Professional Documents
Culture Documents
Abstract—An efficient analytical (EA) method is proposed for inappropriate DG allocation may lead to increased system
optimally installing multiple distributed generation (DG) technolo- losses and system operation costs [9], [10]. It is also a fact that
gies to minimize power loss in distribution systems. Different DG most of the electrical power losses in electric power systems
types are considered, and their power factors are optimally cal-
culated. The proposed EA method is also applied to the problem are dissipated in distribution systems due to heavy currents
of allocating an optimal mix of different DG types with various flowing in primary and secondary feeders. Therefore, there
generation capabilities. Furthermore, the EA method is integrated is a critical need to develop efficient tools that can optimally
with the optimal power flow (OPF) algorithm to develop a new allocate different DG types in distribution systems, thereby
method, EA-OPF which effectively addresses overall system con- reducing losses.
straints. The proposed methods are tested using 33-bus and 69-bus
distribution test systems. The calculated results are validated using Several methods have recently been proposed for the plan-
the simulation results of the exact optimal solution obtained by an ning of distribution systems with DG to minimize losses. These
exhaustive OPF algorithm for both distribution test systems. The methods can be classified as numerical-based (NB), heuristic-
results show that the performances of the proposed methods are based (HB), and analytical-based (AB) methods [10]. The most
superior to existing methods in terms of computational speed and common examples of NB methods are gradient search (GS)
accuracy.
[11], linear programming (LP) [12], optimal power flow (OPF)
Index Terms—DG power factor, distribution systems, optimal [13], and exhaustive search (ES) [14], [15]. The GS, LP, and
DG location, optimal DG size, power loss reduction. OPF algorithms are considered efficient ways for obtaining the
optimal DG sizes at certain locations. The ES algorithm is based
on searching for the optimal DG location for a given DG size
I. INTRODUCTION
or under different load models. Therefore, these methods fail to
represent the accurate behavior of a DG optimization problem
I N recent years, the use of distributed generation (DG) tech- that involves two continuous variables, both optimal DG size
nologies has remarkably increased worldwide due to their and optimal DG location. The HB methods are based on em-
potential benefits. DG units generate power near load centers, ploying advanced artificial intelligence (AI) techniques, such as
avoiding the cost of transporting electric power through trans- genetic algorithms (GAs) [16], [17], particle swarm optimiza-
mission lines. Another benefit of DG is cost savings in elec- tion (PSO) [18], harmony search (HS) [19], and tabu search
tricity production compared with large centralized generation [20]. The main feature of these methods is their computational
stations [1]. Furthermore, renewable DG technologies, such as robustness. They can provide near-optimal solutions but involve
wind power, photovoltaic (PV), and solar thermal systems, are intensive computational efforts.
considered to be one of the fundamental strategies in the fight It is notable that great interest is directed to the AB methods,
against climate change, as they can reduce dependence on fossil as they are easy to implement and fast. AB methods often follow
fuels [2]–[5]. various strategies to simplify the optimization problem, either
With the rapid increase of DG penetration, distribution by assuming uniformly distributed loads as in [21] or by allo-
systems are being converted from passive to active networks. cating only a single DG unit in the entire system [21], [22]. Ref-
Normally, DG units are small in size and modular in structure. erence [23] has proposed a method for determining the optimal
Therefore, their impacts on distribution system operation, locations of multiple DG units, while the corresponding optimal
control, and stability vary depending on their locations and DG sizes are obtained by the Kalman filter algorithm. A load
sizes [6], [7]. One of the most common positive impacts of DG centroid concept [24] is proposed in [25] for allocating mul-
is the ability to reduce distribution system losses [8]. However, tiple DG units. The authors of [26] have proposed an approach
to allocate a single DG unit that operates at unity power factor,
which has recently been extended to an improved analytical (IA)
Manuscript received June 19, 2014; revised October 29, 2014, November 26,
2014, January 20, 2015, March 09, 2015; accepted March 26, 2015. Paper no. method [27]. The IA method involves allocating a single DG
TPWRS-00835-2014. with various capabilities to generate both active and reactive
K. Mahmoud is with the Faculty of Engineering, Aswan University, 81542
power. More recently, the IA method has been upgraded to solve
Aswan, Egypt, and also with the Graduate School of Engineering, Hiroshima
University, 739-8527 Hiroshima, Japan (e-mail: karar.alnagar@aswu.edu.eg). the multiple DG allocation problem [28] and validated by com-
N. Yorino is with the Graduate School of Engineering, Hiroshima University, parison with the exhaustive power flow solution. The main idea
739-8527 Hiroshima, Japan (e-mail: yorino@hiroshima-u.ac.jp).
of the IA method for allocating multiple DG units is to update
A. Ahmed is with the Faculty of Engineering, Aswan University, 81542
Aswan, Egypt. the load data after each time the DG is allocated to determine
Digital Object Identifier 10.1109/TPWRS.2015.2418333 the next DG location. After each DG placement, the calculated
0885-8950 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
(1) (2)
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
MAHMOUD et al.: OPTIMAL DISTRIBUTED GENERATION ALLOCATION IN DISTRIBUTION SYSTEMS FOR LOSS MINIMIZATION 3
C. Generalized RPL Formula With Multiple DG III. GENERALIZED MODELS FOR DIFFERENT DG TYPES
For handling multiple DG allocation, let the list of nodes that Different DG technologies can generally be classified into
are connected to DG units be denoted by NDG. A general RPL three main types based on their active and reactive power gener-
formula can be written as follows: ation characteristics, as illustrated in Fig. 2. The figure describes
the possible energy sources and conversion devices for each DG
type. Combining different energy sources with different energy
converters represents special DG generation characteristics for
each configuration. The bounds of the decision variables, the ac-
tive and reactive DG powers, are specified for each DG type. For
(3) DG type 1, if is equal to zero, its power factor is unity.
DG type 2 represents those that can support reactive powers.
The power factor of DG type 3 may not be specified. By these
where represents a binary matrix whose ele- constraints, optimal values of decision variables will be deter-
ments are defined as follows: mined. Note that the DG power factor will also be determined
in the optimization problem. For a specific DG unit type, if its
if passes through branch optimal active and reactive generated powers are defined, the
(4)
otherwise. interfaced device and the DG technology structure can be opti-
Matrix is employed here to define the list of branches that mally selected and designed [29], [30].
each DG generated power passes through. By using the pro-
IV. PROPOSED EA METHOD
posed mathematical formulations, we can directly evaluate the
losses after installing DG. The main idea of the EA method is based on employing the
proposed RPLR formula as an indicator for the amount of loss
D. Proposed RPLR Formula reduction as a result of installing the DG units. The details of
The basic formulation of RPLR can be expressed as follows: the EA method are listed as follows:
A. Optimal DG Sizing
(5) The first step to solve the problem of DG allocation is by in-
troducing an efficient way to calculate the optimal DG size at a
Substituting (1) and (3) into (5) leads to given bus in a distribution system. Fig. 3 shows the influence
of changing both the active and reactive DG generated power
on the value. The main goal of the optimization
problem is to calculate the optimal DG size to
maximize the value of , i.e., minimize system losses.
The methodology for calculating optimal DG sizes mainly de-
pends on the DG power factor operating conditions as follows.
1) DG With Specified Power Factors: The specified gener-
(6) ation values of DG type 1 and DG type 2 are
treated as negative loads. Therefore, the power factors for DG
Equation(6) is useful for computing the total RPLR by the DG types 1 and 2 can be dealt with as unity and zero, respectively,
by evaluating only the branch losses in the BDG list. There is no during the optimization process.
need to calculate total power loss before and after installing DG The first derivative of the RPLR formula with respect to the
to evaluate the benefits in terms of loss reduction. Substituting real generated power of the DG at bus can be expressed as
the DG power factor into the RPLR formula yields follows:
(7) (8)
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
where
(11b)
(11c)
(12)
(13)
(10)
Solving (13) for the DG reactive powers gives:
The above equation is available for each DG at a typical bus
, so that the set of equations can be organized in matrix nota-
tion as follows: ..
.
..
.
.. .. .. .. ..
. . . . .
.. .. .. .. ..
. . . . .
.. (14)
.
(11a)
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
MAHMOUD et al.: OPTIMAL DISTRIBUTED GENERATION ALLOCATION IN DISTRIBUTION SYSTEMS FOR LOSS MINIMIZATION 5
where
Fig. 5. Simple distribution system with one loop. (a) Meshed distribution
system. (b) Equivalent radial distribution system.
Equations (11) and (14) are solved in a sequential manner and the specified DG generated powers into (6), and by using
to obtain the optimal DG power factors. The solution process the power flow results of the base case, we obtain an estimated
will be repeated until convergence is obtained. The DG power value for the RPLR by
factor mismatch is used as convergence criteria. The proposed
DG sizing algorithm is illustrated in Fig. 4. (16)
It should be noted that the power flow is carried out only once
to obtain the base case loading so that a direct optimal solu-
tion of the DG allocation problem can be efficiently solved. The
complete computational procedure of the proposed EA method
is shown in Fig. 6.
MAHMOUD et al.: OPTIMAL DISTRIBUTED GENERATION ALLOCATION IN DISTRIBUTION SYSTEMS FOR LOSS MINIMIZATION 7
TABLE I
COMPARISON OF DIFFERENT ALGORITHMS FOR THE 33-BUS AND 69-BUS
SYSTEMS WITH DG TYPE 1
TABLE II
POWER LOSS ATTAINED BY EACH METHOD WITH DIFFERENT DG POWER
FACTORS FOR THE 33-BUS SYSTEM
TABLE III (14) in the following iterations to provide the optimal values.
RESULTS OF INSTALLING DG TECHNOLOGIES OF TYPE 3 IN THE TEST SYSTEMS Thus, the RPL minimization is effectively obtained.
Fig. 12 illustrates the effects of increasing the number of DG
units on their total size and the amount of resulting loss re-
duction. For the two systems, the RPLR value and the corre-
sponding total DG size increase dramatically when increasing
the DG number from one to four units. However, the figures
rise only slightly for five and six DG installations. This implies
that the proposed method is also useful for determining the op-
timal number of DG units to be installed to obtain a desired loss
reduction and maximize DG penetration.
To demonstrate the real contribution of calculating the op-
timal DG power factors for loss reduction, the following two
cases are studied. The first case (case 1) is allocating DG with
specified power factors (equal to the total load power factor)
while the second case (case 2) involves implementing the pro-
To compare the amounts of loss reduction attained by posed EA method to calculate the optimal DG power factors. It
methods with different DG power factors, we provide Table II, is interesting to note that the relative loss reduction between the
which compares the RPL by different methods for the 33-bus two cases, calculated by ,
system. The results show that the lowest RPL values are ob- increases with respect to the number of DG units, as shown in
tained with the proposed methods compared with Method 2 Fig. 13. This indicates that calculating the optimal DG power
and Method 3. For instance, for the case of installing three factors can play a vital role in reducing losses, especially in al-
DG units with 0.82 lagging power factors, the RPL values are locating multiple DG units.
reduced to 15.52 and 14.58 kW when using the proposed EA
and EA-OPF methods, respectively. In contrast, Methods 2 D. Optimal Mix of Different DG Types
and 3 provide losses of 23.05 and 53.72 kW, respectively. This This section involves a solution methodology of the optimal
is because Methods 2 and 3 often fail to determine optimal mix of different DG types to minimize RPL. Four scenarios are
locations, especially for multiple DG installations. studied as follows.
• Scenario 1) DG type 1 and DG type 2.
C. DG Type 3 With Unspecified Power Factors
• Scenario 2) DG type 2 and DG type 3.
This section addresses the results of installing DG technolo- • Scenario 3) DG type 1 and DG type 3.
gies with unspecified power factors. Table III summarizes the • Scenario 4) Mix of DG type 1, DG type 2 and DG type 3.
results in terms of optimal DG locations, DG sizes, DG power Here, the main target is to select the optimal scenario among
factors and RPL with DG. The results show that the RPL is them. The optimal DG mix is computed by comparing the
reduced to the lowest value by the proposed methods. This is amount of RPLR for each scenario. The results are illustrated in
reasonable since, unlike the existing methods, the proposed Table IV for the two test systems. It is interesting to note that,
methods compute the optimal power factors to minimize RPL. for all four DG combinations, the RPL is significantly reduced
Fig. 11 shows the convergence characteristics of the EA when compared with the base case. The main observations
method in the 33-bus system for installing three DG units of about the differences among the scenarios are summarized as
type 3 at buses 13, 24, and 30. The DG power factors and the follows:
corresponding RPL computed in each iteration are illustrated in • The maximum RPLR occurs in scenario 4. Therefore, sce-
the figure. At the initial point, iteration 0, the DG power factors nario 4, which involves a mix of the three DG types, is the
are set to 0.82 lagging, which are improved by using (11) and optimal scenario.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
MAHMOUD et al.: OPTIMAL DISTRIBUTED GENERATION ALLOCATION IN DISTRIBUTION SYSTEMS FOR LOSS MINIMIZATION 9
TABLE IV
COMPARISON OF THE SCENARIOS FOR THE TEST SYSTEMS
REFERENCES
[1] H. L. Willis and W. G. Scott, Distributed Power Generation: Planning
and Evaluation. New York, NY, USA: Marcel Dekker, 2000.
[2] O. M. Toledo, D. O. Filho, A. S. A. C. Diniz, J. H. Martins, and M.
H. M. Vale, “Methodology for evaluation of grid-tie connection of dis-
tributed energy resources—Case study with photovoltaic and energy
storage,” IEEE Trans. Power Syst., vol. 28, no. 2, pp. 1132–1139, May
Fig. 13. Relative loss reduction between the two cases for the test systems. 2013.
[3] A. G. Exposito, A. J. Conejo, and C. Canizares, Electric Energy Sys-
tems: Analysis and Operation. Boca Raton, FL, USA: CRC/Taylor
& Francis, 2008.
• Scenario 1 is not recommended, as the RPLR is the lowest [4] X. Zhang, G. G. Karady, and S. T. Ariaratnam, “Optimal allocation
among the scenarios. In addition, the total DG size is the of CHP-based distributed generation on urban energy distribution net-
works,” IEEE Trans. Sustain. Energy, vol. 5, no. 1, pp. 246–253, Jan.
largest, implying that the installation cost is the highest. 2014.
• The scenarios that include DG type 3 tend to provide an [5] X. Zhang, R. Sharma, and Y. He, “Optimal energy management of a
effective RPLR. Therefore, DG type 3 is superior to the rural microgrid system using multi-objective optimization,” in Proc.
IEEE PES Innovative Smart Grid Technol. Conf., Washington, DC,
other two DG types in terms of loss reduction. USA, 2012, pp. 1–8.
Because the proposed methods are based on generalized [6] R. A. Walling, R. Saint, R. C. Dugan, J. Burke, and L. A. Kojovic,
mathematical models, the study can be extended to any DG “Summary of distributed resources impact on power delivery systems,”
IEEE Trans. Power Del., vol. 23, no. 3, pp. 1636–1644, Jul. 2008.
combination case. [7] T. Ackermann and V. Knyazkin, “Interaction between distributed gen-
eration and the distribution network: Operation aspects,” in Proc. IEEE
VII. CONCLUSION T&D Conf., 2002, vol. 2, pp. 1357–1362.
[8] D. Singh and K. S. Verma, “Multiobjective optimization for DG plan-
The paper has presented two methods, EA and EA-OPF, for ning with load models,” IEEE Trans. Power Syst., vol. 24, no. 1, pp.
determining the optimal DG locations and sizes in distribution 427–436, Feb. 2009.
systems to minimize system losses. Two test systems have been [9] Y. M. Atwa, E. F. El-Saadany, M. M. A. Salama, and R. Seethapathy,
“Optimal renewable resources mix for distribution system energy loss
used to validate the proposed methods, and a detailed compar- minimization,” IEEE Trans. Power Syst., vol. 25, no. 1, pp. 360–370,
ison has been conducted with alternative methods in the liter- Feb. 2010.
ature. The proposed methods have been applied to the optimal [10] P. S. Georgilakis and N. D. Hatziargyriou, “Optimal distributed gen-
eration placement in power distribution networks: Models, methods,
DG mix problem to allocate different types of DG units in distri- and future research,” IEEE Trans. Power Syst., vol. 28, no. 3, pp.
bution systems. Furthermore, the proposed methods have been 3420–3428, Aug. 2013.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
[11] N. S. Rau and Y.-H. Wan, “Optimum location of resources in [29] M. J. E. Alam, K. M. Muttaqi, and D. Sutanto, “A three-phase power
distributed planning,” IEEE Trans. Power Syst., vol. 9, no. 4, pp. flow approach for integrated 3-wire MV and 4-wire multigrounded LV
2014–2020, Nov. 1994. networks with rooftop solar PV,” IEEE Trans. Power Syst., vol. 28, no.
[12] A. Keane and M. O'Malley, “Optimal allocation of embedded genera- 2, pp. 1728–1737, May 2013.
tion on distribution networks,” IEEE Trans. Power Syst., vol. 20, no. [30] K. Mahmoud and M. Abdel-Akher, “Analysis of hybrid photovoltaic
3, pp. 1640–1646, Aug. 2005. and wind energies connected to unbalanced distribution systems,” in
[13] G. P. Harrison and A. R. Wallace, “Optimal power flow evaluation of Proc. IEEE Int. Conf. Power and Energy, Kuala Lumpur, 2010, pp.
distribution network capacity for the connection of distributed genera- 79–84.
tion,” roc. Inst. Electr. Eng.—Gener. Transm. Distrib., vol. 152, no. 1, [31] M. A. Kashem, V. Ganapathy, G. B. Jasmon, and M. I. Buhari, “A
pp. 115–122, Jan. 2005. novel method for loss minimization in distribution networks,” in Proc.
[14] D. Zhu, R. P. Broadwater, K. S. Tam, R. Seguin, and H. As- IEEE Int. Conf. Elect. Utility Deregulation Restruct. Power Technol.,
geirsson, “Impact of DG placement on reliability and efficiency with 2000, pp. 251–256.
time-varying loads,” IEEE Trans. Power Syst., vol. 21, no. 1, pp. [32] M. E. Baran and F. F. Wu, “Optimal sizing of capacitors placed on a
419–427, Feb. 2006. radial distribution system,” IEEE Trans. Power Del., vol. 4, no. 1, pp.
[15] D. Singh, R. K. Mirsa, and D. Singh, “Effect of load models in dis- 735–743, Jan. 1989.
tributed generation planning,” IEEE Trans. Power Syst., vol. 22, no. 4,
pp. 2204–2212, Nov. 2007.
[16] K. Vinothkumar and M. P. Selvan, “Fuzzy embedded genetic algorithm
method for distributed generation planning,” Electr. Power Compon.
Syst., vol. 39, no. 4, pp. 346–366, Feb. 2011. Karar Mahmoud (S'14) received the B.S. and
[17] K.-H. Kim, Y.-J. Lee, S.-B. Rhee, S.-K. Lee, and S.-K. You, “Dispersed M.Sc. degrees in electrical engineering from Aswan
generator placement using fuzzy-GA in distribution systems,” in Proc. University, Aswan, Egypt, in 2008 and 2012,
IEEE Power Eng. Soc. Summer Meeting, 2002, vol. 3, pp. 1148–1153. respectively. He is currently working toward the
[18] W. Prommee and W. Ongsakul, “Optimal multiple distributed gener- Ph.D. degree at the Electric Power and Energy
ation placement in microgrid system by improved reinitialized social System Laboratory (EPESL), Graduate School of
structures particle swarm optimization,” Eur. Trans. Electr. Power, vol. Engineering, Hiroshima University, Hiroshima,
21, no. 1, pp. 489–504, Jan. 2011. Japan.
[19] M. F. Shaaban, Y. M. Atwa, and E. F. El-Saadany, “DG allocation Since 2010, he has been an Assistant Lecturer
for benefit maximization in distribution networks,” IEEE Trans. Power with the Faculty of Engineering, Aswan University,
Syst., vol. 28, no. 2, pp. 639–649, May 2013. Aswan, Egypt. He was a Research Assistant with
[20] M. E. H. Golshan and S. A. Arefifar, “Optimal allocation of distributed the Aswan Power Electronics Application Research Center (APEARC), Egypt,
generation and reactive sources considering tap positions of voltage from 2009 to 2012. His research interests include modeling, analysis, control,
regulators as control variables,” Eur. Trans. Electr. Power, vol. 17, no. and optimization of distributed systems with distributed generations.
3, pp. 219–239, May 2007.
[21] H. L. Willis, “Analytical methods and rules of thumb for modeling
DG-distribution interaction,” in Proc. IEEE Power Eng. Soc. Summer
Meeting, Seattle, WA, Jul. 2000, vol. 3, pp. 1643–1644.
[22] C. Wang and M. H. Nehrir, “Analytical approaches for optimal place- Naoto Yorino (M'90) received the B.S., M.Sc., and
ment of distributed generation sources in power systems,” IEEE Trans. Ph.D. degrees in electrical engineering from Waseda
Power Syst., vol. 19, no. 4, pp. 2068–2076, Nov. 2004. University, Waseda, Japan, in 1981, 1983, and 1987,
[23] S.-H. Lee and J.-W. Park, “Selection of optimal location and size of respectively.
multiple distributed generations by using Kalman filter algorithm,” He is a Professor, Vice Dean, Faculty of Engi-
IEEE Trans. Power Syst., vol. 24, no. 3, pp. 1393–1400, Aug. 2009. neering, Hiroshima University, Hiroshima, Japan.
[24] T. Xiao-bo and W. Xue-hong, “A new method of distributed generation He was with Fuji Electric Co. Ltd., Japan, from 1983
optimal placement based on load centroid,” in Proc. IEEE Power and to 1984. He was a Visiting Professor with McGill
Energy Eng. Conf., 2011, pp. 1–5. University, Montreal, QC, Canada, from 1991 to
[25] A. Elmitwally, “A new algorithm for allocating multiple distributed 1992.
generation units based on load centroid concept,” Alexandria Eng. J., Dr. Yorino is a member of CIGRE, iREP, and
vol. 52, no. 4, pp. 655–663, Dec. 2013. ESCJ. He was a Vice President of PE&S, the IEE of Japan.
[26] N. Acharya, P. Mahat, and N. Mithulananthan, “An analytical ap-
proach for DG allocation in primary distribution network,” Int. J.
Electr. Power Energy Syst., vol. 28, no. 10, pp. 669–678, Dec. 2006.
[27] D. Q. Hung, N. Mithulananthan, and R. C. Bansal, “Analytical expres- Abdella Ahmed received the B.S. and M.Sc. degrees in electrical engineering
sions for DG allocation in primary distribution networks,” IEEE Trans. from Assiut University, Egypt, 1974 and 1978, respectively, and the Ph.D. de-
Energy Convers, vol. 25, no. 3, pp. 814–820, Sept. 2010. gree from Hamburg University, Hamburg, Germany, in 1984.
[28] D. Q. Hung and N. Mithulananthan, “Multiple distributed generator He is a Professor with the Faculty of Engineering, Aswan University, Aswan,
placement in primary distribution networks for loss reduction,” IEEE Egypt. His research interests including electrical power and high-voltage
Trans. Ind. Electron., vol. 60, no. 4, pp. 1700–1708, Apr. 2013. engineering.