimal benefits of 8 4 12 0.00 0.26 0.00 65
distributed generation using genetic algorithms’, Electr. Power Syst. 9 5 7 0.05 0.12 0.01 70
Res., 2010, 80, (7), pp. 869– 877
14 Singh, R.K., Goswami, S.K.: ‘Optimum siting and sizing of distributed 10 6 7 0.03 0.08 0.01 130
generations in radial and networked systems’, Electr. Power Compon. 11 6 8 0.01 0.04 0.00 32
Syst., 2009, 37, (2), pp. 127– 145 12 6 9 0.00 0.21 0.00 65
15 Haghifam, M.R., Falaghi, H., Malik, O.P.: ‘Risk-based distributed 13 6 10 0.00 0.56 0.00 32
generation placement’, IET Gener. Transm. Distrib., 2008, 2, (2),
14 6 28 0.02 0.06 0.01 32
pp. 252–260
16 Wang, C., Nehrir, M.H.: ‘Analytical approaches for optimal placement 15 8 28 0.06 0.20 0.02 32
of distributed generation sources in power systems’, IEEE Trans. 16 9 10 0.00 0.11 0.00 65
Power Syst., 2004, 19, (4), pp. 2068–2076 17 9 11 0.00 0.21 0.00 65
17 Jabr, R.A., Pal, B.C.: ‘Ordinal optimisation approach for locating and 18 10 17 0.03 0.08 0.00 32
sizing of distributed generation’, IET Gener. Transm. Distrib, 2009, 3,
(8), pp. 713– 723 19 10 20 0.09 0.21 0.00 32
18 Celli, G., Ghiani, E., Mocci, S., Pilo, F.: ‘A multiobjective evolutionary 20 10 21 0.03 0.07 0.00 32
algorithm for the sizing and siting of distributed generation’, IEEE 21 10 22 0.07 0.15 0.00 32
Trans. Power Syst., 2005, 20, (2), pp. 750– 757 22 12 13 0.00 0.14 0.00 65
19 Gözel, T., Hocaoglu, M.H.: ‘An analytical method for the sizing and
23 12 14 0.12 0.26 0.00 32
siting of distributed generators in radial systems’, Electr. Power Syst.
Res., 2009, 79, (6), pp. 912– 918 24 12 15 0.07 0.13 0.00 32
20 Miettinen, K.: ‘Nonlinear multiobjective optimization’ (Springer, 1999) 25 12 16 0.09 0.20 0.00 32
21 Momoh, J.A.: ‘Electric power system applications of optimization’ 26 14 15 0.22 0.20 0.00 16
(Marcel Dekker Inc., New York, 2001) 27 15 18 0.11 0.22 0.00 16
22 Srinivas, M., Patnaik, L.M.: ‘Genetic algorithms: a survey’, Computer,
1994, 27, (6), pp. 17–27 28 15 23 0.10 0.20 0.00 16
23 Haupt, R., Haupt, S.: ‘Practical genetic algorithms’ (Wiley-Interscience, 29 16 17 0.08 0.19 0.00 16
2004) 30 18 19 0.06 0.13 0.00 16
24 Wang, K.: ‘A new fuzzy genetic algorithm based on population 31 19 20 0.03 0.07 0.00 32
diversity’, Proc. 2001 IEEE Int. Symp. on Computational Intelligence
32 21 22 0.01 0.02 0.00 32
in Robotics and Automation, 2001, pp. 108 –112
25 Xu, H., Vukovich, G.: ‘A fuzzy genetic algorithm with effective search 33 22 24 0.12 0.18 0.00 16
and optimization’. Int. Joint Conf. on Neural Networks, 1993, 34 23 24 0.13 0.27 0.00 16
pp. 2967– 2970 35 24 25 0.19 0.33 0.00 16
26 Lee, M.A., Takagi, H.: ‘Dynamic control of genetic algorithms using 36 25 26 0.25 0.38 0.00 16
fuzzy logic techniques’. Fifth Int. Conf. on Genetic Algorithms,
Urbana-Champaign, IL, 1993, pp. 76–83 37 25 27 0.11 0.21 0.00 16
27 El-khattam, W., Bhattacharya, K., Hegazy, Y., Salama, M.M.A.: 38 27 29 0.22 0.42 0.00 16
‘Optimal investment planning for distributed generation in a 39 27 30 0.32 0.60 0.00 16
competitive electricity market’, IEEE Trans. Power Syst., 2004, 19, 40 28 27 0.00 0.40 0.00 65
(3), pp. 1674– 1684
41 29 30 0.24 0.45 0.00 16
28 California Energy Commission: ‘California Distributed Energy
Resources Guide’, 15 March 2010, www.energy.ca.gov/distgen/
286 IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276 –287
& The Institution of Engineering and Technology 2011 doi: 10.1049/iet-gtd.2010.0199
www.ietdl.org
Table 6 Fuzzy control rules
Input Output
fA/fB fW/fA D Pc Pm Np
IET Gener. Transm. Distrib., 2011, Vol. 5, Iss. 2, pp. 276– 287 287
doi: 10.1049/iet-gtd.2010.0199 & The Institution of Engineering and Technology 2011