1 s2.0 S030626191400227X Main
1 s2.0 S030626191400227X Main
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
h i g h l i g h t s
a r t i c l e i n f o a b s t r a c t
Article history: This paper presents new analytical expressions to efficiently capture the optimal power factor of each
Received 4 January 2014 Distributed Generation (DG) unit for reducing energy losses and enhancing voltage stability over a given
Received in revised form 17 February 2014 planning horizon. These expressions are based on the derivation of a multi-objective index (IMO), which
Accepted 1 March 2014
is formulated as a combination of active and reactive power loss indices. The decision for the optimal
Available online 22 March 2014
location, size and number of DG units is then obtained through a benefit–cost analysis. Here, the total
benefit includes energy sales and additional benefits, namely energy loss reduction, network upgrade
Keywords:
deferral and emission reduction. The total cost is a sum of capital, operation and maintenance costs.
Distributed generation
Emission reduction
The methodology was applied to a 69-bus industrial distribution system. The results showed that the
Loss reduction additional benefits are imperative. Inclusion of these in the analysis would yield faster DG investment
Network upgrade deferral recovery.
Optimal power factor Ó 2014 Elsevier Ltd. All rights reserved.
Voltage stability
http://dx.doi.org/10.1016/j.apenergy.2014.03.005
0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
D.Q. Hung et al. / Applied Energy 124 (2014) 62–72 63
Nomenclature
AEy actual annual emission of the system with DG units NPV net present value
(Ton CO2) OMy annual operation, maintenance and fuel costs ($/year)
ALossy actual annual energy loss of the system with DG units pfDGi power factor of DG unit at bus i
(MW h) PDGi, QDGi, SDGi active, reactive and apparent power sizes of DG
AVSM average voltage stability margin of the system unit, respectively at bus i
B present value benefit over a planning horizon ($) PDi, QDi active and reactive power of load, respectively at bus i
BCR benefit and cost ratio Pi, Qi net active and reactive power injections, respectively at
C present value cost over a planning horizon ($) bus i
CDG capital cost of DG ($/kW) PLDG, QLDG total system active and reactive power losses with DG
CEy cost of each ton of generated CO2 ($/Ton CO2) unit (MW), respectively
CLossy loss value ($/MW h) PL, QL total system active and reactive power losses without
d discount rate DG unit (MW), respectively
EIy emission incentive ($/year) Ry annual energy sales ($/year)
LF load factor or average load level of the system over a TEy annual emission target level of the system without DG
planning horizon (Ton CO2)
LFbase load factor or average load level of the system over the TLossy annual energy loss target level of the system without
base year DG (MW h)
ILP, ILQ active and reactive power loss indices, respectively VSM voltage stability margin
IMO multi-objective index |Vi|, di voltage magnitude and angle, respectively at bus i
IRR internal rate of return Zij ijth element of impedance matrix (Zij = rij + jxij)
LIy loss incentive ($/year) d growth rate of demand a year
N number of buses kmax maximum loading
ND network deferral benefit ($/kW) DAVSM an increase in the average voltage stability margin
Ny planning horizon (years)
allocation for voltage stability at the distribution system level has and power factor of DG for minimizing power losses [26]. In [21],
attracted the interest of some recent research efforts. For instance, three different analytical approaches were presented to determine
DG units are located and sized using different methods: iterative the location, size and power factor of renewable DG (i.e., biomass,
techniques based on Continuous Power Flow (CPF) [8] and a hybrid wind and solar PV) for minimizing energy losses. A dual index-
of model analysis and CPF [28], power stability index-based meth- based analytical approach was proposed to find the location, size
od [29], numerical approach [30,31], simulated annealing algo- and power factor of DG for minimizing power loss and improving
rithm [32] and PSO [33–35]. However, the cost–benefit analyses loadability [38]. Finally, a self-correction algorithm was proposed
of DG planning have been ignored in the works presented above. to specify the size and power factor of PV and battery energy
Furthermore, a few recent studies have indicated that network storage units for minimizing energy losses and enhancing voltage
investment deferral and emission reduction are other attractive stability [39]. The above review shows that a few works have dis-
options for DG planning. For instance, an optimal power flow- cussed the optimal power factor of DG units. However, the size of
based method was successfully developed to place and size DG DG units obtained from the existing studies may not match the
units for postponing network upgrade [4]. An immune-GA method standard sizes available in the market. Furthermore, a comprehen-
was presented for placing and sizing DG units to reduce the total sive benefit–cost study on multiple DG allocation with optimal
emission while minimizing the total cost as a sum of electricity power factor while considering the issues of energy loss and volt-
purchased from the grid, installation, operation and network rein- age stability has not been reported in the literature.
forcement costs [36]. An improved honey bee mating optimization This paper aims at expending the previous preliminary study in
approach was also proposed for locating and sizing DG units to re- [40] where analytical expressions were developed based on a sin-
duce the total emission while minimizing the capital, fuel, opera- gle objective to identify the optimal power factor of each DG unit
tion and maintenance costs, voltage deviation and energy loss for minimizing energy losses. In this paper, analytical expressions
[5]. In addition, an planning framework was also developed for are presented based on a multi-objective index (IMO) to determine
PV integration by reducing the installation, operation and mainte- the optimal power factor for reducing energy losses and enhancing
nance costs and the energy imported from the grid [37]. It is obvi- voltage stability in industrial distribution systems over a given
ous from the above review that numerous methodologies have planning horizon. Here, new analytical expressions are developed
developed for DG allocation in distribution systems with different to efficiently determine the optimal power factor of each DG unit
applications. However, most of them have assumed that DG units with a commercial standard size to ease the computational burden.
operate at a pre-defined power factor. Depending on the nature In this study, it is assumed that DG units are owned and operated
of loads served, DG operation at optimum power factor may have by distribution utilities. To make the work comprehensive, in addi-
positive impacts on system losses, voltage stability, and system tion to the analytical expressions presented to specify the optimal
capacity release. power factor, a benefit–cost analysis is carried out in the paper to
Recently, a few studies have presented DG allocation while con- determine the optimal location, size and number of DG units. The
sidering the optimal power factor, to which the active and reactive total benefit as a sum of energy sales, energy loss reduction, net-
power injections of each DG are optimized simultaneously. For in- work upgrade deferral and emission reduction is compared to
stance, a rule of thumb for DG operation was developed for mini- the total cost including capital, operation and maintenance costs.
mizing power losses [20]. For this rule, it is recommended that The rest of the paper is structured as follows: Section 2 de-
the power factor of DG should be equal to the system load factor. scribes the modeling of loads and DG units. Section 3 presents ac-
A PSO-based method was presented to identify the location, size tive and reactive power loss indices and a combination of both
64 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72
known as the IMO. Methods of assessing the energy loss and volt- voltage regulation, but they are not fast enough to compensate
age stability, and analyzing the benefit and cost with DG units are for transient events [43,44]. As fast response devices, synchronous
also presented in this section. Section 4 presents analytical expres- machines-based DG technologies (e.g., gas turbine engine) are al-
sions to capture the optimal power factor of DG units and a com- lowed to control reactive power for voltage regulation.
putational procedure for DG allocation. Section 5 presents and As reported in [45], gas turbine engines use a turbine spun by
discusses a case study on a 69-bus test distribution system. Finally, the gases of combustion to rotate an electric generator. For DG
the key contributions and conclusions of the work are summarized application, gas turbine engines have smaller sizes than any other
in Section 6. source of rotating power and provide higher reliability than recip-
rocating engines. They also have superior response to load varia-
2. Load and DG modeling tions and excellent steady state frequency regulation when
compared to steam turbines or reciprocating engines. Moreover,
2.1. Load modeling they can operate on a wide range of fuels such as natural gas, waste
gas, methane, propane and diesel. In addition, gas turbine engines
The system considered under the study is assumed to follow the require lower maintenance and produce lower emissions than
industrial load duration curve as shown in Fig. 1, including four reciprocating engines.
discrete load bands (maximum, normal, medium and minimum) For the above reasons, gas turbine engine-based DG units are
that change as the load grows over a planning horizon. The load adopted in this study. As synchronous machines, the DG units
factor or average load level of the system over the base year, LFbase are capable of delivering active power and injecting or absorbing
can be defined as the ratio of the area under load curve to the total reactive power. It is assumed that the units offer a constant energy
duration (four load bands: 8760 h). That means LF base ¼ supply at rated capacity. Given PDGi and QDGi values which corre-
P8760 p:u: loadðtÞ spond to the active and reactive power of DG unit injected at bus
t¼1 8760
, where p.u. load (t) is the demand in p.u. at period t.
i, the power factor of DG unit at bus i (pfDGi) can be expressed as
Assuming the growth rate of demand a year (d), the load factor
follows [21]:
or average load level of the system over a given planning horizon
(Ny), LF can be calculated as: PDGi
pfDGi ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3Þ
1 X8X
Ny
760
p:u: loadðtÞ PDGi þ Q 2DGi
2
LF ¼ ð1 þ dÞy ð1Þ
Ny y¼1 t¼1 8760
The dependence of loads on the voltage and time at period t can 3. Problem formulation
be expressed as [41]:
3.1. Active power loss index
n n
Pi ðtÞ ¼ Poi ðtÞ V i p ðtÞ; Q i ðtÞ ¼ Q oi ðtÞ V i q ðtÞ ð2Þ
The total active power loss (PL) in a system with N buses can be
where Pi and Qi are respectively the active and reactive power injec-
expressed as [46]:
tions at bus i, Poi and Qoi are respectively the active and reactive
loads at bus i at nominal voltage; Vi is the voltage at bus i; X
N X
N
np = 0.18 and nq = 6.0 are respectively the active and reactive indus- PL ¼ ½aij ðPi Pj þ Q i Q j Þ þ bij ðQ i Pj P i Q j Þ ð4Þ
i¼1 j¼1
trial load voltage exponents [41].
where
2.2. DG modeling r ij rij
aij ¼ cosðdi dj Þ; bij ¼ sinðdi dj Þ
V iV j V iV j
Given the fact that most of the DG units are normally designed
to operate at unity power factor under the standard IEEE 1547 [42]. V i \di is complex voltage at bus i; rij + jxij = Zij is the ijth element of
Consequently, inadequacy of reactive power support for voltage impedance matrix [Zbus]; Pi and Pj are respectively active power
regulation may exist in distribution systems, given a high DG pen- injections at buses i and j; and Qi and Qj are respectively the reactive
etration. It is likely that shortage of reactive power support may be power injections at buses i and j.
an immediate concern at the distribution system level in the fu- The active and reactive power injected at bus i where a DG unit
ture. Conventional devices such as switchable capacitors, voltage is installed can be expressed as [20]:
regulators and tap changers are actually employed for automatic
Pi ¼ P DGi PDi ð5Þ
Q i ¼ Q DGi Q Di ð6Þ
p.u.
where PDGi and QDGi are respectively the active and reactive power
1.0 Maximum
injections from DG at bus i; PDi and QDi are respectively the active
load
and reactive power of a load at bus i.
0.8 Substituting Eqs. (5) and (6) into Eq. (4), we obtain the total ac-
Normal tive power loss with DG unit (PLDG) as follows:
work hour
load X
N X
N
0.6 PLDG ¼ aij ððPDGi PDi ÞPj þ ðQ DGi Q Di ÞQ j Þ
Medium i¼1 j¼1
load
0.4 þbij ððQ DGi Q Di ÞPj ðP DGi PDi ÞQ j Þ ð7Þ
Mimimum
load Finally, the active power loss index (ILP) can be defined as Eq.
0 80 2880 5820 8760 (7) divided by Eq. (4) as follows [38]:
Hour
PLDG
ILP ¼ ð8Þ
Fig. 1. Load duration curve. PL
D.Q. Hung et al. / Applied Energy 124 (2014) 62–72 65
The total reactive power loss (QL) in a system with N buses can The maximum DG penetration, which is calculated as the total
be expressed as [46]: capacity of DG units, is limited to less than or equal to a sum of the
total system demand and the total system loss.
X
N X
N
QL ¼ ½cij ðPi Pj þ Q i Q j Þ þ nij ðQ i Pj Pi Q j Þ ð9Þ X
N X
N X
N X
N
i¼1 j¼1 PDGi 6 PDi þ PL ; Q DGi 6 Q Di þ Q L ð13Þ
i¼2 i¼2 i¼2 i¼2
where
where PL and QL are respectively calculated using Eqs. (4) and (9).
x x The voltage at each bus is maintained close to nominal.
cij ¼ ij cosðdi dj Þ; nij ¼ ij sinðdi dj Þ
V iV j V iV j
V min
i 6 V i 6 V max
i ð14Þ
Substituting Eqs. (5) and (6) into Eq. (9), we obtain the total
where and V min V max
are respectively the lower and upper bounds of
i i
reactive power loss with DG unit (QLDG) as follows:
the voltage at bus i, Vi = 1 p.u. (substation
voltage).
X N h
N X The thermal capacity of circuit n Smax n is less than the maxi-
Q LDG ¼ cij ððPDGi PDi ÞPj þ ðQ DGi Q Di ÞQ j Þ mum apparent power transfer (Sn).
i¼1 j¼1
jSn j 6 Smax
n ð15Þ
þnij ððQ DGi Q Di ÞPj ðPDGi PDi ÞQ j Þ ð10Þ
Finally, the reactive power loss index (ILQ) can be defined as Eq.
3.5. Energy loss and voltage stability
(10) divided by Eq. (9) as follows [38]:
IMO ¼ r1 ILP þ r2 ILQ ð12Þ where Dt is 1 h, which is the time duration of period t.
to kmax2 as defined by (17) [28,31,38,39] and the voltage stability power resources and other devices at the operation stage. How-
margin subsequently improves from VSM1 to VSM2. ever, redundancy of reactive power can lead to reverse power flow,
thereby leading to high losses, voltage rise, system instability, etc.
PD ¼ kPo ; Q D ¼ kQ o ð17Þ
Consequently, it becomes necessary to study the optimal power
where Poi and Qoi correspond to the initial active and reactive power factor of DG units to which the active and reactive power injections
demands, respectively. of each DG are optimized simultaneously [20].
3.6.1. Utility’s benefit In practice, the choice of the best DG capacity may follow com-
The present value benefit (B) in $ given to a utility to encourage mercial standard sizes available in the market or be limited by en-
DG connection over a planning horizon from owning and sitting its ergy resource availability. Given such a pre-specified DG capacity,
own DG units can be expressed as follows: the DG power factor can be optimally calculated by adjusting the
active and reactive power sizes at which the IMO as defined by
X
Ny
Ry þ LIy þ EIy X
N
B¼ þ ND P DGi ð18Þ Eq. (12) can reach a minimum level. Using the Lagrange multiplier
y
y¼1 ð1 þ dÞ i¼2 method, the constrained problem defined by Eq. (12) can be math-
ematically converted into an unconstrained one as follows:
where all annual values are discounted at the rate d; Ry is annual
energy sales ($/year) in year; ND is the network deferral benefit LðPDGi ; Q DGi ; ki Þ ¼ r1 ILP þ r2 ILQ þ ki S2DGi P 2DGi Q 2DGi ð21Þ
($/kW); PDGi is the total DG capacity connected at bus i (kW); and
Ny is the planning horizon (years). The loss incentive LIy ($/year) where ki is the Lagrangian multiplier.
can be written as [3,4]: Substituting Eqs. (8) and (11) into Eq. (21), we obtain:
r1 r2
LIy ¼ CLossy ðTLossy ALossy Þ L¼ PLDG þ Q LDG þ ki S2DGi P2DGi Q 2DGi ð22Þ
PL QL
where CLossy is the loss value ($/MW h), ALossy is the actual annual
energy loss of the system with DG units (MW h), and TLossy is the The necessary conditions for the optimization problem, given
target level of the annual energy loss of the system without DG unit by Eq. (22), state that the derivatives with respect to control vari-
(MW h). When DG units are integrated into the grid for primary en- ables PDGi, QDGi and ki become zero.
ergy supply purposes, the environmental benefit as a result of @L r1 @PLDG r2 Q LDG
reducing the usage of fossil fuel energy resources could be obtained. ¼ þ 2ki PDGi ¼ 0 ð23Þ
@PDGi PL @PDGi Q L @PDGi
The emission incentive EIy ($/year) including the emission produced
by the electricity purchased from the grid and DG units can be for- @L r1 @PLDG r2 Q LDG
mulated as [36]: ¼ þ 2ki Q DGi ¼ 0 ð24Þ
@Q DGi PL @Q DGi Q L @Q DGi
EIy ¼ CEy ðTEy AEy Þ
@L
where CEy is the cost of each ton of generated CO2 ($/Ton CO2); AEy ¼ P2DGi þ Q 2DGi S2DGi ¼ 0 ð25Þ
@ki
is the actual annual emission of a system with DG units (Ton CO2);
TEy is the target level of the annual emission of the system without The derivative of Eqs. (7) and (10) with respect to PDGi and QDGi
DG unit (Ton CO2). are given as:
@PLDG XN
3.6.2. Utility’s cost ¼ 2 ½aij Pj bij Q j ¼ 2aii Pi þ 2Ai ð26Þ
@PDGi j¼1
The present value cost (C) in $ incurred by a distribution utility
over a planning horizon can be expressed as [3,4]:
@Q LDG XN
XNy
OMy XN ¼ 2 ½cij Pj nij Q j ¼ 2cii Pi þ 2C i ð27Þ
C¼ y þ C DG PDGi ð19Þ @PDGi j¼1
y¼1 ð1 þ dÞ i¼2
where OMy is the annual operation, maintenance and fuel costs @PLDG XN
¼ 2 ½aij Q j þ bij Pj ¼ 2aii Q i þ 2Bi ð28Þ
($/year) in year y; CDG is the capital cost of DG ($/kW). @Q DGi j¼1
2r1 2r 2 Step 1: Set the apparent power of DG unit (SDGi) and the maxi-
½aii Pi þ Ai þ ½c Pi þ C i 2ki PDGi ¼ 0 ð30Þ
PL Q L ii mum number of buses to connect DG units.
Step 2: Run load flow for the system without DG unit at the
Substituting Eq. (5) into Eq. (30), we obtain:
average load level over the planning horizon (LF) using Eq. (1).
PDi Y i rP1LAi rQ2 CL i Step 3: Find the optimal power factor of DG unit for each bus
PDGi ¼ ð31Þ using Eq. (3). Place this DG unit at each bus and find the IMO
Y i ki
for each case using Eq. (12).
where Step 4: Locate the optimal bus for DG at which the IMO is min-
imum with the corresponding optimal size and power factor at
r1 aii r2 cii that bus.
Yi ¼ þ
PL QL Step 5: Run multiyear multi-period load flow with the DG size
obtained in Step 4 over the planning horizon. Calculate the
Similarly, substituting Eqs. (28) and (29) into Eq. (24), we
energy loss and its corresponding BCR using Eqs. (16) and
obtain:
(20), respectively.
2r1 2r 2 Step 6: Repeat Steps 3–5 until ‘‘the maximum number of buses is
½aii Q i þ Bi þ ½c Q þ Di 2ki Q DGi ¼ 0 ð32Þ reached’’. These buses are defined as ‘‘a set of candidate buses’’.
PL Q L ii i
Continue to connect DG units to ‘‘these candidate buses’’ by
Substituting Eq. (6) into Eq. (32), we obtain Eq. (33), where Yi is repeating Steps 3–5.
given in Eq. (31). Step 7: Stop if any of the violations of the constraints (Sec-
tion 3.4) occurs or the last iteration BCR is smaller than the pre-
Q Di Y i rP1LBi rQ2 DL i
Q DGi ¼ ð33Þ vious iteration one. Obtain the results of the previous iteration.
Y i ki
Substituting Eqs. (31) and (33) into Eq. (25), we obtain:
5. Case study
1
Y i ki ¼
SDGi
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 5.1. Test systems
r 1 Ai r 2 C i 2 r1 B i r2 D i 2
PDi Y i þ Q Di Y i The proposed methodology was applied to an 11 kV 69-bus ra-
PL QL PL QL
dial distribution system with four feeders that are fed by a 6 MVA
ð34Þ
33/11 kV transformer, as depicted in Fig. 3 [51]. Its complete data
Substituting Eq. (34) into Eqs. (31) and (33), we obtain: can be found in [52]. The total active and reactive power of the
system at the average load level defined by Eq. (1) is 3.35 MW
PDi Y i rP1LAi rQ2 CL i SDGi and 2.30 MVAr, respectively. The proposed methodology was
PDGi ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2 ð35Þ simulated in MATLAB environment.
PDi Y i rP1LAi rQ2 CL i þ Q Di Y i rP1LBi rQ2 DL i
Q Di Y i rP1LBi rQ2 DL i SDGi
Q DGi ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2 ð36Þ
PDi Y i rP1LAi rQ2 CL i þ Q Di Y i rP1LBi rQ2 DL i
Table 2
Location, size and power factor of DG units.
Scenarios DG location DG size (MVA) DG power factor (lag.) Total DG size (MVA) Permissible constraints?
1 DG 62 0.8 0.87 0.8 No
2 DGs 62 0.8 0.87 1.6 No
35 0.8 0.89
3 DGs 62 0.8 0.87 2.4 Yes
35 0.8 0.89
25 0.8 0.89
4 DGs 62 0.8 0.87 3.2 Yes
35 0.8 0.89
25 0.8 0.89
4 0.8 0.85
5 DGs 62 0.8 0.87 4.0 Yes
35 0.8 0.89
25 0.8 0.89
4 0.8 0.85
39 0.8 0.82
0.6
reactive power loss of the system, significantly reduces for the sce-
0.4 nario with four or five DG units when compared to three DG units,
as shown in Fig. 5. In addition, it is observed from Fig. 7 that the
0.2
VSM values with and without DG units reduce with respect to a
0.0 yearly demand growth of 3%. Hence, the lowest VSM values are
1 2 3 4 5 found in the year 15.
Number of DGs Table 3 shows a summary of the results of energy losses with-
out and with DG units for each scenario over the planning horizon
Fig. 5. Indices (ILP, ILQ and IMO) for the system with various numbers of DG units of 15 years. The energy savings due to loss reduction is beneficial. A
over planning horizon. maximum energy savings is achieved for the scenario with five DG
units when compared to three and four DG units. Table 3 also pre-
sents a summary of the results of voltage stability with and with-
1500
No DG out DG units over the planning horizon. The average voltage
3 DGs stability margin of the system (AVSM) is calculated as a sum of
Energy loss (MWh)
1200
4 DGs the VSM values of all years divided by the total planning horizon.
5 DGs
900 An increase in the average voltage stability margin (DAVSM) is ob-
served after 3–5 DG units are installed in the system. It is observed
600 from Table 3 that the VSM for each scenario increases with respect
to an increase in the number of DG units installed in the system as
300
well as a reduction in the overall energy loss of the system.
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
5.3.3. Benefit and cost analysis
Year
Table 4 presents the results of the total benefit and cost for
Fig. 6. Losses of the system with and without DG units over planning horizon. three scenarios (i.e., 3, 4 and 5 DG units) without and with addi-
tional benefits over the planning horizon of 15 years. The addi-
6.0 tional benefit includes the loss incentive (LI), emission incentive
Voltage stability margin
No DG 3 DGs
(EI) and network upgrade deferral (ND). The total benefit (B) is a
4 DGs 5 DGs
4.5 sum of all the additional benefits and the energy sales (R). The total
cost (C) is a sum of the operation, maintenance and fuel cost (OM)
3.0 and the DG capacity cost (CDG RPDGi). Table 5 shows a comparison
of the results for three different scenarios without and with addi-
tional benefits over the planning horizon. The results include the
1.5
benefit–cost ratio (BCR), net present value (NPV = B C), payback
period, and internal rate of return (IRR).
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For exclusion of the additional benefits, it is observed from Ta-
Year ble 5 that the BCR is the same for all the scenarios at 1.288. The best
solution is five DG units as the NPV is highest at k$4162. This solu-
Fig. 7. Voltage stability margin curves for all scenarios over planning horizon. tion generates an IRR of 16.48% and a payback period of 5.6 years.
For inclusion of the additional benefits, it is seen from Table 4 that
the energy sales (R) accounts for around 89–90% of the B, leaving
the VSM increases to 4.5918 from the base case value of 2.8681 the total additional benefit (LI, EI and ND) at roughly 10–11%. It
(without DG units). A similar trend has been found for years is obvious that the R has a significant impact on the BCR when
70 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72
Table 3
Energy loss and voltage stability losses over planning horizon.
Scenarios Energy loss (GW h/15 years) Energy savings (GW h/15 years) Voltage stability
AVSM DAVSM
Base case 14.91 2.1667
3 DGs 5.97 8.94 3.5745 1.4078
4 DGs 4.46 10.45 3.5758 1.4091
5 DGs 4.35 10.56 3.6855 1.5188
Table 4
Analysis of the present value benefit and cost for different scenarios over planning horizon.
Table 5
Comparison of different scenarios over planning horizon.
4 DGs 5 DGs
installed. This factor has an impact on the BCR. As shown in Table 5, 150
Transformer upgrade required
the BCR slightly drops from 1.450 to 1.438 when the number of DG
units is increased from three to five, respectively. The best solution 100
is three DG units with the highest BCR of 1.450. This solution gen- Trans. rating
erates an NPV of k$3993, an IRR of 32.65% and a payback period of
50
2.80 years. In general, in the absence of the additional benefits, the
optimal number of DG units is five, while in the presence of the Trans. upgrade deferred
additional benefits, this figure is three. Inclusion of the additional 0
benefits in the study can lead to faster investment recovery with 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Year
Fig. 9. Percentage ratio of the total demand plus loss to the thermal transformer
7.5 limit over planning horizon.
3 DGs
6.0 4 DGs
NPV (million $)
a higher BCR, a higher IRR and a shorter payback period when com-
5 DGs
4.5 pared to exclusion of the additional benefits. In addition, it is
shown from Table 5 that the NPV value will increase further with
3.0 increasing DG units installed. However, the system cannot accom-
modate more than five DG units due to the violation of the maxi-
1.5 mum DG penetration constraint as defined by Eq. (13).
Fig. 8 shows an increase in the NPV with the corresponding
0.0
10 20 30 40 50 60 70 80 90 100 average DG utilization factors over the planning horizon of
15 years for three scenarios (i.e., 3, 4 and 5 units) with the addi-
Average utilization of DG (%)
tional benefits. It is observed from the figure that given a certain
Fig. 8. NPV at various average DG utilization factors for different scenerios over average DG utilization factor in the range of 0–100%, the NPV is
planning horizon. highest for the scenario with five DG units, while this figure is
D.Q. Hung et al. / Applied Energy 124 (2014) 62–72 71
lowest for the scenario with three DG units. In addition, for each [10] Borges CLT, Falcão DM. Optimal distributed generation allocation for
reliability, losses, and voltage improvement. Int J Elect Power Energy Syst
scenario, the NPV is maximum when the utilization factor is
2006;28(6):413–20.
100% as estimated in Table 5. However, in practice, this factor [11] Abdullah MA, Agalgaonkar AP, Muttaqi KM. Assessment of energy supply and
may be less than 100% due to interruption for maintenance and continuity of service in distribution network with renewable distributed
others. Consequently, the respective NPV will be reduced as shown generation. Appl Energy 2014;113:1015–26.
[12] Zhang P, Li W, Li S, Wang Y, Xiao W. Reliability assessment of photovoltaic
in Fig. 8. power systems: review of current status and future perspectives. Appl Energy
Fig. 9 shows the percentage ratio of the total demand plus total 2013;104:822–33.
loss to the thermal limit (6 MVA) of the transformer over the plan- [13] Naderi S, Pouresmaeil E, Gao WD. The frequency-independent control method
for distributed generation systems. Appl Energy 2012;96:272–80.
ning horizon with a demand growth of 3%. For the scenarios with [14] Mashhour E, Moghaddas-Tafreshi SM. Bidding strategy of virtual power plant
DG units, the curves are plotted at the average DG utilization factor for participating in energy and spinning reserve markets – Part I: problem
of 100%. Obviously, without DG connection, an investment would formulation. IEEE Trans Power Syst 2011;26(2):949–56.
[15] Rueda-Medina AC, Padilha-Feltrin A. Distributed generators as providers of
be needed to add a new transformer before year 4. However, the reactive power support – a market approach. IEEE Trans Power Syst
selected three-DG scenario can defer in upgrading this current 2013;28(1):490–502.
transformer (6 MVA) to 11 years. A higher deferral is achieved for [16] Yuen C, Oudalov A, Timbus A. The provision of frequency control reserves from
multiple microgrids. IEEE Trans Ind Electron 2011;58(1):173–83.
the scenarios with four or five DG units. [17] Katiraei F, Aguero JR. Solar PV integration challenges. IEEE Power Energy Mag
2011;9(3):62–71.
[18] Eftekharnejad S, Vittal V, Heydt GT, Keel B, Loehr J. Impact of increased
6. Conclusions penetration of photovoltaic generation on power systems. IEEE Trans Power
Syst 2013;28(2):893–901.
[19] Wang C, Nehrir MH. Analytical approaches for optimal placement of
This paper has developed an investment planning framework distributed generation sources in power systems. IEEE Trans Power Syst
for integrating multiple Distributed Generation (DG) units in 2004;19(4):2068–76.
industrial distribution systems where the DG units are assumed [20] Hung DQ, Mithulananthan N, Bansal RC. Analytical expressions for DG
allocation in primary distribution networks. IEEE Trans Energy Convers
to be owned and operated by utilities. In this framework, analytical 2010;25(3):814–20.
expressions are proposed to efficiently identify the optimal power [21] Hung DQ, Mithulananthan N, Bansal RC. Analytical strategies for renewable
factor of DG units for minimizing energy losses and enhancing distributed generation integration considering energy loss minimization. Appl
Energy 2013;105:75–85.
voltage stability. The decision for the optimal location, size and [22] Ochoa LF, Harrison GP. Minimizing energy losses: optimal accommodation and
number of DG units is achieved through a benefit–cost analysis. smart operation of renewable distributed generation. IEEE Trans Power Syst
The total benefit includes energy sales and three additional bene- 2011;26(1):198–205.
[23] Atwa YM, El-Saadany EF. Probabilistic approach for optimal allocation of wind-
fits including loss reduction, network upgrade deferral and emis-
based distributed generation in distribution systems. IET Renew Power Gener
sion reduction. The total cost is a sum of capital, operation and 2011;5(1):79–88.
maintenance costs. The results obtained on a 69-bus test distribu- [24] Atwa YM, El-Saadany EF, Salama MMA, Seethapathy R. Optimal renewable
resources mix for distribution system energy loss minimization. IEEE Trans
tion system indicated that the additional benefits, particularly the
Power Syst 2010;25(1):360–70.
loss incentive have a significant impact on decision-making about [25] Singh D, Verma KS. Multiobjective optimization for DG planning with load
the total number of DG units or the amount of DG capacity in- models. IEEE Trans Power Syst 2009;24(1):427–36.
stalled. The additional benefits together accounted for 10–11% of [26] Kansal S, Kumar V, Tyagi B. Optimal placement of different type of DG sources
in distribution networks. Int J Elect Power Energy Syst 2013;53:752–60.
the total benefit when compared to the energy sales of 89–90%. [27] Abu-Mouti FS, El-Hawary ME. Optimal distributed generation allocation and
Inclusion of these benefits in the study can lead to faster invest- sizing in distribution systems via artificial bee colony algorithm. IEEE Trans
ment recovery with a high benefit–cost ratio, a high internal rate Power Del 2011;26(4):2090–101.
[28] Ettehadi M, Ghasemi H, Vaez-Zadeh S. Voltage stability-based DG placement
of return and a short payback period. in distribution networks. IEEE Trans Power Del 2013;28(1):171–8.
When DG units are owned by DG developers, the additional [29] Aman MM, Jasmon GB, Mokhlis H, Bakar AHA. Optimal placement and sizing of
benefits should be shared between the distribution utility and a DG based on a new power stability index and line losses. Int J Elect Power
Energy Syst 2012;43(1):1296–304.
DG developer to encourage DG connection. In this situation, the [30] Esmaili M. Placement of minimum distributed generation units observing
proposed methodology could be used as guidance for the utility power losses and voltage stability with network constraints. IET Gener Transm
on how to plan and operate DG units to obtain the additional Distrib 2013;7(8):813–21.
[31] Al Abri RS RS, El-Saadany EF, Atwa YM. Optimal placement and sizing method
benefits.
to improve the voltage stability margin in a distribution system using
distributed generation. IEEE Trans Power Syst 2013;28(1):326–34.
[32] Injeti SK, Prema Kumar N. A novel approach to identify optimal access point
References and capacity of multiple DGs in a small, medium and large scale radial
distribution systems. Int J Elect Power Energy Syst 2013;45(1):142–51.
[1] El-Khattam W, Bhattacharya K, Hegazy Y, Salama MMA. Optimal investment [33] Aman MM, Jasmon GB, Bakar AHA, Mokhlis H. A new approach for optimum
planning for distributed generation in a competitive electricity market. IEEE DG placement and sizing based on voltage stability maximization and
Trans Power Syst 2004;19(3):1674–84. minimization of power losses. Energy Convers Manage 2013;70:202–10.
[2] Viral R, Khatod DK. Optimal planning of distributed generation systems in [34] Kayal P, Chanda CK. Placement of wind and solar based DGs in distribution
distribution system: a review. Renew Sustain Energy Rev 2012;16(7):5146–65. system for power loss minimization and voltage stability improvement. Int J
[3] Siano P, Ochoa LF, Harrison GP, Piccolo A. Assessing the strategic benefits of Elect Power Energy Syst 2013;53:795–809.
distributed generation ownership for DNOs. IET Gener Transm Distrib [35] Hien NC, Mithulananthan N, Bansal RC. Location and sizing of distribution
2009;3(3):225–36. generation units for loadability enhancement in primary feeder. IEEE Syst J
[4] Piccolo A, Siano P. Evaluating the impact of network investment deferral on 2013;7(4):797–806.
distributed generation expansion. IEEE Trans Power Syst 2009;24(3):1559–67. [36] Soroudi A, Ehsan M, Zareipour H. A practical eco-environmental distribution
[5] Niknam T, Taheri SI, Aghaei J, Tabatabaei S, Nayeripour M. A modified honey network planning model including fuel cells and non-renewable distributed
bee mating optimization algorithm for multiobjective placement of renewable energy resources. Renew Energy 2011;36(1):179–88.
energy resources. Appl Energy 2011;88(12):4817–30. [37] Kucuksari S, Khaleghi AM, Hamidi M, Zhang Y, Szidarovszky F, Bayraksan G,
[6] Niknam T. A new HBMO algorithm for multiobjective daily Volt/Var control in et al. An integrated GIS, optimization and simulation framework for optimal
distribution systems considering distributed generators. Appl Energy PV size and location in campus area environments. Appl Energy
2011;88(3):778–88. 2014;113:1601–13.
[7] Martinez-Rojas M, Sumper A, Gomis-Bellmunt O, Sudrià-Andreu A. Reactive [38] Hung DQ, Mithulananthan N. Loss reduction and loadability enhancement
power dispatch in wind farms using particle swarm optimization technique with DG: a dual-index analytical approach. Appl Energy 2013;115:233–41.
and feasible solutions search. Appl Energy 2011;88(12):4678–86. [39] Hung DQ, Mithulananthan N, Bansal RC. Integration of PV and BES units in
[8] Hedayati H, Nabaviniaki SA, Akbarimajd A. A method for placement of DG units commercial distribution systems considering energy loss and voltage stability.
in distribution networks. IEEE Trans Power Del 2008;23(3):1620–8. Appl Energy 2014;113:1162–70.
[9] Brown RE, Freeman LAA. Analyzing the reliability impact of distributed [40] Hung DQ, Mithulananthan N. Assessing the impact of loss reduction on
generation. In: Proc IEEE Power Eng Soc Summer Meet 2001;2:1013–8. distributed generation investment decisions. In: Proc in Australasian
72 D.Q. Hung et al. / Applied Energy 124 (2014) 62–72
universities power engineering conference (AUPEC) Hobart, TAS, Australia, 29 [47] Ochoa LF, Padilha-Feltrin A, Harrison GP. Evaluating distributed generation
September–3 October; 2013. impacts with a multiobjective index. IEEE Trans Power Del 2006;21(3):1452–8.
[41] Asper SG, Nwankpa CO, Bradish RW, Chiang H-D, Concordia C, Staron JV, et al. [48] Ochoa LF, Padilha-Feltrin A, Harrison GP. Evaluating distributed time-varying
Bibliography on load models for power flow and dynamic performance generation through a multiobjective index’’. IEEE Trans Power Del
simulation. IEEE Trans Power Syst 1995;10(1):523–38. 2008;23(2):1132–8.
[42] IEEE 1547 standard for interconnecting distributed resources with electric [49] El-Zonkoly AM. Optimal placement of multi-distributed generation units
power systems; October 2003. including different load models using particle swarm optimisation. IET Gener
[43] Yeh HG, Gayme DF, Low SH. Adaptive VAR control for distribution circuits with Transm Distrib 2011;5(7):760–71.
photovoltaic generators. IEEE Trans Power Syst 2012;27(3):1656–63. [50] Canizares CA, Alvarado FL. Point of collapse and continuation methods for
[44] Turitsyn K, Sulc P, Backhaus S, Chertkov M. Options for control of reactive large AC/DC systems. IEEE Trans Power Syst 1993;8(1):1–8.
power by distributed photovoltaic generators. Proc IEEE 2011;99(6):1063–73. [51] Das D. A fuzzy multiobjective approach for network reconfiguration of
[45] Willis HL, Scott WG. Distributed power generation: planning and distribution systems. IEEE Trans Power Del 2006;21(1):202–9.
evaluation. New York: Marcel Dekker Inc.; 2000. [52] Harrison GP, Piccolo A, Siano P, Wallace AR. Hybrid GA and OPF evaluation of
[46] Elgerd IO. Electric energy system theory: an introduction. New York: McGraw- network capacity for distributed generation connections. Electr Power Syst Res
Hill Inc.; 1971. 2008;78(3):392–8.