Power Distribution Systems Planning With Distributed

Thermal and Wind Generation

Leonardo W. Oliveira, Flávio V. Gomes, Edimar J. Oliveira, Ângelo R. Oliveira, Abílio M. Variz, Hélio A. Silva
Department of Electrical Engineering
Federal University at Juiz de Fora (UFJF)
Juiz de Fora, Brazil
[email protected]

Abstract—This paper presents a new approach for distribution optimization approaches known as heuristic techniques have
systems planning with thermal and wind distributed generation been developed. Among this sort of methods, there are genetic
(DG). This approach determines the optimal placement of bio- algorithm [6], [7], evolutionary programming [8], differential
mass-based thermal and wind distributed generation aiming at evolution [9] and particle swarm optimization (PSO) [10].
minimizing the energy loss, emissions and the total investment
and operation costs, as well as improving the voltage regulation. Reference [11] proposes a PSO-based approach for optim-
The costs for the purchased energy from grid and thermal gen- al allocation of renewable DGs including wind power incorpo-
eration based on biomass resource are considered. The wind rating the DG reactive capability. PSO is also used in [12] to
power is given by pre-established nominal power and capacity select the proper installation sites for wind turbines consider-
factor. To solve the mixed integer nonlinear problem, a genetic ing pre-specified range for the power output and power factor,
algorithm (GA) with embedded optimal power flow (OPF) tool and in [13] to place the DG at optimal location with optimal
is proposed. The OPF is used to determine the size of the ther- size and unity power factor. In [14] and [15], genetic algo-
mal generators. Results are presented for well-known systems rithm and PSO are applied to define the optimal size of wind
from literature. turbines taking into account the investment costs.
Index Terms—biomass-based power, distributed generation, Genetic algorithms have recently been proposed for DG
genetic algorithm, wind power. placement in [6], [16]. According to [5], the choice of the ob-
jectives related with DG planning impacts the placement of
I. INTRODUCTION the distributed resources and the rule for this issue is still an
open question.
The penetration of distributed generation in power distri-
bution systems has been increased due to some aspects as a Following the lines of research on the aforementioned
response to the growing load demand. Moreover, other aspects question, the present paper proposes an approach for optimal
have been contributed to the distributed generation capability allocation of distributed generation based on thermal and wind
as the deregulation of power systems, technological advances power, considering biomass resources for the thermal power.
that lead to small generating units with high efficiency and This approach combines a genetic algorithm to place the DG
system loss reduction [1]-[4]. The economical and environ- with an optimal power flow [17] tool for sizing the thermal
mental constraints over the construction of new transmission units. The size and capacity factors of the wind generators are
lines for long distance power delivery also favor the spread of pre-specified values. The objectives are to minimize the net-
DG. The environmental benefits increase when the DG uses work energy losses, emissions and the total investment and
renewable resources, as the biomass and wind power. operation costs, while improving the system voltage regula-
tion. The cost for purchased energy from grid is included in
The potential advantages of DG depend on its proper the total operation cost. Two radial distribution test systems of
placement in the distribution system [1], [5]. If non-optimal 33 and 69 nodes are used for presenting the results.
placements are chosen for the DG penetration, these advan-
tages can be low, null, or the DG can even negatively affect II. PROPOSED MODELING
the system. As example, the DG can increase the system
losses if it is not optimally placed [5]. The optimization problem for optimal distribution systems
planning aims at minimizing the energy loss, emissions and
If the distribution system planning considers the possibility the total investment and operation costs, while improving the
of installation of multiple DGs, the optimal placement be- voltage levels subjected to the network constraints. The objec-
comes a mixed integer, nonlinear and non-convex optimiza- tive function of the problem is based on [11] and formulated
tion problem [1]. To solve this kind of problem, a class of as

The authors gratefully acknowledge the contributions of FAPEMIG,

CAPES, CNPq, INERGE and the ‘‘Bio-inspired and Heuristic Optimiza-
tion’’ research group of UFJF.
 ⎡ NTh NWd
⎤ pf th,min ≤ pf th ≤ 1 , (9)
⎢ ∑ ( c th ⋅ S th + OC th ) + ∑ ( cwd ⋅ Swd + OCwd ) ⎥
Min ⎢ t =1 wd =1
⎥ .(1)
⎢⎣ +c pl ⋅ T ⋅ L pl + cgr ⋅ T ⋅ Pgr + cem ⋅ egr ⋅ T ⋅ Pgr ⎥⎦ Pth2 ,k + Q 2th,k ≤ Sth,max , (10)

Where NTh and NWd are the number of thermal and wind Vfb, Vtb ≥ Vmin , (11)
generators, respectively; cth and cwd are the cost factors
($/kVA) for calculating capital costs for thermal and wind xth,k + xwd,k < 2 , (12)
generator, respectively; Sth and Swd are the installed capacity
(kVA) of thermal and wind generator, respectively; OCth and NTh + NWd ≤ NDGmax . (13)
OCwd are the operation costs ($) for thermal and wind genera-
tion, respectively; Lpl is the total active power loss (kW) and Where xth,k and xwd,k are integer values associated with the
cpl is the respective cost ($/kWh); T is the total hours of opera- decisions of thermal and wind generation, respectively, at
tion; cgr is the cost factor ($/kWh) for the purchased energy busbar k (‘0’ for no allocation and ‘1’ for allocation); Pth,k and
from grid and Pgr is the respective power (kW); cem is the cost Pwd,k are the active power outputs (kW) of the thermal and
factor ($/kg) for emissions and egr is the emission factor wind generator at busbar k, respectively; Qth,k and Qwd,k are the
(kg/kWh) associated with Pgr. reactive power outputs (kVAr) at busbar k from thermal and
The operation and maintenance cost for wind generation wind generators, respectively; PLk and QLk are the active
(OCwd) is given by cow·T·Pwd, where cow is the cost factor (kW) and reactive (kVAr) power loads at busbar k, respective-
($/kW) and Pwd is the active power output (kW) of the wind ly; Ωk is the set of busbars directly connected to busbar k; Pkm
generator. The operation cost for thermal generation compris- and Qkm are the active (kW) and reactive (kVAr) power flows,
es a term associated with the operation and maintenance, respectively, through the branch that connects busbars k and
which includes the fuel cost, and another term associated with m; cfwd is the capacity factor of wind-based DG units; pfth and
emissions, as in pfwd are the power factors for thermal and wind-based DG
units, respectively; Sth,max is the maximum allowable thermal
DG capacity (kVA); Vmin is the lower limit for the system
OC th = cot ⋅ T ⋅ Pth + cem ⋅ efu ⋅ T ⋅ Pth . (2) voltage (V); NDGmax is the maximum number of DG units.
Equations (4) and (5) consist of the active and reactive
Where cot is the cost factor ($/kWh) for the biomass-based power balance constraints, respectively, defined for each bus-
thermal generation, Pth is the respective active power output bar of the network. The active power output of wind-based
(kW) and efu is the emission factor (kg/kWh) for the power DG units is given by their capacity factor and rated power as
supplied by biomass generators. in (6). The capacity factor for thermal generators is assumed to
The total active power loss (Lpl) is given by be 1. The reactive power outputs of DGs are obtained from the
respective active power outputs and operating power factors as
in (7) and (8). These power factors may need to be regulated
Nbr
as in (9) [11].
L pl = ∑ g b ⋅ ⎡⎣ Vfb2 + Vtb2 − 2 ⋅ Vfb ⋅ Vtb ⋅ cos ( θfb − θtb ) ⎤⎦ .(3)
b =1 The DG capacity at each busbar is limited by constraint
(10) due to land space and network constraints [11]. The con-
Where Nbr is the total number of branches; gb is the is the straint formulated in (11) ensure that all busbars have their
conductance of branch b; Vfb and Vtb are the voltage magni- voltages above a minimum value and the constraint in (12)
tudes at the end busbars of branch b; θfb and θtb are the phase means that a same busbar cannot receive thermal and wind
angles at the end busbars of branch b. generators considering the land space limit [11]. The con-
The objective function formulated in (1) is subjected to the straint in (13) limits the DG penetration and has to be planned
following constraints in advance at the planning stage [11]. It can be highlighted
that the voltage limit constraint leads to an improvement of the
system voltage regulation.
x th,k ⋅ Pth,k + x wd,k ⋅ Pwd,k − PL k + ∑P
m∈Ωk
km = 0, (4)
III. PROPOSED METHODOLOGY
The proposed methodology for power distribution systems
x th,k ⋅ Q th,k + x wd,k ⋅ Q wd,k − QL k + ∑ Q km = 0 , (5) planning with distributed thermal and wind generation embeds
m∈Ωk an optimal power flow in a genetic algorithm. The genetic
algorithm handles the combinatorial feature of the problem
Pwd,k = cf wd ⋅ Pwdr , (6) due to the integer variables (xth,k , xwd,k) associated with the
decisions about the placements of the distributed resources.
Q th,k = pf th ⋅ Pth,k , (7) Every element of the GA (chromosome) encodes a candi-
date siting solution in a vector (Ind) that has Nbc positions,
Q wd,k = pf wd ⋅ Pwd,k , (8) where Nbc is the number of busbars candidate for DG sites.
Each position of the vector (gene) can receive an integer value
from ‘0’ to ‘2’. Equation (14) gives the codification where bc
is a busbar candidate for DG placement.

⎧0, if there is no DG at bc
⎪
Ind( bc) = ⎨1, if there is a thermal DG at bc . (14)
⎪ 2, if there is a wind DG at bc
⎩

The codification in (14) allows meeting the constraint in

(12) because it does not envisage thermal and wind genera-
tions in a same busbar.
The solution where no allocation occurs in any busbar is
named base case. Firstly, the initial population of the GA is
formed by candidate solutions randomly generated, and then
the base case is added to this population. This procedure helps
the search process finding for solutions where the investment
in DG is attractive. The generation process of the initial popu-
lation meet the constraint in (13) related to the maximum
number of DG units.
Each chromosome of the current population is decoded in
a corresponding siting solution for which the optimal power
flow optimizes the active and reactive power outputs for the
thermal generators, Pth,k and Qth,k, and the purchased energy
from grid, Pgr. The OPF consists of the model presented in (1)- Figure 1. Proposed algorithm.
(11) wherein the discrete variables xth,k and xwd,k are set to
fixed values according to the respective chromosome of the Step-1 generates the initial population with the base case
GA. In the OPF problem, the constraints in (12) and (13) do included as described before in such a way that the constraints
not need to be added because they are automatically met in the in (12)-(13) are met. The individuals of the initial population
solution codified by the chromosome. are decoded as DGs sites and the OPF of Step-2 optimizes the
thermal DG generations aiming at minimizing the objective
In the OPF (1)-(11), the active power outputs from the function in (1) subjected to the constraints (2)-(11). After that,
wind generators, Pwd,k, are set as fixed values in function of the apparent thermal generations are rounded as in (16) to
pre-specified nominal power and capacity factor. This factor is calculate the objective function (1) in Step-3.
associated with the wind resources in the regions where the
distributed units are installed [11]. The reactive power from Steps-4 to 6 apply the genetic operators:
wind units is calculated based on the active power and pre- (i) ‘selection’, using the roulette wheel technique to select
specified power factor. two parent chromosomes from the population according to
From the solution of OPF (1)-(11), the continuous appar- their fitness;
ent power of the thermal generators are calculated as (ii) ‘crossover’, where one cutoff is applied to the middle
2 2 points of a random pair of chromosomes from the previously
Sc th,k = P th , k +Q th , k . (15) selected ones to form a new child chromosome, until the num-
ber of new individuals is the same as the population size;
Where Sth,k is the continuous apparent power (kVA) of the
(iii) ‘mutation’, which consists of adding or subtracting
thermal DG placed at busbar k. The discrete size of the ther-
the value ‘1’ from a random position of the chromosomes of a
mal unit at busbar k is calculated based on the result of (15),
relative small random set.
the limit Sth,max and the discrete interval capacity dith (kVA), as
in The crossover and mutation processes were developed
specifically to meet the constraints (12)-(13) considering the
structure of a chromosome as depicted in (14).
⎛ Sc ⎞
Sth,k = ceil ⎜ th,k ⎟ ⋅ di th . (16) After Step-6, if the convergence criterion is met, then the
⎝ di th ⎠ algorithm terminates; otherwise, the counter of generations is
incremented and the algorithm returns to Step-2 with the up-
Where ceil(x) is a function that rounds ‘x’ to the next dated population. The converge is achieved when the number
higher discrete value. Therefore, the discrete value Sth,k (16) is of generations exceeds a maximum value or when a better
used to calculate the objective function in (1). solution than the current best one is not found for a given
number of generations.
Fig. 1 summarizes the steps of the proposed algorithm that
will be briefly discussed next.
 The parameters of the GA were obtained from the litera- TABLE I. RESULTS FOR 33-BUS TEST SYSTEM – ANALYSIS 1
ture [6] also for the optimal DG placement problem: (i) popu- Method Base Case Proposed Method
lation size 50; (ii) crossover probability 90%; (iii) mutation Allocation -- Busbars 14, 30 (wind)
probability 20%; (iv) elitism 1 chromosome; (v) maximum Investiment cost for
-- 5.66e05
generation number 200; (vi) number of consecutive genera- wind unit ($)
tions that a better chromosome has not been found 25. Operation cost for
-- 6.92e05
wind unit ($)
Investiment cost for
IV. RESULTS biomass-based unit ($)
-- --
The results obtained for two well-known test systems from Operation cost for
-- --
literature are presented here. In all case studies, the overall biomass-based unit ($)
Grid energy cost ($) 6.07e07 5.60e07
planning period was 30 years as in [11] and the cost and emis-
Loss Cost ($) 3.14e06 2.46e06
sion factors were obtained from [11]. As in [18], the nominal Grid emission cost ($) 1.85e07 1.71e07
power of the wind generators was 700 kW and the respective Biomass unit emission
power factor was 0.875, close to the value used in [18], 0.9. -- --
cost ($)
For the considered power factor 0.875, the nominal apparent Total cost ($) 8.23e07 7.68e07
power of wind units was 800 kVA. The minimum and maxi- Minimum Voltage (V) 0.913 0.927
mum capacities of biomass-based thermal generation were
100 kVA and 1500 kVA, respectively, with discrete interval Analysis 2: This analysis considers the voltage limit as
capacity of 100 kVA (15 discrete capacity levels) [11]. Differ- 0.95 pu whereas the other conditions are the same as in the
ent values for capacity factor of wind units, power factor limit previous analysis. For this voltage constraint, the base case
of biomass-based thermal generators, voltage limits and max- where there is no allocation is not feasible. Therefore, the DG
imum number of DG units are used to evaluate the results. planning allows the system operation with this severe con-
A. 33-Bus Test System straint. The results of the proposed method for this analysis are
presented in Table II, which includes the capacity of the bio-
This case study aims at assessing the proposed methodolo- mass-based thermal unit.
gy for optimal placement of distributed thermal and wind gen-
eration in a 12.66 kV 33-node radial distribution system [19].
The system has a total load of 3715 kW. Fig. 2 shows their TABLE II. RESULTS FOR 33-BUS TEST SYSTEM – ANALYSIS 2
structure. Method Proposed Method
Busbar 16 (wind)
Allocation
Busbar 28 (biomass)
Investiment cost for wind unit ($) 2.83e05
Operation cost for wind unit ($) 3.46e05
Investiment cost for biomass-based unit ($) 3.44e06
Operation cost for biomass-based unit ($) 1.96e08
Grid energy cost ($) 3.85e07
Loss Cost ($) 9.25e05
Grid emission cost ($) 1.17e07
Biomass unit emission cost ($) 1.83e04
Total cost ($) 2.51e08
Capacity of biomass-based unit (kVA) 1500 (busbar 28)
Minimum Voltage (V) 0.954
Figure 2. 33-bus test system [18].

Analysis 1: In the first analysis, the capacity factor of From Tables I and II, it can be observed that in Analysis 1
wind-based DG units is 18.8%, which is considered as a rela- there was no allocation of thermal distributed generation, whe-
tive high factor for this kind of generation [11]. Also accord- reas Analysis 2 gave as result the placement of one biomass-
ing to [11], a maximum of two nodes can be selected for in- based thermal unit at busbar 28. This is due to the more severe
stallation of DG units (NDGmax = 2) and the power factor con- voltage constraint considered in Analysis 2 that requires more
straint is relaxed by setting its lower limit as zero. The voltage investment in DG. As the capacity factor of biomass-based
limit is 0.90 pu. Table I presents the results for this analysis, thermal units are greater than that of wind generators and only
including all the related costs and the minimum voltage, for two units can be placed, one thermal unit is placed in Analysis
the base case, where there is no allocation, and for the solution 2 despite their high costs in relation to the wind generation due
obtained by the proposed method. The difference between the to the higher amount of DG power required in Analysis 2. As
total costs is equal to $ 5,561,000.00 and the proposed model a result, the total cost of Analysis 2 is greater than the cost in
leads to a better voltage profile. This model also gives a lower Analysis 1. However, as the emission factor of biomass-based
emission cost. units is smaller than that of the energy from the grid, the total
emission cost of Analysis 2, given by the summation of the
respective costs for the grid and the biomass unit, is smaller
than the emission cost of Analysis 1. This result shows the
environmental benefits provided by the penetration of renew-
able energy sources in power systems.
 Analysis 3: As stated in the previous analysis, the capacity TABLE IV. RESULTS FOR 69-BUS TEST SYSTEM – ANALYSIS 1
factor of DG units is a parameter that affects the DG planning. Method Base Case Proposed Method
Therefore, Analysis 3 considers a lower capacity factor for Allocation -- Busbars 60, 62 (wind)
wind generation, 5.0%, to further evaluate this aspect. The low Investiment cost for
-- 5.66e05
capacity factor is due to supposed poor wind resources. The wind unit ($)
other planning conditions are the same of Analysis 2. Table III Operation cost for
-- 6.92e05
wind unit ($)
presents the results for Analysis 3. Investiment cost for
-- --
biomass-based unit ($)
TABLE III. RESULTS FOR 33-BUS TEST SYSTEM – ANALYSIS 3 Operation cost for
-- --
biomass-based unit ($)
Method Proposed Method Grid energy cost ($) 6.24e07 5.76e07
Busbar 16 (biomass) Loss Cost ($) 3.49e06 2.68e06
Allocation
Busbar 31 (biomass) Grid emission cost ($) 1.91e07 1.76e07
Investiment cost for wind unit ($) -- Biomass unit emission
-- --
Operation cost for wind unit ($) -- cost ($)
Investiment cost for biomass-based unit ($) 5.04e06 Total cost ($) 8.50e07 7.90e07
Operation cost for biomass-based unit ($) 2.92e08 Minimum Voltage (V) 0.909 0.922
Grid energy cost ($) 3.14e07
Loss Cost ($) 6.47e05
Grid emission cost ($) 9.58e06 For this system, when the capacity factor is considered as
Biomass unit emission cost ($) 2.73e04 5.0% there is no allocation of biomass or wind-based genera-
Total cost ($) 3.39e08 tion because the investment in both kinds of DG is not attrac-
800 (busbar 16) tive for this condition due to the relative high costs of bio-
Capacity of biomass-based unit (kVA)
1400 (busbar 31)
Minimum Voltage (V) 0.979
mass-based generation and the low capacity factor of wind
power.
Analysis 2: This analysis considers a maximum of one dis-
As it can be observed from Table III, there was no alloca-
tributed generator to be placed and continuous values for dis-
tion of wind-based generators in Analysis 3 because their ca-
tributed power as in [1]. The maximum capacity of biomass-
pacity factors are low and the requirement for DG power is
based thermal unit is 2300 kVA, only thermal distributed gen-
high due to the severe voltage constraint, 0.95 pu.
eration can be placed, the power factor constraint is relaxed
B. 69-Bus Test System and the voltage limit is 0.95 pu. For this limit, the system can-
not operate without DG allocation. Reference [1] presents a
The 12.66 kV test system is from [20], has one substation case for this system in similar conditions, in which continuous
(SE), a total load of 3802 kW and the topology pictured in Fig. real and reactive power can be allocated independently of each
3 [1], [21]. other, which means that the power factor constraint is relaxed.
SE
Table V presents the results for this analysis. Although the
objective function in [1] is given only by the power loss, the
data of Table V allows comparing the solutions in terms of
each term of the objective function modeled in the present
paper, equation (1), as well as the power loss considered in
[1]. The investment, operation and emission costs for [1] were
calculated in the present work considering the DG based on
biomass resources for comparison purposes.

61 TABLE V. RESULTS FOR 69-BUS TEST SYSTEM – ANALYSIS 2

64 Method Proposed Method [1]

Allocation Busbar 64 (biomass) Busbar 61 (biomass)
Investiment cost for
4.57e06 5.15e06
biomass-based unit ($)
Operation cost for
2.74e08 3.09e08
biomass-based unit ($)
Figure 3. 69-bus test system [1], [21]. Grid energy cost ($) 3.44e07 3.10e07
Loss Cost ($) 6.32e05 3.59e05
Grid emission cost ($) 1.05e07 9.45e06
Analysis 1: This analysis considers the capacity factor of Biomass unit emission
wind-based DG units as 18.8% and a maximum of two nodes cost ($)
2.56e04 2.88e04
for placement of DG [11], the power factor constraint relaxed Total cost ($) 3.24e08 3.54e08
and the voltage limit as 0.90 pu. Table IV presents the results. Minimum Voltage (V) 0.971 0.973
Capacity of biomass-
1992 (busbar 64) 2244 (busbar 61)
based unit (kVA)
Loss (kW) 40.76 23.17
 Analysis 3: In this analysis, only real power is considered REFERENCES
to be allocated in the test system in similar conditions of [1] B.H Dias, L.W. Oliveira, F.V. Gomes, I.C. Silva Jr., and E.J. Oliveira,
another case of [1]. In the present paper, it was done by setting "Hybrid heuristic optimization approach for optimal distributed genera-
the lower power factor to 1.0. The other conditions are the tion placement and sizing," in Proc. 2012 IEEE/PES General Meeting,
same of the previous analysis. pp.6.
[2] J. Olamaie and T. Niknam, "Daily Volt/Var control in distribution
networks with regard to DGs: a comparison of evolutionary methods,"
TABLE VI. RESULTS FOR 69-BUS TEST SYSTEM – ANALYSIS 3 in Proc. 2006 IEEE Power India Conf., pp.6.
[3] T. Ackerman, G. Anderson, and L. Soder, "Distributed generation: A
Method Proposed Method [1] definition," Electr. Power Systems Res., vol. 57, no. 3, pp. 195-204,
Allocation Busbar 64 (biomass) Busbar 61 (biomass) Apr. 2001.
Investiment cost for [4] P. Chiradeja and R. Ramakumar, "An approach to quantify the technic-
3.78e06 4.29e06
biomass-based unit ($) al benefits of distributed generation," IEEE Trans. Energy Conversion,
Operation cost for vol.19, no. 4, pp. 764-773, Dec. 2004.
2.78e08 3.16e08
biomass-based unit ($) [5] P. Kayal and C.K. Chanda, "Placement of wind and solar based DGs in
Grid energy cost ($) 3.49e07 3.12e07 distribution system for power loss minimization and voltage stability
Loss Cost ($) 1.50e06 1.29e06 improvement," Int. J. Electr. Power Energy Systems, vol.53, pp. 795-
Grid emission cost ($) 1.06e07 9.52e06 809, Dec. 2013.
Biomass unit emission [6] V.A. Evangelopoulos and P.S. Georgilakis, "Optimal distributed gener-
2.06e07 2.95e04 ation placement under uncertainties based on point estimate method
cost ($)
Total cost ($) 3.29e08 3.62e08 embedded genetic algorithm," IET Generation, Transmission & Distri-
Minimum Voltage (V) 0.967 0.968 bution, vol. 8, no. 3, pp. 389-400, Mar. 2014.
Capacity of biomass- [7] G. Celli, E. Ghiani, S. Mocci, and F. Pilo, "A multiobjective evolutio-
1647.7 (busbar 64) 1872.7 (busbar 61) nary algorithm for the sizing and siting of distributed generation," IEEE
based unit (kVA)
Loss (kW) 96.59 83.22 Trans. Power Syst., vol.20, no. 2, pp. 750-757, May 2005.
[8] D.K. Khatod, V. Pant, and J. Sharma, "Evolutionary programming
based optimal placement of renewable distributed generators," IEEE
Trans. Power Systems, vol. 28, no. 2, pp. 683-695, May 2013.
From the results of Analyses 2 and 3, it can be observed [9] L.D. Arya, A. Koshti, and S.C. Choube, "Distributed generation plan-
that the DG places obtained by the proposed method and in [1] ning using differential evolution accounting voltage stability considera-
are different from each other. In both analyses, the model of tion," Int. J. Electr. Power Energy Systems, vol. 42, no. 1, pp. 196-207,
[1] found the solution with lower loss whereas the proposed Nov. 2012.
method found the solution with lower total cost, as expected in [10] M.G. Gonzalez, A. Lopez, and F. Jurado, "Optimization of distributed
terms of the respective objective functions. These results generation systems using a new discrete PSO and OPF," Electr. Power
Systems Res., vol. 84, no. 1, pp. 174-180, Mar. 2012.
prove that the objectives related with DG planning impacts its [11] K. Zou, A.P. Agalgaonkar, K.M. Muttaqi, and S. Perera, "Distribution
placement [5]. However, both results identify the same prom- system planning with incorporating DG reactive capability and system
ising region in the system for DG penetration since busbars 61 uncertainties," IEEE Trans. Sustainable Energy, vol. 3, no. 1, pp. 112-
and 64 are close to each other as highlighted in Fig. 3. 123, Jan. 2012.
[12] J.-N. Sheen, M.-T. Tsai, and S.-W. Wu, "A benefits analysis for wind
turbine allocation in a power distribution system," Energy Conversion
V. CONCLUSIONS and Management, vol. 68, pp. 305-312, Apr. 2013.
This paper presented an optimization model for planning [13] N. Jain, S. N. Singh, and S. C. Srivastava, "A generalized approach for
of distribution systems that performs the optimal allocation of DG planning and viability analysis under market scenario," IEEE
Trans. Indust. Electronics, vol. 60, no. 11, pp. 5075-5085, Nov. 2013.
distributed generation from the wind and biomass-based ther- [14] G. Mokryani and P. Siano, "Optimal wind turbines placement within a
mal power. This model consists of a multi-objective problem distribution market environment," Applied Soft Computing, vol. 13, no.
that aims at minimizing the network energy loss, emissions 10, pp. 4038-4046, Oct. 2013.
and costs of investment and operation, including the cost for [15] P. Siano and G. Mokryani, "Assessing wind turbines placement in a
purchased energy from grid. Moreover, the distributed genera- distribution market environment by using particle swarm optimization,"
tion allocation allows improving voltage regulation. For this IEEE Trans. Power Systems, vol. 28, no. 4, pp. 3852-3864, Nov. 2013.
[16] M.H. Moradi and M. Abedini, "A combination of genetic algorithm and
task, a genetic algorithm along with an optimal power flow particle swarm optimization for optimal DG location and sizing in dis-
tool was proposed. The genetic algorithm handles the binary tribution systems," Int. J. Electr. Power Energy Systems, vol. 34, no. 1,
variables that define the distribution power placement while pp. 66-74, Jan. 2012.
the optimal power flow optimizes the thermal power genera- [17] S. Granvile, "Optimal reactive dispatch through interior point me-
tion. The wind power is given in function of pre-specified thods," IEEE Trans. Power Systems, vol. 9, no. 1, pp. 136-146, Feb.
1994.
nominal power and capacity factor. The study cases showed [18] L. Chen, J. Zhong, and D. Gan, " Reactive power planning and its cost
the applicability of the presented model, which leads to a re- allocation for distribution systems with distributed generation," in Proc.
duction in the total cost as well as in the emission and loss 2006 IEEE/PES General Meeting, pp.6.
costs. Besides, the proposed approach showed to be effective [19] M.E. Baran and F.F. Wu, "Network reconfiguration in distribution
in finding promising regions for distributed generation pene- systems for loss reduction and load balancing," IEEE Trans. Power De-
livery, vol. 4, no. 2, pp. 1401-1407, Apr. 1989.
tration considering more than one objective. [20] M.E. Baran and F.F. Wu, "Optimal capacitor placement on radial dis-
tribution systems," IEEE Trans. Power Delivery, vol. 4, no. 1, pp. 725-
ACKNOWLEDGMENT 734, Jan. 1989.
[21] M. F. AlHajri, M. R. AlRashidi, and M. E. El-Hawary, "Hybrid particle
The authors gratefully acknowledge the contributions of swarm optimization approach for optimal distribution generation sizing
FAPEMIG, CAPES, CNPq, INERGE and the ‘‘Bio-inspired and allocation in distribution systems," in Proc. 2007 IEEE Canadian-
and Heuristic Optimization’’ research group of UFJF. Conf. on Electr. and Computer Engineering, pp. 1290-1293.