Академический Документы
Профессиональный Документы
Культура Документы
Optimal Placement
& sizing of
Distributed Generator (DG)
- A Single Objective Approach
CHAPTER - 5
Distributed Generation (DG) for Power Loss Minimization
5.1 Introduction
Distributed generators [2] are very commonly employed to provide active and reactive
power compensation in distribution systems. Installations of Distributed generators are
important for active and reactive power planning of a distribution system. Due to high
concentration of inductive loads in distribution system, power losses are more. The
installations of Distributed generators are necessary for many reasons. The power injections
from DG units located close to the load centers provide an opportunity for system power loss
reduction, cost reduction, voltage profile improvement, voltage stability improvement,
environmental friendliness, postponement system upgrading and increasing reliability. This
impact can be enhanced via optimal DG placement and sizing.
From the literature survey, it is observed that most of these population based
optimization techniques have successfully used to determine optimal placement and sizing of
DG for power loss minimization in RDS. These motivate to introduce new, simple, efficient
and fast population based optimization method to solve optimal DG placement and sizing
problem of RDS.
79
bus and the optimal DG size that would minimize the network active power losses. The
proposed approach addresses the two sub problems simultaneously by using modified TLBO
algorithm that is capable of handling multiple DG planning in a single run. The proposed
algorithm adopts the distribution power flow algorithm developed in chapter - 2.
Three distributed generators that operated at different power factors are installed in to
the system. Two scenarios of each DG source are tested. The first considers the DG units with
a fixed power factor of unity (unity p.f.), while the second has a fixed power factor of 0.866
leading (0.866 p.f. leading). To show the effectiveness and superiority, the performance of the
proposed method was tested on 33 and 69 bus RDS and the results were compared with the
results of other popular optimization techniques.
The rest of the chapter is organized as follows. In Section 5.2, model of a radial
distribution system with the inclusion of DG units is discussed. The problem formulation with
system constraints are addresses in Section 5.3. Application of MTLBO to solve the optimal
DG placement and sizing problem in RDS is explained in Section 5.4. In Section 5.5, DG
placement & sizing evaluation indices are described. Simulation results and analysis are
reported in Section 5.6. Finally, conclusions are drawn in Section 5.7.
• Power flow calculation is performed using S base = 100MVA and V base =12.66KV.
• Three small distributed generators that operated at different power factors i.e. inject
only pure real power (unit p.f.) or inject real power and reactive power (0.866 p. f.
leading) are installed in to the system.
• The bus at which load is connected is considered as the location for DG.
• The limits of DG unit sizes for installation at different systems bus locations
assumed to be 0 to 1.2 MW.
• The upper and lower limits of voltage for each bus are 1.05p.u. and 0.95 p.u.,
respectively.
80
• The maximum allowable number of the parallel DG is one, in each bus.
• The load model used in the simulations is uniform with constant power.
81
A distributed generation (DG) unit can be modeled as either a voltage-controlled bus
(PV bus) or as a complex power injection (PQ bus) in the distribution system. If DGs have
control over the voltage by regulating the excitation voltage (synchronous generator DGs) or
if the control circuit of the converter is used to control P and V independently, then the DG
unit may be modeled as a PV type. Other DGs, like induction generator – based units or
converters used to control P and Q independently, are modeled as PQ types. The PV model
regulates the terminal bus voltage by adjusting their reactive power output. However, it is
preferred to not to use a PV model, since injecting a great amounts of reactive power in order
to raise the bus voltage may result in high field currents and overheating for the generator,
triggering the excitation limit and disconnecting the generator from the network. The most
commonly used DG model is the PQ model. It should be noted that in PQ model, the DG is
considered as negative load. In this work, the DG units are represented as a negative PQ load
model delivering active and reactive power to a distribution system. This gives flexibility in
modeling various types of DG.
In general, DG can be classified into four major types [80] based on their terminal
characteristics in terms of active and reactive power delivering capability as follows :
• Type 1 : DG capable of injecting real power only, like photovoltaic, fuel cell
etc. ( p. f DG = 1 )
• Type 2 : DG capable of injecting both real and reactive power, e.g. synchronous
machines. ( 0 < p. f DG < 1, leading)
• Type 3 : DG capable of injecting real but consuming reactive power, e.g. induction
generators used in the wind farms. ( 0 < p. f DG < 1 , lagging)
• Type 4 : DG capable of injecting reactive power only to improve the voltage profile,
e.g. kvar compensator, synchronous compensator, capacitors etc.( p. f DG = 0 )
Qi = QDGi − QLi = (α
i × PDGi ) − QLi (5.2)
82
where Q DGi = α i × P DGi (5.3)
αi = ( sign ) ( (
× tan cos −1 p. f DGi )) (5.4)
The power factor depends on the DG type and operation condition of the DG. For the
type 1: p. f DG = 1, for the type 2: 0 < p. f DG < 1 and sign = +1, for the type 3: 0 <
With the proposed methodology, it is possible to handle four different types of DGs.
However, in this work, the DG units are modeled as type – 1 DG operated at unity power
factor and type - 2 DG operated at 0.866 power factor leading.
The active power loss of the line section connecting buses i and j derived in chapter -
2 is used in this chapter.
2
Pi + Qi
2
P L o ss ( i , j ) = R ij ⋅ 2 (5.5)
V i
The objective function for total active power loss of the all lines sections is described as
n−1 (P
2
i + Qi2 ) nb (P
2
i + Qi2 )
f 1 = PT ,Loss = ∑ Rij ⋅ 2
= ∑ Rij ⋅ 2 (5.6)
ij =1 Vi ij =1 Vi
83
5.3.1.2 Minimization of Voltage deviation ( f 2 )
The objective function for minimization of voltage deviation is defined as
n
f 2 = ∑ V i , nom − V i (5.7)
i =1
n n
P SUB + ∑ P DG ,i = ∑ P Li + P T , Loss (5.10)
i =1 i =1
n n
Q SU B + ∑ Q D G ,i = ∑ Q Li + Q T , Los s (5.11)
i =1 i =1
Vi
min
≤ V i ≤ V imax (5.12)
84
5.3.2.2.3 Radial structure of the network
P DG ,i ≤ P DG ,i ≤ P DG ,i
min max
(5.16)
Q min max
DG ,i ≤ Q DG ,i ≤ Q DG ,i (5.17)
Step 1: Read the system data, constraints, the population size ( N ), the maximum number of
iterations ( G ), the number of DG units to be installed in the distribution network, limits of
DG placement buses and limits of DG sizes.
Step 2 : The DG placement Buses are positive integers, while the variables that represent the
DG unit size variables are continuous. The placement of DG buses and size of the DGs are
randomly generated and normalized between the maximum and the minimum operating
limits. The placement of DG buses and size of the DGs of j th DG is normalized to b jDG and
j
P DG as given below to satisfy the placement of DG buses and size of the DGs constraints:
+
DG ∈ N , b DG = [b DG ,min
j
b b DG ,max ]
j j
( j
( j
))
b DG = round b DG,min + random( ) × b DG max − b DG,min
(5.18)
j
Where b DG represents DG bus location. Main distribution substation is designated as b jDG = 1
85
P DG ∈ R , P DG = [ 0 P DG ,max ]
j
P DG = P
j j
DG ,min ( (
+ random ( ) × P jDG , max − P jDG ,min )) (5.19)
Select DG placement buses randomly from all the buses and the DG units are installed
in these selected buses. The rating of all the installed DG units, comprise a vector which
represents the grade of different subjects of a particular student and it also represents a
candidate solution for the optimal DG allocation problem. Each set of the feasible solution of
matrix Mi represents a potential solution which is given by
The proposed DG has a pre specified power factor (unity p. f and 0.866 leading p. f ),
so the dimension of the vector is two variables per DG installed (the positive integer bus
number and the DG real power output). Moreover for multiple DG units (n DG) to be
installed in the grid, the vector will have a dimension of (1× 2nDG ) .
Depending upon the population size, initial solution M is created which is given by :
Step 4 : Identify the best solution and assign that solution as the teacher of the class.
Step 5 : Modify the grade of each subject (independent variables of radial distribution
system) of each student based on the teacher knowledge using equation (3.11) in chapter -3.
Step 6 : Update grade of each subject of each student based on the learners’ knowledge by utilizing
the knowledge of some other learner of the same group using equation (3.13) in chapter- 3.
Step 7 : Check whether the independent variables violate the operating limits or not. If any
independent variable is less than the minimum level it is made equal to minimum value and if
it is greater than the maximum level it is made equal to maximum level.
Step 8 : Go to step 2 until the current iteration number reaches the pre specified maximum
iteration number.
86
5.5 DG Placement & Sizing Indices
There are various technical issues that need to be addressed when considering the
presence of distributed generators in distribution systems. To study the effect of DG units on
the performance of power systems, some indices are used as shown in Table 5.1.The details of
indices are described in chapter – 4.
P L = S DG × 100%
DG penetration level
S lo a d
PT , Loss , NO DG − PT , Loss , DG
Active power loss reduction APLR = × 100%
PT , Loss , NO DG
Q T , L o ss , − Q T , L o ss ,
NO DG DG
Reactive power loss reduction RPLR = × 100%
Q T , L o ss , NO DG
VD NO DG − VD
Voltage deviation reduction VDR = DG
× 100%
VD NO DG
VSI NO DG − VSI
Voltage stability index reduction V S IR = DG
× 100%
VSI NO DG
1 n V i − DG
− V i − nom
Voltage deviation index
VDI =
n
∑
i =1
V i − nom
n
∑ V i − DG − V i − NO DG
1 i =1
Voltage profile index VPI = n
× 100%
n ∑ V i − NODG
i =1
n
Qualified load index Q L I = ∑ V i × P Li
i =1
n −1 S
i − DG
Line loading index L L I = m ax
i =1 S i − r a te d
87
A PQ model is considered for DG units in this study system. The DG units are
considered to be working at a specified power factor as mentioned below. In this work, the
two test systems (33 and 69 bus RDS) are considered with existing loading conditions. The
impacts of DG units are studied by comparing the case with and without DG units.
The effectiveness of the proposed method is tested on 33 and 69 bus radial distribution
systems installing two types of three DG units operating at different power factors.
In this section, the results of proposed method for 33 bus RDS for case - I & II are
presented. Table 5.2 shows the before installation objective function values of DG units as
given in chapter - 2.
Table 5.2 : Objective function value of the RDS before DG units installation (Base case)
The detailed performance analysis using proposed (MTLBO) algorithm for 33 bus
RDS after installation of the three DG units are described in Table 5.3 for minimization of the
active power losses. The worst, the best, the arithmetic mean and the Standard deviation of
the objective function for minimization of the active power losses in system 33 bus RDS are
given in Table 5.4. To explain the impact of DG units on the performance of power systems,
some indices are calculated as shown in Table 5.5.
88
Table 5.3 : Performance analysis of the 33 bus RDS after three DG units installation
Table 5.6 gives the results which are compared with other existing techniques for 33
bus RDS respectively. It may be observed from the simulation results that distribution losses
achieved due to installation of DG units in optimal position obtained by different algorithms
are reduced significantly.
33 Bus
Unity p.f. 0.866 p.f. leading
(Case - I) (Case - II)
Best P loss (kW) 73.7061 17.5473
Worst P loss (kW) 76.2018 20.1178
Arithmetic Mean of P loss (kW) 74.9351 19.1021
St. Dev. of P loss (kW) 1.0416 0.9471
33 Bus
Impact index Unity p. f. 0.866 p.f. leading
(Case-I) (Case-II)
DG penetration level (%) 63.16 77.93
Active power loss reduction (%) 64.91 91.65
89
Reactive power loss reduction (%) 64.19 90.62
Voltage deviation reduction (%) 87.14 98.13
Voltage stability index reduction (%) 24.16 28.85
Voltage deviation index 0.020 0.0072
Voltage profile index 0.1111 0.1523
Qualified Load Index 0.0359 0.0363
Line loading Index 0.0389 0.0123
Table 5.6 : Comparison of the proposed method results with previous publications for 33 bus
RDS with three DG units at different power factor
Unity p. f. (Case - I)
90
In addition, it should be pointed out that for 33 bus RDS accomplished with the installation
of three numbers of DG units, the proposed MTLBO algorithm attains active power loss reduction
of 64.91% for unity p.f. and 91.65% for 0.866 p.f. leading which are better than previously
reported methods. Therefore, it can be concluded that MTLBO technique is more efficient than
other techniques in reducing the power loss of 33 bus radial distribution systems.
Obj. Function ( Active power loss) in kW
84
82
80
78
76
74
0 5 10 15 20 25 30 35 40 45 50
Iteration No.
25
24
23
22
21
20
19
18
17
16
0 5 10 15 20 25 30 35 40 45 50
Iteration No.
91
60
Without DG (Base Case)
50 With 3DGs at unity p.f
Active power loss in kW
30
20
10
0
0 5 10 15 20 25 30 35
Branch No.
Fig. 5.4 : Active power loss (kW) before & after DG units installation for 33 bus RDS
Figs. 5.4 depicts active power loss of each bus for case – I & II in 33 bus RDS. It is seen that
the three numbers of DG units injecting active and reactive power at 0.866 p.f. leading(case-II)
results in higher real power loss reduction in the systems as compared to the three DG units
injecting active power only at unity p.f.(case-I) only and without DG (base case).
35
Without DG (Base Case)
30 With 3DGs at unity p.f
Reactive power loss in kVAr
20
15
10
0
0 5 10 15 20 25 30 35
Branch No.
Fig. 5.5 : Reactive power loss (kVAr) before & after DG units installation for 33 bus RDS
Figs. 5.5 gives reactive power loss of each bus for case – I & II in 33 bus RDS. It is seen that
the three numbers of DG units injecting active and reactive power at 0.866 p.f. leading(case-
92
II) results in higher reactive power loss reduction in the systems as compared to the three DG
units injecting active power only at unity p.f.(case-I) and without DG (base case).
1.04
Without DG(Base Case)
1.02 With 3DGs at unity p.f
Voltage magnitude (p.u.)
0.98
0.96
0.94
0.92
0.9
0 5 10 15 20 25 30 35
Bus No.
Fig. 5.6 : Bus voltage level (p. u.) before and after DG units Installation for a 33 bus RDS
Figs. 5.6 gives voltage profile of each bus for case – I & II in 33 bus RDS. The results show
the different voltage levels before installation and after installation of the DG units for
proposed method. Before installation of DG units, voltage level in a 33 bus RDS are low.
After installation of the three DG units, the voltage levels are improved in the proposed
method. It is seen that the three numbers of DG units injecting active and reactive power at
0.866 p.f. leading (case-II) results in higher voltage level in the systems as compared to the
three DG units injecting active power only at unity p.f. (case-I) and without DG (base case).
0.14
Without DG(Base Case)
0.12 With 3DGs at unity p.f
Voltage deviation ( p.u. )
0.08
0.06
0.04
0.02
0
0 5 10 15 20 25 30 35
Bus No.
Fig. 5.7 : Voltage deviation (p. u.) before and after DG units installation for a 33 bus RDS
93
Figs. 5.7 gives bus voltage deviation of each bus for case – I & II in 33 and 69 bus RDS. It is
observed that the three numbers of DG units injecting active and reactive power at 0.866 p.f.
leading (case-II) results in higher voltage deviation reduction in the systems as compared to
the three DG units injecting active power only at unity p.f.(case-I) and without DG (base
case).
1.6
Without DG (Base case)
Voltage stability index (p.u)
1.3
1.2
1.1
1
0 5 10 15 20 25 30 35
Bus No.
Fig. 5.8 : Voltage stability index (p. u.) before and after DG units installation for a 33 bus RDS
Figs. 5.8 gives voltage stability index of each bus for case – I & II in 33 bus RDS. The results
show that the three numbers of DG units injecting active and reactive power at 0.866 p.f.
leading(case-II) results in higher voltage stability index reduction in the systems as compared
to the three DG units injecting active power only at unity p.f.(case-I) and without DG (base
case).
94
In this section, the results of proposed method for 69 bus RDS for case - I & II are
presented. Table 5.7 shows the before installation objective function values of DG units as
given in chapter - 2.
Table 5.7: Objective function value of the RDS before DGs installation (Base case)
The detailed performance analysis using proposed (MTLBO) algorithm for 69 bus
RDS after installation of the three DGs are described in Table 5.8 for minimization of the
active power losses. The worst, the best, the arithmetic mean and the standard deviation of the
objective function for minimization of the active power losses in 69 bus RDS are given in
Table 5.9.To explain the impact of DG units on the performance of power systems, some
indices are calculated as shown in Table 5.10. Table 5.11 gives the results which are
compared with other existing techniques for 69 bus RDS respectively. It may be observed
from the simulation results that distribution losses achieved due to installation of DG units in
optimal position obtained by different algorithms are reduced significantly.
Table 5.8 : Performance analysis of the 69 bus RDS after three DGs installation
95
Table 5.9: Statistics of active power loss in kW for 69 bus RDS
69 Bus
Unity p.f. 0.866 p.f. leading
(Case-I) (Case-II)
Best P loss 71.7734 9.3870
Worst P loss 73.8754 11.6886
Arithmetic Mean of P loss 72.7203 10.5445
St. Dev. of P loss 1.0043 0.9975
69 Bus
Impact index Unity p.f. 0.866 p.f. leading
(Case - I) (Case - II)
DG penetration level (%) 49.58 79.85
Active power loss reduction (%) 68.03 95.82
Reactive power loss reduction (%) 64.78 91.21
Voltage deviation reduction (%) 93.94 99.52
Voltage stability index reduction (%) 25.51 28.75
Voltage deviation index 0.0071 0.0018
Voltage profile index 0.0291 0.0371
Qualified Load Index 0.0372 0.0376
Line loading Index 0.0451 0.0076
In addition, it should be pointed out that for 69 bus system, active power loss reduction
of 68.03% at unity p.f.(case - I ) and 95.82% at 0.866 p.f. leading (case-II) are accomplished
with the installation of three numbers of DG units using the proposed MTLBO algorithm
which is far better than the active power loss reduction of previously reported methods.
Therefore, it is concluded that MTLBO technique is more efficient than other techniques in
reducing the power loss of 69 bus radial distribution system.
96
Table 5.11 : Comparison of the proposed method results with previous publications for 69
bus RDS with three DG units at different power factor
86
84
82
80
78
76
74
72
70
0 5 10 15 20 25 30 35 40 45 50
Iteration No.
97
The convergence characteristics of objective function after the installation of three DG
units obtained by the proposed algorithm for case – I of 69 bus RDS are illustrated in Fig. 5.9.
25
Obj. Function(Active power loss) in kW
20
15
10
5
0 5 10 15 20 25 30 35 40 45 50
Iteration No.
50
Without DG(Base Case)
With 3 DGs at unity p.f.
Active power loss in kW
40
With 3 DGs at 0.866p.f. leading
30
20
10
0
0 10 20 30 40 50 60 70
Branch No.
Fig. 5.11 : Active power loss (kW) before & after DG units installation for 69 bus RDS
Fig. 5.11 depicts active power loss of each bus for case – I & II in 69 bus RDS. It is observed
that the three numbers of DGs injecting active and reactive power at 0.866 p.f. leading (case - II)
98
results in higher real power loss reduction in the systems as compared to the three DG units
injecting active power only at unity p.f.(case - I) and without DG (base case).
20
Without DG(Base Case)
Reactive power loss in kVAr
10
0
0 10 20 30 40 50 60 70
Branch No.
Fig. 5.12 : Reactive power loss ( kVAr ) before & after DG units installation for 69 bus RDS
Fig. 5.12 gives reactive power loss of each bus for case – I & II in 69 bus RDS. It is seen that
the three numbers of DGs injecting active and reactive power at 0.866 p.f. leading (case-II)
results in higher reactive power loss reduction in the systems as compared to the three DGs
injecting active power only at unity p.f. (case-I) and without DG (base case).
1.02
1
Voltage magnitude (p.u.)
0.98
0.96
0.94
Without DG(Base Case)
0.92 With 3 DGs at unity p.f.
With 3 DGs at 0.866p.f. leading
0.9
0 10 20 30 40 50 60 70
Bus No.
Fig. 5.13 : Bus voltage level (p. u.) before and after DG units installation for a 69 bus RDS
Fig. 5.13 depicts each bus voltage in 69 bus RDS. The results show the different voltage levels before
installation and after installation of the DGs for proposed method. Before installation of DGs, voltage
level in 69 bus RDS are low. After installation of the three DGs, the voltage levels are improved in the
proposed method. It is seen that the three numbers of DGs injecting active and reactive power at 0.866
99
p.f. leading (case-II) results in higher voltage level in the systems as compared to the three DGs
injecting active power only at unity p.f. (case-I) and without DG (base case).
0.1
Without DG(Base Case)
With 3DGs at unity p.f
0.08
Voltage deviation (p.u.)
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70
Bus No.
Fig. 5.14 : Bus voltage deviation (p. u.) before and after DG units installation for a 69 bus RDS
Fig. 5.14 gives bus voltage deviation of each bus for case - I & II in 69 bus RDS. It is
observed that the three numbers of DGs injecting active and reactive power at 0.866 p.f.
leading (case-II) results in higher voltage deviation reduction in the systems as compared to
the three DGs injecting active power only at unity p.f.(case-I) and without DG (base case).
1.5
Without DG (Base case)
Voltage stability index (p.u)
1.2
1.1
0.9
0 10 20 30 40 50 60 70
Bus No.
Fig. 5.15 : Voltage stability index (p.u.) before and after DG units installation for a 69 bus RDS
Fig. 5.15 gives voltage stability index of each bus for case - I & II in 69 bus RDS. The results
show that the three numbers of DGs injecting active and reactive power at 0.866 p.f. leading
100
(case-II) results in higher voltage stability index reduction in the systems as compared to the
three DGs injecting active power only at unity p.f. (case-I) and without DG (base case).
5.7 Conclusions
This chapter proposed MTLBO method to solve placement and sizing problems for
DGs simultaneously in 33 and 69 bus radial distribution systems. The proposed method was
implemented for the both systems to minimize the active power losses. The proposed method
stated less objective function values in state of existence DGs. Also this method gives less
active power losses in comparing with the results of other popular optimization techniques.
After DGs installation, the both RDS have shown major improvement in voltage profile and
increase the voltage stability index for the proposed method.
The comparison of the results using the proposed approach to those reported in the
literature confirms its effectiveness and superiority to find remarkable global solutions. The
proposed method introduces the accuracy as well as convergence speed and simplicity. Also
multi-objective studies can be done by the proposed algorithm.
101