See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/4226157

Optimal placement and sizing of distributed generation on radial feeder with

different static load models

Conference Paper · December 2005

DOI: 10.1109/FPS.2005.204319 · Source: IEEE Xplore

CITATIONS READS

85 2,164

4 authors, including:

Tuba Gozel Mehmet Hakan Hocaoglu

Gebze Technical University Istanbul Ticaret University
49 PUBLICATIONS   1,024 CITATIONS    125 PUBLICATIONS   1,864 CITATIONS   

SEE PROFILE SEE PROFILE

Abdulkadir Balikci
Gebze Technical University
19 PUBLICATIONS   429 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

GREENDC View project

Sustainable Energy Demand Side Management for GREEN Data Centers (GreenDC) View project

All content following this page was uploaded by Mehmet Hakan Hocaoglu on 21 May 2014.

The user has requested enhancement of the downloaded file.

 1

Optimal Placement and Sizing of Distributed

Generation on Radial Feeder with Different
Static Load Models
Tuba GÖZEL M. Hakan HOCAOGLU Ulas EMINOGLU Abdulkadir BALIKCI

location and size of DG for minimum power loss could not

Abstract--Due to the increasing interest on renewable sources take generation cost into account.
in recent times, the studies on integration of distributed In order to minimize line losses of power systems, it is
generation to the power grid have rapidly increased. In order to crucially important to determine the size and location of local
minimize line losses of power systems, it is crucially important to generation to be placed. There have been number of studies to
define the size and location of local generation to be placed.
define the optimum location of DG. The mathematical
There have been studies, to define the optimum location of
approaches on the optimum DG placement for minimum
distributed generation. Additional to the current studies in the
literature, in this study, the optimum size and location of power losses are as follows: optimal load flow with second
distributed generation that will be placed in the radial system order algorithm method [4], genetic algorithm and Hereford
will be defined by an analytical method so as to minimize total Ranch algorithm which can find optimum [5], Fuzzy-GA
power loss for the uniformly, centrally and increasingly method [6], tabu search approach [7], the algorithm to
distributed system profile. Additionally the analytical approach determine the near optimal [8], 2/3 rule, which is often used in
will also be evaluated against classical grid search algorithm for capacitor placement studies [9], and analytical approach in
different load models.1 radial as well as networked systems [10].
Index Terms-- distributed generation, radial systems, power Particularly, reference [10] demonstrates an analytical
loss, optimal placement, static load models
approach to determine exclusively the optimal location to
place a DG in radial systems to minimize the total loss of the
I. INTRODUCTION
system. This study takes the size of DG as total load size and
In recent times, due to the increasing interest on renewable in respect of the size of DG obtains the optimal location of
sources such as; hydro, wind, solar, geothermal, biomass and DG in radial systems without optimizing size of DG.
ocean energy, all over the world, the number of studies on In all of the studies, cited above [4]-[10], the loads are
integration of distributed resources to the grid have rapidly generally modeled as constant power or constant current types
increased. Distributed generation (DG), which consists of of loads. Since most of the distribution system loads are
distributed resources, can be defined as electric power uncontrolled, effects of this type of load models on optimum
generation within distribution networks or on the customer sizing and location should be questioned. Accordingly, in this
side of the network [1]. paper, an analytical approach, which is used to optimally size
DG affects the flow of power and voltage conditions at and locate the DG to the radial systems, will be parametrically
customers and utility equipment. These impacts may manifest analyzed by giving particular emphasis to the effects of load
themselves either positively or negatively depending on the models.
distribution system operating characteristics and the DG The study, presented in this paper, undertakes an analysis
characteristics. Positive impacts are generally called “system on the effects of the load modeling to the optimal placement
support benefits”, and include: voltage support and improved and size of DG in a radial feeder which is determined using
power quality; loss reduction; transmission and distribution the analytical approach [10] to minimize the power losses for
capacity release; improved utility system reliability. On uniformly, centrally and increasingly distributed load profile.
account of achieving the above benefits, the DG must be In order to do so, a radial feeder is simulated with different
reliable, dispatchable, of the proper size and at the proper load types to examine the system loss and voltage profile. The
locations [2], [3]. values of DG size and placement for minimizing power loss
Energy cost of distributed resources as compared to fossil which is determined by the analytical approach is validated
resources is generally high whereat the factors of social and against the results obtained by the classical grid search
environmental benefits could not be included in cost account. algorithm for each types of loads.
Accordingly, most of the studies to determine the optimum
II. THEORETICAL ANALYSIS

This work is supported by State Planning Organization of Turkey (Project To simplify the analysis, transmission lines with uniformly
No: 2003K120530). distributed parameters are considered, R and L per unit length
 2

are the same along the feeder while shunt capacitor (C and G) For uniformly distributed load profile, the phasor current
per unit length are neglected. Loads are distributed along the density Id(x) can be used (8). Besides, for centrally and
line with the phasor current density Id(x). The phasor feeder increasingly distributed load profile, the phasor current
current at point x is densities Id(x) can be taken (9) and (10) to calculated total
x
(1) power loss with.
∫
I ( x) = Id ( x)dx
0

Assuming the per unit length impedance of the line is

Z=R+jX (Ω/km), the length of line is u, the end of line is
Id ( x) = I (8)
chosen as reference. Thus, the total power loss is
2
u
x 
Ploss = ∫  ∫ Id ( x) dx  .Rdx (2)
 
0 0 

The voltage drop between point x and receiving end can  I .x 0≤ x≤u/2
Id ( x) =  (9)
be calculated by using (3). 
x x
 I .(u − x) u/2≤ x≤u
Vdrop( x) = ∫ ∫ Id ( x)dx.Zdx
0 0
(3)

Consider a DG is added into the feeder at the location x0,

Id ( x) = I .(u − x) (10)
as injected current source Idg. When the change in the load
current density, resulted from the addition of DG, is neglected,
the feeder current after adding DG can be written as follows:
 x The derivations of total power loss per size of DG, Idg, and
 ∫ Id ( x ) dx 0 ≤ x ≤ x0
 location of DG, x0, are obtained for all distributed load
I ( x ) =  x0 (4)
 Id ( x ) dx − Idg profile. The values of the derivations, being equal to zero, are
 ∫0
x0 ≤ x ≤ u
found out as the optimal size and placement of DG.
Power loss and voltage drop after adding DG is obtained
from (4). The total power loss and voltage drop in the feeder
are given in (5) and (6) respectively. B. The Results of Theoretical Analysis Optimal Size and
2 2 Placement for adding DG

x0 x
 u
x 
Ploss = ∫  ∫ Id ( x) dx  .Rdx + ∫  ∫ Id ( x)dx − Idg  .Rdx (5) The solution of (7) for the uniformly distributed system
   
0 0  x 0 0  2
gives that the optimal size of DG is determined as 3 I.u by
xx
 ∫ ∫ Id ( x)dx.Zdx 0 ≤ x ≤ x0 analytically and its size equals to 67% of the total load.
 (6)
Vdrop( x) =  x00 x0 x x Accordingly, it is obtained that for centrally distributed and
 Id ( x)dx.Zdx + ( Id ( x)dx − Idg )Zdx x0 ≤ x ≤ u
∫∫
 0 0
∫∫
x0 0
increasingly distributed systems, the optimum size of DG
equals to 80% of the total load size. The optimal placement of
A. Determination of Optimal Size and Placement for added DG is provided such as 0.33, 0.446 and 0.225 of radial feeder
DG length from the end of the feeder for uniformly, centrally and
The goal is to minimize total power loss in the system by increasingly distributed system, respectively.
adding a DG on a particular point with optimum size. The
objective is to minimize power losses, Ploss, in the system by Fig. 1 shows that for each system (uniformly, increasingly
injected current, Idg, from a particular place, x0. The main and centrally distributed load profile) the optimal placement
constraints are to restrain the voltages along the feeder within of DG, x0 curve, which changes with the size of DG from
1±0.05pu and maximum DG size must be selected as total zero to total load size, total power loss, Plossx0 curve, which
corresponds the optimal placement of DG, x0 curve, in the
load size ( ∑ I load ).
same way the optimal size of DG, Idg curve, which changes
d d
Ploss Ploss 0 with the placement of DG from the receiving end (zero) to the
dx0 dIdg (7)
sending end (u, the length of line), total power loss, PlossId,
curve, which corresponds the optimal size of DG, Idg curve.
Optimal placement and sizing of DG is determined by The intersection point of the x0 curve and the Idg curve is the
using (7) then the voltages along the feeder are checked to be optimal size and placement for placing DG. In Fig 1, this
within the acceptable range. If the voltages along the feeder intersection point is shown as the minimum of the Plossx0
are not satisfied, optimal sizing and placement of DG are curve and PlossIdg curve.
changed to the nearest values to take the feeder voltages to
the voltage limits.
 3

In Table 1, theoretical results of optimal placement and size

of DG to minimize power loss are given for each distributed
system. The total power loss of the each distributed system
decreases when DG is located by optimally as demonstrated in
[10]. Moreover, when the size of DG is optimized, the
decrease on the power losses becomes more significant. For
the optimal placing of DG with and without optimal size, it is
seen that optimal placement of DG affected by the load profile
when DG is located by optimally with its optimal value. On
the other hand, optimal placement of DG is the same for the
uniformly and centrally distributed load profile when only
optimal location of DG is used.
(a) Uniformly distributed load profile
III. STATIC LOAD MODELS
Load models, used in power system studies, are classified
into two categories: static models and dynamic models. Static
load models for active and reactive power are expressed in a
polynomial or an exponential form. The characteristic of the
exponential load models can be given as:

np nq
V  V  (11)
P = Po   , Q = Qo  
 Vo   Vo 

Where np and nq stand for load exponents, Po and Qo stand

for real and reactive powers at the nominal voltage. V and Vo
(b) Centrally distributed load profile stand for load bus voltage and load nominal voltage,
respectively. Special values of the load exponents can cause
specific load types such as 0: constant power 1: constant
current 2: constant impedance. The values of these
coefficients are determined for different load types in
transmission and distribution systems. Usually data,
determined from experience, could be used for the estimation.
Common values for the exponents of static load model for
different load components are given in Table 2.

(c) Increasingly distributed load profile

Fig. 1 The optimal placement and size for adding DG and total power loss

TABLE 1
THEORETICAL ANALYSIS RESULTS OF A RADIAL FEEDER WITH DIFFERENT LOAD TYPES

Without With DG With DG

DG calculated upon optimal placement calculated upon optimal placement and size
Percent of Percent of
Load Types Total power Total power
Optimal
power loss Total power
Optimal Optimal size
power loss
placement of placement of placement of
loss loss reduction loss reduction
DG (x0) DG (x0) DG (x0)
(%) (%)
1 2 3 1 2
Uniformly Distributed Load I 2 Ru3 / 3 I 2 Ru 3 / 12 υ/2 %75.00 I Ru u
3
I.u %88.89
27 3
23
I 2 Ru 2.10−3 I 2 Ru 5
5
1 2
Centrally Distributed Load 960 I 2 Ru 5 / 320 υ/2 %86.96 0.446.υ I .u %91.52
5

0.29.υ %88.39 9.4.10−3 I 2Ru5 2 2

%92.95
Increasingly Distributed Load 0.133I 2 Ru 5 0.0155I 2 Ru5 0.225.u
5
I .u
 4

TABLE 2 satisfied the optimum values are omitted and again the
COMMON VALUE OF EXPONENTS FOR DIFFERENT STATIC LOAD MODELS
optimum values are searched at the voltage limits. The same
[11], [12]
flowchart is applied for OPDG without changing DG size.
Load Component np nq
In Table 3, optimal values for different static load models
Battery Charge 2.59 4.06
are shown. For OPSDG, optimal DG’s size is decreased in all
Fluorescent Lamps 2.07 3.21
distribution types when load voltage dependence is increased.
Constant Impedance 2 2
Optimal location remains unchanged in centrally distributed
Air Conditioner 0.5 2.5
profile. In uniformly and increasingly distributed types,
Constant Current 1 1
optimal location is shifted one bus from the source. For
Pumps, Funs other Motors 0.08 1.6
OPDG, optimal location can not be determined with high
Compact Fluorescent Lamps 0.95-1.03 0.31-0.46
voltage dependent loads. On account of at high voltage
Small Industrial Motors 0.1 0.6
dependent loads the voltages along the feeder go beyond to
Constant Power 0 0
the limit and the system power loss with OPDG is more than
the case without DG.
Exponential load modelsmay be valid for only a limited
voltage range, which are ± 10% of 1 pu voltage level. For the
BEGIN

high and low voltage magnitude levels, the models are

inadequate for some load type i.e. motors and lighting [11]- SET system
[13]. The load types have important effects of the power param eters

system studies such as; power flow analysis and voltage

stability. Traditionally, most of the conventional load flow
FO R np=nq=0:10
methods, for transmission and distribution systems, use the (np, nq load exponents)
constant-power load model. The constant power load model is
highly questionable, especially for a distribution system where
FO R x0=1:N
most of the buses are uncontrolled. For this reason, static load (x0=bus no for adding DG)
models, given in (11), are more important and it must be taken (N=total bus number)
into distribution systems analysis such as determination of the
optimal size and placement for adding DG in radial systems.
FOR idg=0:ΣPload
(idg=D G size ΣPload=total load )
IV. SIMULATIONS AND ANALYSIS
A radial feeder with 13-bus, as shown in Appendix Fig A1, Calculate
power flow and pow er loss
is used as a test network to analyze the effects of load
modeling on the optimum size and location of DG by varying
load exponents from 0 to 10 having load profile with
uniformly, centrally and increasingly distributions. A power
Find
flow algorithm, developed for radial distribution systems with m inimum power loss
voltage sensitive loads [13], is employed for the solution.
O m it
System parameters, load sizes for all load types and total load unsuitable values
sizes are given in Appendix Table A1. 0.95pu≤V ≤1.05pu
are all bus voltages
Two different cases are examined. In the first case, optimal in acceptable range? No
placement is obtained by adding DG, whose size is equal to
the size of total load to minimize power loss (OPDG). In the Y es
second case, optimal placement and size for adding DG is Determ ine x0, idg
obtained (OPSDG). OPDG and OPSDG are calculated by a for m inimum
power loss
grid search algorithm for each load model. General flowchart
for grid search algorithm to minimize power losses is shown
in Fig. 2. The grid search algorithm is applied by adding DG Print all optim al
to each bus, changing the size of DG from 0% of total load values for np=nq=0:10
power to 100% of total load power DG in the ratio of 10% and
varying load exponents from 0 to 10 having load profile with
uniformly, centrally and increasingly distributions. The END

optimal size and placement of placing DG to minimize power

loss is determined. Thereafter, the voltages along the feeder Fig. 2 General flowchart of grid search algorithm
are checked to be in acceptable range. If the voltages are not
 5
TABLE 3
OPTIMAL SIZE AND PLACEMENT FOR PLACING DG WITH DIFFERENT STATIC LOAD MODELS

Power losses, given in Fig. 3, are calculated according to

different placement along the feeder and different sizing of
DG with constant power, current and impedance for each
distributed load profile. From Fig. 3, it is seen that optimal
size and placement of DG reach an agreement with the
theoretical analysis for constant power, current and impedance
load models.

V. CONCLUSION

(a) Uniformly distributed load profile

This study presents and evaluates an analytical method
which is used to determine the optimal placement and sizing
of DG in a radial feeder, so as to minimize total power loss for
the uniformly, centrally and increasingly distributed system
profile. The optimal size and location of the DG, which is
determined by the analytical approach detailed here, is also
evaluated against classical grid search algorithm for different
load models. It is indicated that; optimal size and placement of
DG are different for each distributed load profile; optimal size
and placement of the theoretical analysis are valid for constant
power, current and impedance load models. It is found that
while optimum size of the DG is heavily under influence of
(b) Centrally distributed load profile the load models, the optimum location does not change with
the chosen model.

In this study, the optimal size and placement of DG with

various load types are determined for different static load
models. In practice, the load profile is not as given in this
study even so general information concerning sizing and
placement of DG is assured. The determination of the optimal
size and placement for multiple DGs, and with a networked
system will be undertaken by considering the other constraints
as a future work.

(c) Increasingly distributed load profile

Fig. 3 Total power losses in respect of placing DG each size and location
with constant power, current and impedance
 6

VI. APPENDIX

Fig. A1 .A Radial Feeder

TABLE AI
PARAMETERS OF THE SYSTEM IN FIGURE 2

R = 0.538 Ω/km X = 0.4626 Ω/km R=0.8608 pu X=0.74016 pu

Line Parameters
Line length between two buses : 2.5km Bus Voltage = 12.5 kV Total
Load at each bus [MW] Load
Load Profile
1 2 3 4 5 6 7 8 9 10 11 12 13 [MW]
Uniformly distributed 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 3.9
Centrally distributed 0.06 0.12 0.18 0.24 0.3 0.36 0.42 0.36 0.3 0.24 0.18 0.12 0.06 2.94
Increasingly distributed 0.09 0.12 0.15 0.18 0.21 0.24 0.27 0.3 0.33 0.36 0.39 0.42 0.45 3.51

VII. REFERENCES VIII. BIOGRAPHIES

[1] T. Ackermann, G. Anderson, L. Söder, “Distributed Generation: a Tuba Gözel was born in Konya in Turkey. She received
Definition”, Electric Power System Research 57, 2001, pp.195-204
the B.Sc degree in Electrical-Electronics Engineering from
[2] P. P. Barker, R. W. de Mello, “Determining the Impact of Distributed
Generation on Power Systems: Part 1 – Radial Distribution Selcuk University in 1994 and M. Sc degree in Electrical-
Systems”, IEEE PES Summer Meeting, Vol.3, 2000, pp.1645-1656 Electronics Engineering from Gebze Institute of
[3] N. Hadjsaid, J. F. Canard, and F. Dumas, “Dispersed Generation Technology in 2002. She has worked as research assistant
Impact on Distribution Networks”, IEEE Compt. Appl. Power, vol.
and has continued to Ph.D. program at the same institute.
12, Apr. 1999, pp.22–28
[4] S. Rau and Y.H. Wan, “Optimum Location of Resources in M. Hakan Hocaoglu was born in Hatay in Turkey. He
Distributed Planning”, IEEE Trans. Power Syst., vol. 9, , Nov. 1994, received the B.Sc. and M.Sc. degrees from Marmara
pp.2014–2020 University, Turkey. He obtained the Ph.D. degree in 1999
[5] K. H. Kim, Y. J. Lee, S. B. Rhee, S. K. Lee, and S.-K. You, from Cardiff School of Engineering, UK. From 1988 to
“Dispersed Generator Placement Using Fuzzy-GA in Distribution
Systems”, IEEE PES Summer Meeting, Vol. 3, July 2002, pp.1148– 1993, he worked at Gazinatep University, Turkey as a
1153 Lecturer. Since 1999, he has been with the Electronics
[6] J. O. Kim, S. W. Nam, S. K. Park, and C. Singh, “Dispersed Engineering Department of Gebze Institute of Technology,
Generation Planning Using Improved Hereford Ranch Algorithm”, Turkey as an Assistant Professor. He is a member of IEE.
Electric Power System Research , vol. 47, no. 1, Oct. 1998, pp.47–55
[7] K. Nara, Y. Hayashi, K. Ikeda, and T. Ashizawa, “Application of Ulas Eminoglu was born in Kars in Turkey, on
Tabu Search to Optimal Placement of Distributed Generators”, IEEE November 25, 1978. He received the B. Sc degree in
PES Winter Meeting, 2001, Volume: 2, pp.918 – 923 Electrical-Electronics Engineering from Inonu University
[8] T. Griffin, K. Tomsovic, D. Secrest, and A. Law, “Placement of in 2000 and M. Sc degree in Electrical-Electronics
Dispersed Generation Systems for Reduced Losses”, 33rd Annu.
Hawaii Int. Conf. Systems Sciences, Maui, HI, 2000; pp.1-9
Engineering from Nigde University in 2003. He joined
[9] H. L. Willis, “Analytical Methods and Rules of Thumb for Modeling Gebze Institute of Technology in 2003 as research assistant
DG-Distribution Interaction”, IEEE PES Summer Meeting, vol. 3, and has studied distribution systems load flow analyses and
Seattle, WA, July 2000, pp. 1643–1644 power electronics. He has continued to Ph.D. program at
[10] C. Wang, M. H. Nehrir, “Analytical Approaches For Optimal
Placement Of DG Sources In Power Systems”, IEEE Trans. on
the same institute from 2003.
Power Syst., Vol. 19, No. 4, November 2004; pp. 2068–2076 Abdulkadir Balikci was born in Kahramanmaras,
[11] T. V. Cutsem and C. Vournas, “Voltage Stability of Electric Power Turkey. He received his B.Sc. degree from Gazi
Systems,” Power Electronics and Power System Series, Kluwer, University, Ankara, Turkey in 1992. From 1994 to 2003,
1998
[12] C. W. Taylor, “Power System Voltage Stability,” Electric Power
he pursued his M.Sc. and Ph.D. degrees in the Department
Research Institute: McGraw-Hill; USA 1994: pp 17-135. of Electric and Computer Engineering, Polytechnic
[13] U. Eminoglu, M. H. Hocaoglu, “A New Power Flow Method for University, Brooklyn, New York. He received his M.Sc.
Radial Distribution Systems Including Voltage Dependent Load Degree in January 1997 and his Ph.D. Degree in June
Models”, Electric Power System Research , vol. 76, no. 1, Sept.
2005, pp.106–114
2003. He is currently working at Gebze Institute of
Technology, Gebze, Turkey as an Assistant Professor.

View publication stats