You are on page 1of 19

www.ietdl.

org

Published in IET Generation, Transmission & Distribution


Received on 18th January 2009
Revised on 3rd August 2009
doi: 10.1049/iet-gtd.2009.0026

Special Issue selected papers on Electricity Markets:


Analysis & Operations

ISSN 1751-8687

Optimal distributed generation location using


mixed integer non-linear programming in
hybrid electricity markets
A. Kumar1 W. Gao2
1
Department of Electrical Engineering, National Institute of Technology, Kurukshetra, India
2
Department of Electrical and Computer Engineering, Centre for Energy Systems Research, Tennessee Tech University, USA
E-mail: wgao@tntech.edu

Abstract: This study presents mixed integer non-linear programming (MINLP) approach for determining optimal
location and number of distributed generators in hybrid electricity market. For optimal location of distributed
generation (DG), rst the most appropriate zone has been identied based on real power nodal price and real
power loss sensitivity index as an economic and operational criterion. After identifying the suitable zone,
mixed integer non-linear programming approach has been applied to locate optimal place and number of
distributed generators in the obtained zone. The non-linear optimisation approach consists of minimisation of
total fuel cost of conventional and DG sources as well as minimisation of line losses in the network. The
pattern of nodal real and reactive power prices, line loss reduction and fuel cost saving has been obtained.
The results have also been obtained for pool electricity market model for comparison. The impact of demand
variation on the results has also been obtained for both the market models. The proposed MINLP-based
optimisation approach has been applied for IEEE 24 bus reliability test system.

Nomenclature Nl number of lines in the system


Ng set of generators Pmin max
gi , Pgi minimum and maximum real power
generation limits
Pg active power pool generator-i
Qmin max
gi , Qgi minimum and maximum reactive power
Ci fuel cost of pool generator-i
generation limits
agi , bgi , cgi cost coefcients in $/h, $/MWh, $/MWh2
Vimin, Vimax upper and lower voltage magnitude limits
Pi real power injection at bus-i
dmin
i , dmax
i upper and lower voltage angle limits
Qi reactive power injection at bus-i
Sij , Sij  Sijmax line ow limit
Pgi , Qgi real and reactive power generation at bus-i
lpi, lqi , qpl, qql , corresponding Lagrange multipliers of
Pdi , Qdi real and reactive power demand at bus-i mmax
i , mmin
i , hi
max
, the respective equality and inequality
PGT, QGT total real and reactive power generation hi , gi , gmin
min max
i , constraints
PDT, QDT total real and reactive power demand zmax
i , zmin
i , cl

PLT, QLT total real and reactive power loss Pgp and Pgb vector of pool and bilateral generation
with sufx gp and gb for pool and
Vi voltage magnitude at bus-i;
bilateral generation
di voltage angle at bus-i
Pdp and Pdb vector of pool and bilateral demand with
Yij Gij Bij (i j)th element of Y-bus matrix sufx dp and db for pool and bilateral
Nb number of buses in the system demand

IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298 281
doi: 10.1049/iet-gtd.2009.0026 & The Institution of Engineering and Technology 2010
www.ietdl.org

Pfp and Pfb vector of line ows because of the pool environment policies inuencing fuel choices for power
and bilateral transactions with sufx fp generation have developed the growing interest in the
and fb for real power ows for pool and distributed generation (DG) as an important option of
bilateral transactions energy production in the near future. With electricity
ACDF ac distribution factors market undergoing tremendous transformation, more price
volatility in the market, ageing infrastructure and changing
Tij and Tij0 secure and proposed bilateral transactions
regulatory environments are demanding users and electric
Tmax
ij maximum transaction amount utilities to harness benets of DG [1, 2]. DG is expected to
bij weight factor indicating the importance of a play key role in future competitive markets because of their
a particular transaction economic viability and based on the study of Electric Power
aij , bij , gij , jij loss coefcients Research Institute (EPRI) and Natural Gas Foundation,
30% of power generation share will be of DG [3, 4]. Many
Rij and Xij real and imaginary parts of ijth element of
denitions of DG have appeared in the literature based on
the bus impedance matrix [Zbus]
their size, technologies, location, power delivery area and
x state vector of variables V, d operational constraints with their economical and
u control parameters, Pgi , Qgi , PDGi , QDGi , operational benets [5 7]. The DG technologies may
Pgb , Pgp comprise small gas turbines, micro-turbines, fuel cells, wind
p xed parameters Pdi , Pdb , Pdp , Qd, Tij and solar energy. DG can be connected in an isolated or an
int integrated way in the power system network and issues
j an integer variable with values f0,1g
relating to policy of integrating DGs into power system
r is the demand variation factor planning and their impacts on steady-state power system
NDG set of distributed generators operation, contingency analysis, protection coordination as
Nl total number of lines in the network well as dynamic behaviour were discussed in [7, 8].
PDGi active power of distributed generator
With increasing share of DG, the planning of the system
connected at bus-i
in the presence of DGs will require the assessment of
ai , bi , ci cost coefcients for conventional several factors such as the number and the capacity of units,
generators [24] best possible location in the network and impact of DG on
aDGi , bDGi , cDGi cost coefcients for distributed generators the system operation characteristics such as system losses,
[22] voltage prole, stability and reliability issues [9].
l0pil, l0pjl base case nodal price for real power at
bus-i and bus-j for line l connected The optimal placement and sizing of DGs in the
between bus-i and bus-j distribution network based on different objectives have been
PLl (Pijl Pjil) real power loss in the line l connected reported in the literature. A Lagrangian-based approach was
between bus-i and bus-j is the sum of proposed to determine optimal location of DGs considering
line ows from bus-i to bus-j and bus-j economic as well as stability limits in [10]. Application of
to bus-i tabu search for optimal placement of distributed generators
with an objective of minimising losses was proposed in [11].
(Pijl , Pjil ) line ows from bus-i to bus-j and bus-j A genetic algorithm (GA)-based approach for DG
to bus-i allocation minimising power losses in a network was
PDGi and QDGi real and reactive generation for distributed proposed in [12]. Optimal sizing and siting decisions for
generators DG capacity planning using heuristic approach was
Pmin max
DGi, PDGi minimum and maximum generation proposed in [13]. A new integrated model for distributed
limits on distributed generators system planning with an objective of minimising investment
Qmin max costs, operating costs and payments for compensation of
DGi, QDGi minimum and maximum generation
limits on distributed generators losses was proposed in [14]. Optimal allocation of
embedded generation in distribution network using linear
Nmax
DG maximum number of distributed programming approach including short-circuit constraints of
generator to be located in the network network was proposed in [15]. A multi-objective
optimisation approach using evolutionary algorithm with
an objective of minimising cost of energy losses, network
upgrading and service interruptions for sizing and siting
1 Introduction of DG in distribution systems has been presented in [16].
A GA-based methodology for optimal DG allocation and
With liberalisation of electricity markets leading to stronger sizing in distribution systems, in order to minimise network
competition, technological developments in the eld of losses and to guarantee high level of reliability and voltage
generation, increasing importance of security of power prole was proposed in [17]. An analytical expression based
supply and diversication of energy sources, and recent on real power loss sensitivity to calculate optimal DG size

282 IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-gtd.2009.0026
www.ietdl.org

and optimal location of DG minimising power losses in a number of distributed generators in the identied zone
distribution network was proposed in [18]. A hybrid GA- based on mixed integer non-linear programming-based
OPF approach was proposed for nding optimal location for approach and (iii) to nd the impact of demand variation.
connecting a predened numbers of DGs in a distribution The cost of distributed generators has also been considered
network in [19]. A fuzzy multi-objective optimisation along with fuel costs of conventional power plants with
considering voltage drop, system losses, short-circuit their cost coefcients [23]. Nodal price at each bus and real
capacity and system operation cost for DG allocation was power loss sensitivity factors have been determined and
proposed for a small distribution system in [20]. utilised to nd an appropriate zone for optimal placement
of distributed generators. Subsequently, to nd optimal
A theoretical insight to a competitive market integration number and location of DG in the obtained zone, mixed
mechanism for DG in a pool-based system was proposed integer non-linear programming-based approach has been
in [21]. The mechanism encompasses both energy and applied. Pattern of nodal prices, fuel cost and line loss
capacity payment procedures in the wholesale market with reduction have been determined. The results have been
DG units located in distribution level. Authors introduced obtained for a pool as well as hybrid electricity market
a framework for implementing optimal DG capacity model. The impact of the demand variation on nodal price,
investment as an attractive option in distribution system fuel costs and real power loss has also been studied for both
planning in deregulated electricity market [22]. A social the market models. The proposed methodology has been
welfare maximisation and prot maximisation-based applied for IEEE 24 bus reliability test system [27].
optimal power ow methodologies were proposed for
optimal location of DGs in pool-based electricity markets
[23]. A multi-objective optimisation approach considering
losses reduction and voltage prole improvement for DG 2 Selection of zone for optimal
allocation using GA was proposed in [24]. placement of DG
Based on the literature review, the proposed approaches Optimal location of DG penetration in the network has
for DG location has been studied for radial network and become a crucial issue because of the expectation of more
are applicable to pool-based market models. Several market DG utilisation in the coming future. DG optimal
structures and transactions exist to achieve a competitive placement can provide both economical and operational
electricity environment. Three basic models based on the advantages. In a wide area power system network, the
types of transactions are: pool model, bilateral contract identication of most suitable area for DG penetration is
model and hybrid market model [25]. A pool is dened as very important. Thus, both economic and operational
a centralised market place that clears the market for the aspects should be considered for nding most appropriate
buyers and sellers. Electric power sellers/buyers submit bids zone. Nodal price is an important indicator of price of unit
to the pool for the amount of power that they are willing to MW injection at each node and congestion in the
trade in the market. In a deregulated electricity market transmission network. Line loss sensitivity provides
structure, under competition and open access, the different information about pattern of losses in the transmission
transactions may take place directly between buyers and network with the unit power injection at each node. To
sellers. These transactions may be bilateral, multilateral and obtain the most appropriate zone, nodal price variation at
ancillary services transactions [26]. The hybrid model each bus and line loss sensitivity has been utilised as
combines various features of the previous two models. economical and operational criteria.

In hybrid electricity market model, large number of Once the most appropriate zone has been identied based
transactions is expected between buyers and sellers for more on nodal price and line loss sensitivity, mixed integer non-
exible and economic market operation. These transactions linear programming approach can be utilised to nd the
need to be evaluated ahead of their scheduling time to optimal location and the number for DG in that
check their feasibility with respect to system operating appropriate zone.
conditions. Infeasible transactions can alter the economic
schedule, cause congestion and threaten system security and
stability of the network. Therefore the issue of DG optimal 2.1 A real power nodal price
location in hybrid electricity market model needs to be
addressed. The power system has grown to a very vast
determination (pool market model)
network and for better planning studies of DG penetration, The price for real and reactive powers at bus-i is the
the identication of most suitable area and optimal location Lagrangian multiplier value of equality constraints, that is,
of DG in that area is required. the real and reactive power ow equations of bus-i. The
Lagrangian multipler values are calculated by solving the
The main contribution of the present work is: (i) to nd rst-order necessary condition of Lagrangian, partial
suitable zone for DG placement based on economic and derivatives of the Lagrangian with respect to every variable
operational criteria, (ii) to determine optimal location and concerned.

IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298 283
doi: 10.1049/iet-gtd.2009.0026 & The Institution of Engineering and Technology 2010
www.ietdl.org

Mathematically, the normal dispatch problem with a fuel (b) Reactive power generation limit: This includes the upper
cost minimisation can be written as and lower reactive power generation limits of generators and
other reactive sources at bus-i
X X
Min Ci (Pgi ) (agi bgi Pgi cgi Pgi2 ) (1)
i[Ng i[Ng Qmin max
gi  Qgi  Qgi , i 1, 2, . . . , Nq (13)

1. Equality constraints (c) Voltage limit: This includes the upper and lower voltage
magnitude limits at bus-i
(a) Power ow equations corresponding to both real and
reactive power balance equations as load ow constraints Vimin  Vi  Vimax , i 1, 2, . . . , Nb (14)
can be written for all the buses as
XNb
(d) Phase angle limit: This includes the upper and lower
Pi Pgi  Pdi Vi Vj [Gij cos(di  dj )
angle limits at bus-i
j1

Bij sin(di  dj )] 8i 1, 2, . . . , Nb (2) dmin


i  di  dmax
i , i 1, 2, . . . , Nb (15)
Nb
X
Qi Qgi  Qdi Vi Vj [Gij sin(di  dj ) (e) Line ow limits: These constraints represent maximum
j1 power ow in a transmission line and are based on thermal
and stability considerations. Sl is MVA line ows from bus-
Bij cos(di  dj )] 8 i 1, 2, . . . , Nb (3)
i to bus-j. The line ow limit can be written as
(b) System real and reactive power balance equations: dene
total power generation real and reactive (PGT, QGT), total Sl  Slmax (16)
power demand real and reactive (PDT, QDT) and total real
and reactive power loss (PLT, QLT), the system real and The Langrangian function for the nodal price determination
reactive power balance equations can be written as can be written as a function of Pi and Qi as

PGT  P LT  PDT 0 (4) "


X X Nb
X
QGT  QLT  QDT 0 (5) L(Pi , Qi ) Ci (Pi ) (lpi ) Pi  Vi Vj
i[Ng i[Nb j1
" ##
Using general loss formula, total real and reactive power loss Nb
X
can be expressed as [28]  Gij cos(di  dj ) Bij sin(di  dj ) (lqi )
i1
" #
Nb X
X Nb Nb
X h i
PLT [aij (Pi Pj Qi Qj ) bij (Qi Pj  Qj Pi )] (6)  Qi  Vi Vj Gij sin(di  dj )  Bij cos(di  dj )
i1 j1 j1

Nb X
X Nb
qpl (PGT  PDT  PLT ) qql (QGT  QDT  QLT )
QLT [gij (Pi Pj Qi Qj ) jij (Qi Pj  Qj Pi )] (7)
Ng Ng
i1 j1 X X
Rij mmax max
i (Pi  Pi ) mmin min
i (Pi  Pi )
aij cos(di  dj ) (8) i1 i1
jVi Vj j
Nq Nq
X X
Rij hmax max
i (Qi  Qi ) hmin min
i (Qi  Qi )
bij sin(di  dj ) (9) i1 i1
jVi Vj j
Nb Ng
Xij X X
gij cos(di  dj ) (10) gmax
i (Vi
max
 Vi ) gmin
i (Vi  Vi
min
)
jVi Vj j i1 i1

Xij Nb
X XNg
jij sin(di  dj ) (11) zmax max
zmin min
jVi Vj j i (di  di ) i (di  di )
i1 i1

2. Inequality constraints X
Nl
cl (Slmax  Sl ) (17)
l 1
(a) Real power generation limit: This includes the upper and
lower real power generation limits of generators at bus-i
Knowing Lagrangian function, real and reactive power nodal
price at any bus-i can be determined as the partial derivative
Pgimin  Pgi  Pgimax , i 1, 2, . . . , Ng (12) of the Lagrangian function with respect to injected real

284 IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-gtd.2009.0026
www.ietdl.org

and reactive power equated to zero as @L=@Pi 0 and and non-rm [26]. These bilateral transactions can affect the
@L=@Qi 0. loading pattern of transmission facilities that may require
system operators (SO) to reschedule the system generating
The marginal price of real and reactive power at each units in order to accommodate them. A bilateral transaction
generator node can be obtained as must be accommodated without any violation of system
P  operating constraints such as transmission interface limits,
equipments ratings and system economic dispatch. These
@ i[Nb Ci (Pi )
lpi mmax
i  mmin
i
negotiated transactions have to be within available transfer
@Pi (18) capability of the transmission system to avoid occurrence of
   
@P @QLT any congestion in the transmission network. There are some
qpl 1  LT  qql intrinsic properties associated with the transactions such as
@Pi @Pi
column rule, row rule, range rule and ow rule that have
    been well dened in [31].
@PLT @Q
lqi hmax
i  hmin
i  qpl qql 1  LT (19)
@Qi @Qi For the determination of nodal price for hybrid electricity
market model, the additional power ow equations can be
The general form of Lagrange equation (17) can be written as added to modify the power ow injections at any bus-i.
These equations are
X
m X
n
X X
L(X , l, m) F (X ) lk hk (X ) mj gj (X ) (20) P db Tij , P gb Tij (22)
k1 j1
i j

At the optimal point, the following conditions must be P g P gb P gp (23)


satised as
P d P db P dp (24)

@L 
0, m  0 if gj (x ) 0 and P fb ACDF(P gb  P db ) (25)
@mi x,l,m
P fp ACDF(P gp  P dp ) (26)
mi 0 if gj (x ) , 0 (21)
P f P fb P fp (27)
Inequality constraints will be active only if the gradient of
the function and constraints are opposite as (rF )T rg  Distribution factors have been determined for calculating
0 ) mi  0, where X are the variables, li are the Lagrange power ows because of bilateral transactions [32]. The
multipliers corresponding to all equality constraints and mi bilateral transaction matrix (Tij ) and elements of the matrix
are the Lagrange multipliers corresponding to inequality are the secure bilateral transactions between seller bus-i and
constraints. In (17), these Lagrange multipliers have been buyer bus-j that have been determined solving a non-linear
represented with different symbols for each equality and optimisation problem minimising the deviations of
inequality constraints for distinction. transactions from the proposed transactions. Equations
(22) (27) are real power demand for bilateral transactions,
Derivatives of the total real and reactive power losses w.r.t. real power generation to meet bilateral transactions, total
real and reactive power injection are called loss sensitivity generation, total demand, real power ow for bilateral and
factors. pool demand and total real power ow considering both
pool and bilateral demands.
The optimisation problem has been formulated in general
algebraic modeling system (GAMS) using SNOPT solver For obtaining nodal price for hybrid electricity market,
[29]. MATLAB and GAMS interfacing has been used to an optimisation problem discussed in Section 3.1,
solve load ow at base case to obtain load ow data and minimising fuel cost subject to all equality and inequality
other parameters required for modelling algebraic equation constraints along with the additional equality constraints
in GAMS [30]. from (22) to (27) have been formulated and solved using
NLP. The new Lagrange equation can be formulated and
the respective derivatives can be obtained with these
2.1.1 Real power nodal price determination additional equality constraints for hybrid electricity market
(hybrid market): In a hybrid electricity markets, in model.
addition to pool demand, there can be transactions
between generation companies (GENCOs) and distribution
companies (DISCOs) acting as seller and buyers to sell and 2.1.2 Optimal secure transactions determination:
purchase power. A transaction is a bilateral exchange of In the hybrid market model, secure bilateral transactions have
power between GENCOs and DISCOs and can be rm been utilised in addition to pool demand. An objective

IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298 285
doi: 10.1049/iet-gtd.2009.0026 & The Institution of Engineering and Technology 2010
www.ietdl.org

function for determining secure bilateral transactions is 3 General OPF formulation in


to minimise the deviations of the transactions to be
determined from the proposed transactions as the presence of DG considering
XX demand variation
Min bij (Tij  Tij0 )2 (28) A general multi-objective mixed integer non-linear
i j
programming approach for hybrid electricity markets
considering the cost characteristics of distributed
Tij and T0ij are the secure and proposed bilateral transactions,
generators together with the fuel cost of other conventional
respectively, between GENCO bus-i and DISCO bus-j.
generators and the line loss minimisation has been
bij is a weighting factor indicating the importance of a
formulated to nd optimal location and number of
particular transaction which is taken as unity in the present
distributed generators
work, however, it can be any value for planning and
operational studies [31].
Min F (x, u, p, jint ) (37)
Operating constraints

1. Equality constraints: Real and reactive power injections at


subject to equality and inequality constraints dened as
bus-i are

Nb
X
Pi Pgi  r  Pdi Vi Vj [Gij cos(di  dj ) h(x, u, p, jint ) 0 (38)
j1

Bij sin(di  dj )] 8i 1, 2, . . . , Nb (29)


Nb
g(x, u, p, jint )  0 (39)
X
Qi Qgi  Qdi Vi Vj [Gij sin(di  dj )
j1
where x is state vector of variables V, d; u are the control
 Bij cos(di  dj )] 8i 1, 2, . . . , Nb (30) parameters, Pgi , Qgi , PDGi , QDGi, Pgb , Pgp, r; p are the
xed parameters Pdi, Pdb , Pdp , Qd, Tij; and jint is an integer
In addition to these equality constraints, the other equality variable with values f0,1g. The zero value represents
constraints to determine secure bilateral transactions are the absence and one value represents presence of the
same as (22) (27). distributed generator in the network.

2. Inequality constraints: These are the constraints on


Objective function F is
power generation limits of real and reactive power, real
power ow limit, transaction amount limit, voltage and
angle limits
Min F (x, u, p, jint )
8X 9
P min
g  P g  P max
g (31) >
> (a bgi Pgi cgi Pgi2 ) >
>
< i[N gi =
g
Qmin  Qg  Qmax (32) X
g g >
> int 2
(aDGi bDGi PDGi cDGi PDGi )>
>
: ji  ;
i[NDG
jP f j  P max
f (33) (N " ! #)
XL X X
0  Tij  Tijmax (34) l0pil  l0pjl  PLl (40)
l 1 i[sb j[eb
Vimin  Vi  Vimax (35)

dmin
i  di  dmax
i (36) The rst part of the objective function is the total
fuel cost with conventional and distributed generators.
Equation (34) is the limit on transactions with as Tijmax The second part of the objective function is the line loss in
maximum transaction amount. The problem of line l. The multiplying factor for the line losses is the
secure bilateral transactions determination has been wheeling cost which is the difference of marginal cost for a
modelled in GAMS using (22) (36). The problem for line l between bus-i and bus-j of real power associated
secure bilateral transactions have been modelled in GAMS with line l at base case [33]. Sets sb and nb are the start
and solved using SNOPT solver based on sequential bus and end bus in the second part of an objective
quadratic programming (SQP) algorithm for large- function. The bid curve coefcient data have been given in
scale constrained optimisation to determine transactions, Appendix 1.
Tij [29].

286 IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-gtd.2009.0026
www.ietdl.org

The line ows from bus-i to bus-j and bus-j to bus-i are generation limits. The data for upper and lower reactive
given as power limits have been given in Appendix 2.

Pijl Vi2 Gij  Vi Vj (Gij cos(di  dj ) Bij sin(di  dj )) (c) Optimal number of distributed generators: This includes
the limit on number of maximum distributed generators in
(41)
the network
Pjil Vj2 Gij  Vi Vj (Gij cos(di  dj )  Bij sin(di  dj ))
N
X DG
(42) NDG jint max
 NDG (47)
i
i1
3.1 Equality constraints
Power ow equations corresponding to both real and reactive The results have been obtained by solving the mixed integer
power balance equations are equality constraints that can be non-linear programming problem in GAMS using a discrete
modied in the presence of distributed generation for all continuous optimisation package (DICOPT) solver [29].
the buses as The ow diagram for solving the formulated optimisation
problem has been shown in Fig. 1a and interfacing of
Pi Pgi jint
i  PDGi  r  Pdi MATLAB and GAMS environment to obtain the solution
has been shown in Fig. 1b.
Nb
X
Vi Vj [Gij cos(di  dj ) Bij sin(di  dj )]
j1
4 Results and discussions
8i 1, 2, . . . , Nb (43) The proposed approach for an optimal distribution
generation location has been applied to IEEE 24-bus
Qi Qgi jint
i  QDGi  Qdi
reliability test system [27]. This network contains 32
Nb
X generators distributed among ten buses, and 38 branches
Vi Vj [Gij sin(di  dj )  Bij cos(di  dj )] (line plus transformers). The data for the system are on
j1 100 MVA base. Appropriate zones have been identied for
8i 1, 2, . . . , Nb (44) an optimal location of DG. These zones are based on
nodal price and real power loss variation at each bus. The
buses lying in the zones with higher variation of nodal
All other equality constraints (4), (5) and (22) (27) remain
price and loss sensitivity have been considered as most
same as dened in Section 2. r is the demand variation
suitable place for DG placement. These zones based on
factor for both pool and bilateral demand. It represents the
higher to lower variation have been termed as Zone 1,
change in the operating point of the system.
Zone 2 and so on.

3.2 Inequality constraints


4.1 Zones based on nodal price
In addition to the inequality constraints (31) (36) dened in
Section 2, two new inequality constraints have to be added in Based on the problem formulation as described in Section 2,
an OPF model with DG. the nodal prices of real power have been determined for both
pool as well as hybrid electricity market model. The real
Power generation limit: This includes the upper and lower power nodal prices for a pool model are given in Table 1.
real power generation limits of generators at bus-i
The price zones have been formed based on the nodal price
(a) Real power generation limit variation. The zone with higher price variation has been
termed as Zone 1 and the zone based on low price and less
min
PDGi max
 PDGi  PDGi , i 1, 2, . . . , NDG (45) variation of nodal price has been termed as Zone 2. In
Zone 2, the magnitude or prices at some buses may be
where PDGimin max
, PDGi are the minimum and maximum relatively higher compared to Zone 1; however, if their
generation limits. variation is lower compared to the prices in zone, they have
been placed in Zone 2. The variation of nodal real power
(b) Reactive power generation limit: This includes the upper price is shown in Fig. 2. The price zones are shown in Fig. 3.
and lower reactive power generation limits of distributed
generators at bus-i
4.2 Zones based on real power loss
Qmin max sensitivity factors
DGi  QDGi  QDGi , i 1, 2, . . . , NDG (46)
Based on exact loss formula, the real power loss in a system is
where, Qmin max
DGi , QDGi are the minimum and maximum given as dened in (6). The real power loss sensitivity can be

IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298 287
doi: 10.1049/iet-gtd.2009.0026 & The Institution of Engineering and Technology 2010
www.ietdl.org

Figure 1 Flow diagram for optimal location of distributed generation using MATLAB and GAMS environment
a Flow diagram for solving optimisation problem
b Interfacing of MATLAB and GAMS environment

288 IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-gtd.2009.0026
www.ietdl.org

Table 1 Real power nodal price in $/MWh at each bus for


Zone 1 and Zone 2

Bus no. Zone 1 Bus no. Zone 2


1 35.753 13 28.741
2 35.483 14 28.761
3 31.107 15 27.327
4 33.530 16 27.361
5 33.798 17 26.355
6 31.354 18 26.219
7 19.373 19 27.603
8 30.774 20 27.571
9 30.437 21 26.187
10 30.900 22 23.300
11 29.720 23 27.460
12 29.803
24 29.000 Figure 3 Zones based on real power nodal price

Table 2 Loss sensitivity factors for Zone 1, Zone 2 and


Zone 3

Bus Zone 1 Bus Zone 2 Bus Zone 3


no. no. no.
1 20.0056 11 20.0211 17 0.0669
2 20.0088 12 20.0189 18 0.0741
3 0.0028 13 0.0058 21 0.0782
4 20.0418 14 0.0012
5 20.0394 15 0.0466
Figure 2 Nodal price variation at each bus 6 20.0681 16 0.0430
7 20.2120 19 0.0378
obtained as
8 20.1699 20 0.0433

@PLT X BN 9 20.0249 22 0.0110


2 (aij Pj  bij Qj ) (48)
@Pi j1
10 20.0390 23 0.0501
24 0.0134
Real power loss sensitivity factors are calculated at each bus
and the zones are formed based on their variation at each
bus in the network. The values of the loss sensitivity factors factors at the respective buses. The magnitude of sensitivity
obtained at each bus are shown in Table 2 and their factors at some of the buses may be higher than the
variation at each bus is shown in Fig. 4. The zones sensitivity factors in the Zone 1, however, if their variation
obtained with loss sensitivity factors are shown in Fig. 5. is comparatively lower, these buses have been placed in the
With loss sensitivities obtained at each bus, three zones Zone 2 and Zone 3. Zone 1 with higher variation of loss
have been formed named as Zone 1, Zone 2 and Zone sensitivity factors has been considered a candidate zone
3. Zone 1 is the most sensitive zone from loss sensitivity for the placement of distributed generators. In Zone 3,
point of view because of their higher variation and Zone 2 however, magnitude of loss sensitivity is higher; however,
and Zone 3 are termed as low sensitive zones compared to their variation is comparatively lower than the variation of
Zone 1 because of their low variations of loss sensitivity loss sensitivity for the buses lying in Zone 2.

IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298 289
doi: 10.1049/iet-gtd.2009.0026 & The Institution of Engineering and Technology 2010
www.ietdl.org

fuel cost without and with the presence of DGs. Reductions in


average marginal price of real power and real power loss
reduction for both pool and hybrid market model have also
been determined. Based on number of DGs optimally located
at each selected load buses in the candidate zone, ve different
cases have been considered for both the pool as well as hybrid
market model. The results have also been obtained without
the presence of DGs for comparison. There can be more than
one distributed generator with different possibilities to place in
the selected zone at the load buses. The optimal location of
distributed generators among these buses has been obtained
using the mixed integer-based approach. Results have been
obtained considering different cases with different number of
distributed generators. The different cases without and with
optimally located DGs at the selected buses lying in the most
appropriate zone have been categorised as:
Figure 4 Real power loss sensitivity variation at each bus
Case 0: Without distributed generator.

Case 1: With one distributed generator.

Case 2: With two distributed generator.

Case 3: With three distributed generator.

Case 4: With four distributed generator.

Case 5: With ve distributed generator.

4.3.1 Results for pool electricity market model:


The nodal price variations for both real and reactive power
at each bus have been obtained without and with the
presence of DG for the different cases. The price variations
are shown in Figs. 6 and 7. It is observed from Fig. 6 that
in the presence of DGs, the nodal prices have been
considerably reduced and the variation of real power prices
has also become uniform at all the buses. With the
presence of DGs, it is observed that two price zones can be
represented by a single price zone. Thus, the consumers in
both the zones will pay similar price. The best results have
been obtained with four numbers of DGs. However,
locating more than four DGs, it is observed that only small
Figure 5 Zones based on real power loss sensitivity factors
reduction in real power nodal prices is found. With
penetration of more DGs in the network, the improvement
Based on the most sensitive zones obtained using nodal in the result is found to be marginal.
price and line loss sensitivity factors, a common zone for
placement of DG can be obtained considering both Fig. 7 shows the reactive power price variation without and
economic and operational advantage. Load buses 3, 4, 5, 6, with DGs. It is observed from Fig. 7 that the reactive power
8 and 10 lying in the common Zone 1 have been price is high at nodes 6 and 10 when compared to other buses
considered for obtaining an optimal location of DG at since at these nodes the reactive power absorption is quite
these respective buses. DG location has not been high because of the presence of reactor and transformers.
considered at generator buses. The DG buses have been The Lagrangian multiplier for reactive power can be both
modelled as PV buses in the network. positive and negative. The reactive power price negative
sign indicates the negative sign associated with Lagrange
multiplier corresponding to reactive power balance
4.3 Results for IEEE 24 bus test system equation, however, the price has to be paid for reactive
The results have been obtained for nodal real and reactive price power absorption as well as for injecting reactive power in
variations at each bus, loss reduction in the network and overall the network. For Case 3, the reactive power price at these

290 IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-gtd.2009.0026
www.ietdl.org

Figure 6 Nodal real power price variation without and with DGs (pool model)

Figure 7 Nodal reactive power price variation without and with DGs (pool model)

nodes reduces considerably because of the reduction of


reactive line ows as real power is available locally due to
DGs. Thus, DGs can be help to reduce reactive losses in
a system because of power available locally to loads and
maintaining better voltage prole. The average nodal price
variation at each bus is shown in Fig. 8. It is observed that
the nodal price reduces considerably in the presence of
DGs and becomes almost uniform. With DGs, the price of
real power is almost uniform in both the zones and
customers pay similar price in both the zones. After Case
3, there is no signicant reduction obtained in average
marginal price. Figure 8 Average nodal price for real power (pool model)

The real power loss reduction in the presence of DGs


is shown in Fig. 9. With DGs real power loss reduction is
found to be quite considerable in the network because of
the local real power supply and reducing line losses. The
fuel cost reduction in the presence of DGs is shown in
Fig. 10. The fuel cost reduces for all the cases. Minimum
fuel cost is found to be with DGs for Case 4 with four
numbers of DGs. For Case 5, the reduction is found to be
marginal. With more penetration of DGs, it is observed
that there is no considerable reduction of overall fuel cost.
The optimal size of distributed generators obtained for
different cases has been shown in Fig. 11. The number in Figure 9 Real power loss reduction (pool model)

IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298 291
doi: 10.1049/iet-gtd.2009.0026 & The Institution of Engineering and Technology 2010
www.ietdl.org

Figure 10 Fuel cost saving (pool model) Figure 12 Penetration level of DG (pool model)

non-linear optimisation problem as explained in Section 2


[31]. The bilateral transactions are given in Table 3.

The nodal price variations for both real and reactive power
at each bus have been obtained without and with the presence
of DG. The nodal price variations are shown in
Figs. 14 and 15. It is observed from Fig. 14 that the nodal
prices were having higher variations in both the zones
without DGs.

With DGs, the price variations and their magnitude were


Figure 11 DG size in p.u. (pool model) reduced in both the zones and variation of price is found to
be minimal because of the local supply of real power and
small parentheses with Cases 1 5 (. . .) in Fig. 11 represents reducing line losses. The optimal number of DGs is found
the optimal bus location of DG. to be three for hybrid model to obtain minimum value of
the objective function. With more penetration of DGs, it is
The penetration level of DGs has been determined as the found that there is no considerable reduction in the real
ratio of DG size to the total demand in the system. The DG power nodal prices. The average nodal price variation with
share for different cases has been shown in Fig. 12. Optimal DGs is shown in Fig. 16. It is observed that the average
number of DGs required obtaining best results of fuel cost nodal price at each node decreases with the presence of
savings and loss reduction is found to be four. The voltage DG and customers in both the zones will have almost the
prole without and with DG is shown in Fig. 13. same price to pay for energy and are beneted to utilise the
cheaper energy.
4.3.2 Results for hybrid electricity market model:
For the simulation of hybrid market model, optimal secure The reactive power nodal price variations without and
bilateral transactions have been determined by solving the with DGs are shown in Fig. 15. It is observed that at buses

Figure 13 Voltage prole without and with DG for pool model

292 IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-gtd.2009.0026
www.ietdl.org

Table 3 Secure bilateral transaction matrix

Value of transaction between generator and load bus (p.u.)


T(1,1) 0.493 T (1,2) 0.299 T(1,3) 0.199 T(1,15) 0.077 T(2,8) 0.463 T(2,10) 0.331
T(2,13) 0.307 T(2,15) 0.400 T (2,18) 0.199 T(7,1) 0.046 T(7,2) 0.154 T(7,4) 0.139
T(7,5) 0.153 T (7,6) 0.298 T(7,7) 0.361 T(7,9) 0.256 T(7,10) 0.223 T(7,13) 0.435
T(7,15) 0.892 T(13,3) 0.025 T (13,4) 0.125 T (13,13) 0.422 T (13,14) 0.420 T(13,16) 0.491
T(13,18) 1.466 T(13,19) 0.339 T(13,20) 0.628 T (15,13) 0.120 T (15,14) 0.527 T(15,19) 0.339
T(15,20) 0.628 T(16,7) 0.616 T (16,9) 0.598 T (16,13) 0.026 T(18,3) 0.624 T(18,6) 0.379
T(18,8) 0.015 T(18,18) 0.146 T(18,15) 0.192 T(21,3) 0.025 T(21,4) 0.107 T(21,5) 0.107
T(21,9) 0.053 T(21,10) 0.246 T(21,13) 0.063 T (21,14) 0.103 T (21,15) 0.080 T(21,10) 0.256
T(22,3) 0.025 T(22,5) 0.171

Figure 14 Nodal price for real power without and with DGs (hybrid market model)

Figure 15 Nodal price for reactive power (hybrid market model)

IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298 293
doi: 10.1049/iet-gtd.2009.0026 & The Institution of Engineering and Technology 2010
www.ietdl.org

Figure 16 Average nodal price variation for hybrid market


model
Figure 19 DG size in p.u. (hybrid model)

Figure 17 Real power loss reduction for hybrid market


model
Figure 20 Penetration level of DG (hybrid model)

3, 6, 10 and 24, the reactive power price is comparatively high


compared to other buses in the system because of higher
reactive power absorption at these nodes due to the
presence of reactor and transformers. The reactive power
prices reduce considerably with DGs because of supply of
local real power. Local real power supply from DGs can
help to reduce reactive losses in the lines and maintaining
better voltage prole.

The real power loss reduction in the presence of DGs is


Figure 18 Overall fuel cost saving for hybrid market model shown in Fig. 17. The real power loss reduction is found to

Figure 21 Voltage prole without and with DGs for hybrid market model

294 IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-gtd.2009.0026
www.ietdl.org

Figure 22 Nodal price for real power ($/h) with demand variation (r 1.05 to rmax 1.25) with Case 0 and Case 3 (pool
model)

Figure 23 Nodal price for reactive power ($/MVARh) with demand variation (r 1.05 to rmax 1.25) with Case 0 and
Case 3 (pool model)

Figure 24 Fuel Cost ($/h) with demand variation (r 1.05 to rmax 1.25) with Case 0 and Case 3 (pool model)

be considerable with the presence of DGs in the network found that with more penetration of DGs, there is no
because of the local power supply to the loads and reduced considerable reduction of the overall fuel cost and losses in
real and reactive power losses in the lines. The fuel cost the network.
reduction in the presence of DGs is shown in Fig. 18.
The minimum fuel cost is obtained for Case 3, although The optimal DG size in p.u. obtained for different cases is
compared to Case 1, there was marginal reduction shown in Fig. 19. The number in small parentheses with
obtained in the fuel cost. The fuel cost reduction for Cases 1 5 (. . .) in Fig. 19 represents the optimal bus
bilateral model was found to be comparatively less than the location for distributed generators in the network. The
fuel cost reduction for a pool-based market model. It is penetration level of DG for different cases is shown in

IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298 295
doi: 10.1049/iet-gtd.2009.0026 & The Institution of Engineering and Technology 2010
www.ietdl.org

Fig. 20. For hybrid market model, optimal number of DGs


to obtain best results for fuel cost savings and loss
reduction is three. The voltage prole without and with
DGs are shown in Fig. 21.

4.3.3 Results with demand variation for pool and


hybrid market model: The impact of real power demand
variation (r) on nodal price variation, fuel cost as well as real
power loss has also been determined for both pool as well as
hybrid market model. The demand variation impact on nodal
Figure 25 Real and reactive power loss with demand
price without DG (Case 0) and with DGs (Case 3) for pool
variation (r 1.05 to rmax 1.25) with Case 0 and Case
market model has been shown in Figs. 22 and 23. It has been
3 (pool model)
observed that with the demand variation, the nodal prices
tend to increase at each bus. However with DGs, the nodal
prices reduce at each bus at different loading point (Case 3).
The impact of demand variation on fuel costs, real power and
reactive power loss has been shown in Figs. 2426. The fuel
cost increases with different loadability factor and with DGs,
the fuel cost reduces at different operating points. It has been
observed that real power and reactive power loss increases
with increase in real power demand. With DGs (Case 3),
there is considerable reduction in real and reactive power loss
at a different operating point of the system.

Similar observations have been observed for hybrid market


model and real power loss variation and fuel cost saving for
Figure 26 Reactive power loss with demand variation Case 0 and Case 3 have been shown in Figs. 27 and 28 for
(r 1.05 to rmax 1.25) with Case 0 and Case 3 (pool different operating point of the system considering demand
model) variation.

5 Conclusions
In this paper, zones based on nodal price and line loss sensitivity
as economic and operational criteria have been obtained for
optimal location of DG. The zone with highest variation in
the nodal price and line losses has been considered for optimal
placement. Then, mixed integer non-linear programming
approach has been presented for optimal location and optimal
number of distributed generators in the suitable zone. The
pattern of real and reactive power nodal price, real power loss
and fuel cost saving have been determined for pool as well as
hybrid electricity market. The nodal price variations are
reduced drastically in both electricity market models. It is
Figure 27 Real power loss with demand variation (r 1.01
observed that with DGs, the real power losses reduce
to rmax 1.02) with Case 0 and Case 3 (hybrid model)
considerably for both market models and thus there can be
additional saving of energy and fuel cost. Optimal number of
DGs were found to be four for pool model and three for
hybrid model to obtain maximum fuel cost saving and losses
reduction in the network. The impact of different operating
point with demand variation has also been studied for nodal
price variation, fuel cost and real power loss without and with
DGs for both electricity markets. It is observed that with
increase in demand, the nodal price increases and at maximum
demand the nodal prices become very high, however, with
optimal DG location; there is considerable reduction in the
nodal prices at different operating points in the system for
Figure 28 Fuel cost ($/h) with demand variation (r 1.01 both market models. Similarly, fuel cost and losses tend to
to rmax 1.02) with Case 0 and Case 3 (hybrid model) increase with increase in demand and DG plays an important

296 IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-gtd.2009.0026
www.ietdl.org

role in saving fuel cost and reduces losses in the network. With [13] EL-KHATTAM W., BHATTACHARYA K., HEGAZY Y., SALAMA M.M.A.:
more penetration of DGs, no considerable improvement has Optimal investment planning for distributed generation
been obtained. With optimally located DGs, the consumers in a competitive electricity markets, IEEE Trans. Power
will benet by paying lower cost of energy and there will be Syst., 2004, 19, (3), pp. 1674 1684
overall improvement in terms of technical and economical
benets in the energy market operation. [14] EL-KHATTAM W., HEGAZY Y., SALAMA M.M.A.: An integrated
distributed generation optimization model for distributed
planning, IEEE Trans. Power Syst., 2005, 20, (2),
6 References pp. 1158 1165

[1] Impact of increasing contribution of dispersed [15] KEANE A., OMALLEY M.: Optimal allocation of distributed
generation on power system nal report. CIGRE WG generation on distribution networks, IEEE Trans. Power
37 23, September 1998 Syst., 2005, 20, (3), pp. 1640 1646

[2] DUGAN R.C., PRICE S.K.: Issues for distributed generations [16] CELLI G., GHIANI E., MOCCI S., PILO F.: A multi-objective
in the US. Proc. IEEE PES, Winter Meeting, January 2002, evolutionary algorithm for the sizing and siting for
vol. 1, pp. 121 126 distributed generation, IEEE Trans. Power Syst., 2005, 20,
(2), pp. 750 757
[3] Electric Power Research Institute web-page,
April 2008, http://www.epri.com/gg/newgen/disgen/ [17] BORGES C.L.T., FALCAO D.M.: Optimal distributed generation
index.html allocation for reliability, losses, and voltage improvement,
Electr. Power Energy Syst., 2006, 28, pp. 413 420
[4] Gas Research Institute: Distributed power generation:
a strategy for a competitive energy industry (Gas Research [18] ACHARYA N., MAHAT P., MITHULANANTHAN N.: An analytical
Institute, Chicago, USA, 1998) approach for DG allocation in primary distribution
network, Electr. Power Energy Syst., 2006, 28, pp. 669 678
[5] ACKERMANN T., ANDERSSON G., SODER L. : Distributed
generation: a denition, Electr. Power Syst. Res., 2001, [19] HARRISON G.P., PICCOLO A., SIANO P., WALLACE A.R.: Hybrid GA
57, pp. 195 204 and OPF evaluation of network capacity for distribution
generation connections, Electr. Power Energy Syst., 2008,
[6] EL-KHATTAM W., SALAMA M.M.A.: Distributed generation 78, pp. 392 398
technologies, denitions and benets, Electr. Power Syst.
Res., 2004, 71, pp. 119 128 [20] CANO E.B.: Using fuzzy optimization for distributed
generation allocation. Proc. Power Engineering, Large
[7] CHIRADEJA P., RAMKUMAR R.: An approach to quantify for Engg. System Conf., 10 12 October 2007
technical benets of distributed generation, IEEE Trans.
Energy Convers., 2004, 19, (4), pp. 764 773 [21] ESTEVEZ G.A., BEHNKE R.P., AVILA R.T., VARGAS L.S.: A competitive
market integration model for distributed generation, IEEE
[8] DONKELAAR M.: A survey of solutions and options Trans. Power Syst., 2007, 22, (4)
for integration of distributed generation in electricity
supply systems, J. Energy Environ., 2004, 15, (2), [22] PORKAR S., ABBASPOUR-TEHRANI FARD A., SAADATE S.: An approach
pp. 323 332 to distribution system planning by implementing distributed
generation in a deregulated electricity market. Proc. Power
[9] LOPES J.A.P., HATZIARGYRIOU N., MUTALE J., DJAPIC P., JENKINS N.: System Large Engg. System Conf., 10 12 October 2007,
Integrating distributed generation into electric power pp. 9095
systems: a review of drivers, challenges and opportunities,
Electr. Power Syst. Res., 2007, 77, pp. 1189 1203 [23] GAUTAM D., MITHULANANTHAN N.: Optimal DG placement in
deregulated electricity market, Electr. Power Syst. Res.,
[10] ROSEHART W., NOWICKI E.: Optimal placement of distributed 2007, 77, pp. 1627 1636
generation. Proc. 14th PSC Conf., 2002, pp. 15
[24] SADIGHIZADEH M., REZAZADEH A.: Using genetic algorithm
[11] NARA K., HAYASHI Y., IKEDA K., ASHIZAWA T.: Application of tabu for distributed generation allocation to reduce losses and
search to optimal placement of distributed generators. improve voltage prole, Proc. World Acad. Sci. Eng.
Proc. IEEE PES Winter Meeting, 2001, pp. 288 294 Technol., 2008, 27, pp. 251 256

[12] CELLI G., PILLO F. : Optimal distributed generation [25] SHAHIDEPOUR M., ALOMOUSH M.: Restructured electrical
allocation in MV distribution networks. Proc. IEEE PES power systems, operation, trading, and volatility (Marcel
Conf. on PICA, 2001, pp. 81 86 Dekker, Inc., New York, 2001)

IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298 297
doi: 10.1049/iet-gtd.2009.0026 & The Institution of Engineering and Technology 2010
www.ietdl.org

[26] GALIANA F.D., ILIC M.: A mathematical framework for the 8 Appendix 2
analysis and management of power transactions under open
access, IEEE Trans. Power Syst., 1998, 13, (2), pp. 681687 Reactive power limits of distributed and conventional
generators
[27] IEEE Reliability Test System: A report prepared by the
reliability test system task force of the applications of Bus no. Distributed Bus no. Conventional
probability methods subcommittee, IEEE Trans. Power generators generators
Appar. Syst., 1979, PAS-98, pp. 2047 2054 (p.u.) (p.u.)

[28] ELGERD I.O.: Electric energy system theory (An Qmin


g Qmax
g Qmin
g Qmax
g

Introduction, McGraw Hill, 1971) 3 0.05 0.15 1 20.25 0.50

[29] BROOKE A., KENDRICK D., MEERAUS A., RAMAN R., RASENTHAL R.E.: 4 0.03 0.20 2 20.25 0.50
A users guide, GAMS software (GAMS Development 5 0.03 0.20 7 0.50 2.00
Corporation, 1998)
6 0.02 0.20 13 0.00 2.40
[30] MATLAB and GAMS: interfacing optimization and 8 0.03 0.15 15 0.50 2.30
visualization software (Michael C. Ferris, 1999)
10 0.02 0.25 16 20.50 0.80
[31] CHENG J.W.M., GALIANA F.D., MCGILLS D.T.: Studies of bilateral 18 20.50 2.00
contracts with respect to steady state security in a
deregulated environment, IEEE Trans. Power Syst., 1998, 21 20.50 2.00
13, (3), pp. 1020 1025 22 20.60 0.96

[32] SHARMA A.K.: New secure bilateral transaction matrix 23 21.25 3.10
determination using AC distribution factors and impact of
TCPAR and TCSC on its pattern, Electr. Power Compon.
Syst., 2007, 35, pp. 921 943

[33] HAPP H.H.: Cost of wheeling methodologies, IEEE Trans.


Power Syst., 1994, 9, (1), pp. 147 156

7 Appendix 1
Bid curve data for distributed generators and conventional
generators [23, 27]

Bus Distributed Bus Conventional


no. generators no. generators
aDG bDG cDG agi bgi cgi
3 0.003 20.0 0.0 1 0.2917 35.07 3591.39
4 0.004 19.0 0.0 2 0.0 64.96 306.7
5 0.010 20.0 0.0 7 0.0322 19.18 1940.98
6 0.005 18.0 0.0 13 0.0322 19.18 649.99
8 0.002 15.0 0.0 15 0.0628 27.22 1829.71
10 0.003 20.0 0.0 16 0.0191 14.86 552.8
18 0.0191 14.86 1105.6
21 0.0086 30.0 1992.36
22 0.0112 14.17 927.15
23 0.0017 17.55 1160.23

298 IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 2, pp. 281 298
& The Institution of Engineering and Technology 2010 doi: 10.1049/iet-gtd.2009.0026
Copyright of IET Generation, Transmission & Distribution is the property of Institution of Engineering &
Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email articles for
individual use.

You might also like