Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Electric Power Systems Research 77 (2007) 1627–1636

Optimal DG placement in deregulated electricity market


Durga Gautam, Nadarajah Mithulananthan ∗
Electric Power System Management, Energy Field of Study, Asian Institute of Technology, P.O. Box 4, Klong Luang, Pathumthani 12120, Thailand
Received 19 August 2006; received in revised form 14 November 2006; accepted 16 November 2006
Available online 29 December 2006

Abstract
This paper presents two new methodologies for optimal placement of distributed generation (DG) in an optimal power flow (OPF) based wholesale
electricity market. DG is assumed to participate in real time wholesale electricity market. The problem of optimal placement, including size, is
formulated for two different objectives, namely, social welfare maximization and profit maximization. The candidate locations for DG placement
are identified on the basis of locational marginal price (LMP). Obtained as lagrangian multiplier associated with active power flow equation for
each node, LMP gives the short run marginal cost (SRMC) of electricity. Consumer payment, evaluated as a product of LMP and load at each load
bus, is proposed as another ranking to identify candidate nodes for DG placement. The proposed rankings bridges engineering aspects of system
operation and economic aspects of market operation and act as good indicators for the placement of DG, especially in a market environment. In
order to provide a scenario of variety of DGs available in the market, several cost characteristics are assumed. For each DG cost characteristic, an
optimal placement and size is identified for each of the objectives. The proposed methodology is tested in a modified IEEE 14 bus test system.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Distributed generation; Locational marginal price; Optimal power flow; Electricity market; Social welfare

1. Introduction the local load and effectively reduce the load. The placement of
DG, however, should be carried out with due consideration to
DGs are considered as small power generators that comple- its size and location. The placement should be optimal in order
ment central power stations by providing incremental capacity for the maximum benefit of DG implemented in the network.
to power system. Although DGs may never replace the central Improper placement in some situations can reduce benefits and
power stations, these can be an attractive option when constraints even jeopardize the system operation.
in transmission network prevent economic, or least expensive, Numerous techniques are proposed so far to address the via-
supply of energy reaching demand. However, penetration and bility of DGs in power system. Capacity investment planning of
viability of DG at a particular location is influenced by tech- distributed generation under competitive electricity market from
nical as well as economic factors. The technical merits of DG the perspective of a distribution company is proposed in Ref. [2].
implementation include voltage support, energy-loss reduction, An approach for optimal design of grid connected DG systems
release of system capacity, and improve utility system reliability in relation to its size and type to satisfy on-site reliability and
[1]. Economical merit, on the other hand, encompasses hedge environmental requirements is presented in Ref. [3]. Besides,
against high electricity price. This incentive is enhanced with several optimization tools, including artificial intelligence tech-
vertical unbundling of utilities and market mechanisms such as niques, such as genetic algorithm (GA), tabu search, etc., are
real time pricing. By supplying loads during peak load periods, also proposed for achieving the optimal placement of DG. An
where the cost of electricity is high, DG can best serve as a price optimization approach using GA for minimizing the cost of net-
hedging mechanism. work investment and losses for a defined planning horizon is
DG can have a great value in a highly congested area where presented in Ref. [4]. GA has been used to obtain penetration
LMPs are higher than elsewhere. In such situation, it can serve level of DG for minimizing the total cost of operation includ-
ing fixed and variable cost in Ref. [5]. The method for optimal
placement of DG for minimizing real power losses in power dis-
∗ Corresponding author. Tel.: +66 2 524 5405; fax: +66 2 524 5439. tribution system using GA is proposed in Ref. [6]. The gradient
E-mail address: mithulan@ait.ac.th (N. Mithulananthan). and second order methods to determine the optimal location for

0378-7796/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsr.2006.11.014
1628 D. Gautam, N. Mithulananthan / Electric Power Systems Research 77 (2007) 1627–1636

the minimization of losses or line loading is employed in Ref. power balance equation in OPF. The generator and customer bids
[7]. An iterative method that provides an approximation for the are taken as inputs to OPF. The base case OPF based on social
optimal placement of DG for loss minimization is demonstrated welfare maximizing algorithm evaluates the generation dispatch,
in Ref. [8]. Analytical methods for determining optimal loca- demands and prices at each of the nodes. The nodal prices so
tion of DG with the aim of minimizing power loss are proposed obtained are indicator for identifying candidate nodes for DG
in Ref. [9]. Placement and penetration of distributed generation placement. The placement is intended to meet the demand at a
under LMP based standard market design with the objective of lower price by changing the dispatch scenario.
generation cost minimization is proposed in Ref. [10]. Optimal The profit maximization problem is viewed from the perspec-
placement of DG with Langrangian based approach using tra- tive of DG owner, who chooses to place DG at the load nodes.
ditional pool based OPF and voltage stability constrained OPF In order for them to achieve maximum revenue out of the dis-
formulations is proposed in Ref. [11]. patched power, placement and size of DG chosen should reduce
Present study encompasses the placement of DG in a pool the LMP to a value that maximizes the profit. As higher LMP
based wholesale electricity market with centralized dispatch. value might considerably lower the revenue making the profit
DG is considered as a negative load. The placement problem is negative.
formulated for the two different objectives, namely, maximizing
social welfare and maximizing the profit of DG owner. 2.1. Social welfare maximization
The paper is organized in five sections. Section 2 sets out
the OPF formulation dealing with social welfare maximization The objective function is formulated as quadratic benefit
problem. Section 3 presents the methodology adopted to evalu- curve submitted by the buyer (DISCO) minus quadratic bid
ate the placement of DG wherein the rankings used to identify curve supplied by seller (GENCO) minus the quadratic cost
the candidate nodes for the placement are also discussed. The function supplied by DG owner.
OPF results and inferences drawn from the same are covered in N
Section 4. Several cases have been considered to depict possible 
max (Bi (PDi ) − Ci (PGi )) − C(PDGi ) (1)
scenarios and results have been shown in graphical and tabular
i=1
format. The conclusions that can be drawn from the analysis are
presented in Section 5. Alternatively, the maximization problem (1) can be formu-
lated as a minimization problem with multiplying the objective
2. Problem formulation function by −1.
N

The problem is formulated with two distinct objective
min (Ci (PGi ) − Bi (PDi )) + C(PDGi )
functions, namely, social welfare maximization and profit max- (2)
i=1
imization. Social welfare is defined as the difference between
total benefit to consumers minus total cost of production [12]. subject to
It is the sum of producers’ surplus and consumers’ surplus as
shown in Fig. 1. In general term, it represents the surplus to 2.2. Equality constraints
society and is maximum when the market price is equal to the
marginal cost of producing the last unit of electricity [12]. The network for the transmission of electric energy is mod-
The traditional OPF algorithm for cost minimization is modi- eled via the power balance equation at each node in the network.
fied to incorporate the demand bids, in addition to the generation The sum of power flows, active and reactive, injected into a node
bids. LMP is determined as the lagrangian multiplier of the minus the power flows extracted from the node has to be zero.
N

Pi = PGi + PDGi − PDi = vi [vj {Gij cos(δi − δj )
j=1

+Bij sin(δi − δj }] (3)

N

Qi = QGi − QDi = vi [vj {Gij sin(δi − δj )
j=1

−Bij cos(δi − δj }] (4)

2.3. Inequality constraints

Generation limits:
The generating plants have a maximum and minimum gen-
Fig. 1. Social surplus with quadratic supply and demand curves. erating capacity beyond which it is not feasible to generate due
D. Gautam, N. Mithulananthan / Electric Power Systems Research 77 (2007) 1627–1636 1629

to technical or economic reasons. Generating limits are speci- For generator bus,
fied as upper and lower limits for the real and reactive power
outputs. PDi = 0
Real power generation limits:
PDGi = 0
PGi
min
≤ PGi ≤ PGi
max

2.4. Profit maximization


Reactive power generation limits:
The profit maximization formulation constitutes two nested
Gi ≤ QGi ≤ QGi
Qmin max
blocks. The inner block is handled by the independent sys-
tem operator (ISO). In order to achieve the short-run economic
Line flow limit: optimum, ISO collects the electric power bids from suppliers,
The line flow limit specifies the maximum power that a consumers and DG placement and size from DG owner. The
given transmission line is capable of transmitting under given DG owner being one of the market participants, lies outside the
conditions. The limit can be based on thermal or stability con- block and submits the DG size they are willing to penetrate in
siderations. Thermal limits are usually considered for shorter the market. ISO then runs OPF taking into consideration the net-
lines. The following constraint checks for the absolute power work constraints. The objective of this OPF is to minimize the
flow both at sending and receiving ends of particular line to be total costs. This block allows overall control and coordination of
within the upper limit of the line. generation and transmission. The LMP obtained from the OPF
is used by the DG owner in order to calculate the profit which
Sij ≤ Sijmax is evaluated as revenue minus cost for the particular DG. The
process is iterative as LMP is also a function of DG penetration.
The profit with DG placement at each of the node is evaluated
Sji ≤ Sji
max
as:
Bus voltage limit: Profiti = λi × PDGi − C(PDGi ) (5)
Voltage limits refer to bus voltage to remain within an allow-
able narrow range of levels. where PDGi denotes the DG size at node i; λi denotes
the LMP at node i after placing DG; C(PDGi ) = aDGi +
vmin ≤ vi ≤ vmax bDGi PDGi + cDGi (PDGi )2 denotes the cost characteristic of DG
i i
at node i.
where N denotes the total number of buses in the system; PGi The optimization process will identify the node and corre-
denotes real power generated at bus i; PDi denotes real power sponding optimal DG size that will bring maximum profit to the
demand at bus i; PDGi denotes the power supplied by the DG DG owner.
at bus i. Bi (PDi ) = aDi + bDi PDi − cDi (PDi )2 , denotes purchaser
benefit functions at bus i; Ci = (PGi )aGi + bGi PGi + cGi (PGi )2 , 3. Methodology
denotes the producer offer (bid) price at bus i;
C(PDGi ) = aDGi + bDGi PDGi + cDGi (PDGi )2 , denote the For a specific combination of supplier and demand bid
cost characteristic of DG at bus i; vi denotes the voltage at curves, the base case OPF first calculates different electricity
bus i; δi denotes the power angle at bus i; Bij denotes the prices for different nodes in the network. The nodal prices are
susceptance of the line ij; Gij denotes the conductance of the obtained from the lagrangian multipliers of the non-linear equal-
line ij; QGi denotes reactive power generated at bus i; PGi max ity constraints. The increasing functions for supplier bids and
and PGi denotes upper and lower real power generation
min decreasing functions for the consumer bids are treated as the
marginal cost or benefits of the bidder. The difference in prices
limits of generator at bus i; Qmax Gi and QGi denote upper and
min
results from active line constraints and losses in the transmission
lower reactive power generation limits of generator at bus i;
system.
vmax
i and vmin
i denote upper and lower limits of voltage at bus
To identify candidate nodes for the placement of DG, two
i; Sij denotes the complex power transfer from bus i to bus j;
rankings are defined, namely, LMP based ranking and consumer
Sji denotes the complex power transfer from bus j to bus i;
payment (CP) based ranking.
Sijmax and Sjimax denote the complex power flow limit for line ij

and line ji.


3.1. Locational marginal price (LMP) based ranking
For base case OPF,
LMP is the lagrangian multipliers associated with the active
PDGi = 0 power flow equations for each bus in the system. LMP at any
For load bus, node in the system is the dual variable for the equality con-
straint at that node [13]. LMP is generally composed of three
PGi = 0 components, a marginal energy component (same for all buses),
1630 D. Gautam, N. Mithulananthan / Electric Power Systems Research 77 (2007) 1627–1636

a marginal loss component and a congestion component. Con- One scenario might be where price is high but load is relatively
sidering the case of real power spot price at bus i, LMP is given small, while in the other, price is relatively low but load is high.
by: The ranking based on consumer payment is intended to focus
N
on the later scenario wherein total nodal payment is given the
∂PL  L
∂Pij priority rather than the high price. The ranking will have overall
LMPi = λ + λ + μLij (6)
∂Pi ∂P effect of reducing dominant loads in the system. In effect, LMP
ij=1
goes down and the dominant customer would be better off, as the
LMPi = λ + λL,i + λC,i (7) amount they need to pay would be less compared to no DG case.
The candidate nodes are iteratively selected for the
where λ is the marginal energy component at the reference bus
placement. The placement is carried out with several cost charac-
which is same for all buses, λL,i = λ(∂PL /∂Pi ) is the marginal loss
teristics assumed for DG. As the placement technique is intended
component and λC,i = μLij (∂Pij /∂Pi ) is the congestion compo-
to bring down the LMP, DG with operating cost higher than LMP
nent. Thus, the spot price at each bus is location specific and
will find no incentive for placement. The DG with operating cost
differs by the loss component and the congestion component.
lower than those bided by supplier is expected to have higher
Theoretically, this location-based price equals the economically
penetration while the one with higher cost is expected to have
efficient market value of electricity at that point, factoring into
smaller penetration.
account constraints everywhere in the system.
Higher LMP implies a greater effect of active power flow 4. Simulation results and discussion
equations of the node on total social welfare of the system. In
other words, higher LMP implies higher the generation pressed The effects of DG penetration under the two scenarios,
by demand at that node. It thus provides indication that for namely, social welfare maximization and profit maximization,
the objective of social welfare maximization, injection of active are discussed in detail. The analysis is extended for the various
power at that node will improve the net social welfare. As the cost characteristics assumed for the DG.
DG is assumed to inject real power at a node, the node with high-
est LMP will have first priority for DG placement. Accordingly, 4.1. Cost characteristics used for DG
the load buses are ranked in descending order of LMPs with the
first node in the order as the best candidate for DG placement as Wide varieties of DG technologies with varying operating
shown below. characteristics are available in the market. To depict the vari-
⎡ ⎤ ation, assumptions are made for the cost characteristics. CHP
LMP1 units, due to their heat recovery system can deliver power at
⎢ ⎥
⎢ LMP2 ⎥ much cheaper price than the central generation. The technolo-
⎢ ⎥
⎢ ⎥ gies such as fuel cells are characterized by their high cost while
LMP = ⎢ LMP3 ⎥ (8)
⎢ . ⎥ technologies such as reciprocating engines and gas turbines lie
⎢ . ⎥
⎣ . ⎦ somewhere in the middle. In order to accommodate the vari-
LMPn eties of DG units, assumptions are made on the basis of the
cost characteristics of central generation. Table 1 shows the cost
where n is the number of load locations. characteristics of DGs considered in this work.
The cost comparison among the various units is made as per
Best location = index {max(LMP)} (9)
the incremental cost. Incremental cost is a function of power out-
put of the unit where slope indicates cost to produce incremental
3.2. Consumer payment based ranking
quantity and intercept indicates no load cost. Other conditions
remaining the same, the lesser the slope, the lower the incre-
CP calculated as the product of LMP and capacity of load is
mental cost and higher the penetration. The crossing over of
considered as another criterion to segregate candidate nodes for
two different incremental cost characteristics reveals that opera-
DG placement. Thus, the CP evaluated at the load bus i is the
tional cost effectiveness depends on power output. The crossing
product of LMP and load at bus i.
over is determined by no load cost and slope of the curve. The
⎡ ⎤
CP1
⎢ ⎥ Table 1
⎢ CP2 ⎥
⎢ ⎥ Distributed generation data
⎢ ⎥
CP = LMPi × Loadi = ⎢ CP3 ⎥ (10) DG ID aDG bDG cDG
⎢ . ⎥
⎢ . ⎥
⎣ . ⎦ DG1 0.002 15 0
DG2 0.004 19 0
CP4 DG3 0.04303 20 0
DG4 0.25 20 0
Best location = index {max(CP)} (11)
DG5 0.1 30 0
CPi reflects the total amount the consumer at node i need to DG6 0.01 40 0
DG7 0.003 43 0
pay for the electricity. The ranking is influenced from the fact that
market for DG placement can be viewed from two standpoints. Note: aDG , bDG , cDG are quadratic cost coefficient of DG.
D. Gautam, N. Mithulananthan / Electric Power Systems Research 77 (2007) 1627–1636 1631

Table 2
Ranking based on LMP
Rank Bus PD (MW) LMP ($/MWh)

1 14 33.91 54.644
2 11 39.68 54.413
3 10 12.65 52.229
4 9 26.26 50.698
5 13 10.7 50.501
6 7 19.89 49.204
7 4 56.05 47.758
8 12 11.18 46.658
9 5 26.52 43.636

is extended for various cost characteristics assumed for the


DG.
The system used in this study, modified IEEE 14 bus test
system, consists of 9 load buses and 5 generators. The loads are
Fig. 2. Cost characteristic of various DG.
assumed to be elastic with power factor of 0.91 (lagging). The
maximum social welfare is found to be 4425.31 $/h. The total
unit cheaper due to lower no load cost can prove to be expensive real power loss in the system is 7.042 MW. The generation, load
beyond certain power output if the slope is large and vice versa. and LMP corresponding to the maximum social welfare for the
The cost characteristic of DG units considered in modified base case are determined at each node. Results revealed that
IEEE 14 bus test system is shown in Fig. 2. generator buses have lower values of LMP compared to the load
The cost characteristics considered have wide variety of buses.
slopes and accordingly, intersection at several points. Hence,
the comparative study of operational cost among the units relies 4.3. Candidate nodes for DG placement
on power output.
The incremental cost characteristics of various DGs consid- The system has a maximum load of 56.05 MW at node 4.
ered in this study is shown in Fig. 3. As the quadratic component Contrary to the node with maximum load, the highest LMP of
of DG1 and DG2 is very small, their incremental cost is almost 54.64 $/MWh is recorded at node 14 as shown in Table 2. This
constant for the entire range of output. Same is the case with shows high LMP should not necessarily be at the node with high
the DG6 and DG7. However, DG3, DG4 and DG5 show mono- load. Load exceeding the transmission capacity at a particular
tonically increasing incremental cost with crossover at several location might lead to high LMP. However, due to the loop flow,
points. loads at other nodes and overall network configuration do play a
role in determining LMP. The ranking of the load buses accord-
4.2. Base case analysis ing to LMP and consumer payment are shown in Tables 2 and 3,
respectively.
The social welfare maximization problem encompasses the
welfare of consumers as well as producers. The analysis
4.4. DG placement for social welfare maximization

The optimal DG size for each of the load bus is determined


from the social welfare maximizing problem. Results revealed
that there is an optimal DG size at each of the load bus for
which the net social welfare is maximum. However, the max-

Table 3
Ranking based on consumer payment
Rank Bus PD (MW) LMP ($/MWh) Consumer payment ($/h)

1 4 56.05 47.758 2676.84


2 11 39.68 54.413 2159.11
3 14 33.91 54.644 1852.98
4 9 26.26 50.698 1331.33
5 5 26.52 43.636 1157.23
6 7 19.89 49.204 978.67
7 10 12.65 52.229 660.70
8 13 10.7 50.501 540.36
9 12 11.18 46.658 521.64
Fig. 3. Incremental cost characteristics of various DG.
1632 D. Gautam, N. Mithulananthan / Electric Power Systems Research 77 (2007) 1627–1636

imum net social welfare obtained from these optimal sizes is


different from one load bus to another. Another worth notice-
able point is that the placement as well as penetration of DG
is found to vary with the cost characteristics used. Even for the
same load bus, different optimal sizes are obtained when differ-
ent cost characteristics are used. The cheaper the unit the higher
the penetration and so is the net social welfare. This shows DG
penetration as well as social welfare is a function of DG cost
characteristics.
The study has been carried out to identify the optimal place-
ment and penetration when the DG is cheaper or expensive than
the existing central generation. The results associated with two
expensive DGs, namely, DG6 and DG7 are presented as sample
results. However, summary of the results for all DGs are given
in the end of this section.

4.5. Placement of DG6


Fig. 5. Optimal DG size at respective nodes with DG6.
The maximum net social welfare that can be achieved when
the placement of DG6 is carried out at different load buses is 4.6. Placement of DG7
shown in Fig. 4.
The corresponding optimal DG size at each of the load bus The maximum net social welfare that can be achieved when
is also shown in Fig. 5. For instance, if placement is to be car- the placement of DG7 is carried out each of the load buses is
ried out at node 14, optimal size of DG for the social welfare shown in Fig. 7. The maximum net social welfare of 4483.04 $/h
of 4577.18 $/h is 42.84 MW. Similarly, for placement at bus is obtained when the placement is made at node 14. Corre-
11 the optimal size giving the social welfare of 4563.29 $/h is sponding optimal DG size is found to be 25.33 MW. The smaller
48.69 MW and so on. It is interesting to note that net social wel- optimal size compared to the placement of DG6 can be attributed
fare is maximized for the case of DG at node 14. Hence, the to higher incremental cost of DG7.
optimal placement of DG6 is node 14 with the optimal size of The optimal size of DG after placing DG7 at each of these
42.84 MW. The social welfare maximization is found to capture buses is shown in Fig. 8. From the figure it is revealed that no
the first candidate node of LMP ranking given in Table 2. More- DG is selected for node 5. As apparent from Table 2, node 5 is
over, the ranking is found to capture first four candidate nodes the last candidate node for DG placement. Hence, the placement
accurately in the same order. is found to follow the ranking based on LMP.
The variation of net social welfare with respect to DG size for The variation of net social welfare with respect to DG size
node 14 is shown in Fig. 6. As apparent from the figure, beyond for the placement of DG7 at node 14 is shown in Fig. 9.
the optimal size there is a reduction in net social welfare. For Results revealed that placements as well as sizes vary with
non-optimal size same social welfare can be obtained for two the cost characteristics. The summary of results corresponding
different sizes of DG. However, maximum net social welfare is to the placement of all the seven DGs considered in the study is
obtained only for the optimal DG size. given in Table 4.

Fig. 4. Net social welfare at respective nodes with DG6. Fig. 6. Social welfare vs. DG size for the placement of DG6 at node 14.
D. Gautam, N. Mithulananthan / Electric Power Systems Research 77 (2007) 1627–1636 1633

Table 4
Result summary for the placement of DG with different cost characteristics
DG Best location Optimal DG Social welfare Remarks
size (MW) ($/h)

DG1 Bus 4 202.62 8460.47 CP based ranking


DG2 Bus 4 195.05 7586.12 CP based ranking
DG3 Bus 9 141.28 6427.09 –
DG4 Bus 14 41.94 4993.77 LMP based ranking
DG5 Bus 14 50.38 4848.79 LMP based ranking
DG6 Bus 14 42.84 4577.18 LMP based ranking
DG7 Bus 14 25.33 4483.04 LMP based ranking

the penetration and so is the net social welfare. Hence, the lower
incremental cost followed by higher penetration is found to favor
Fig. 7. Net social welfare at respective nodes with DG7. consumer payment based ranking given in Table 3.

4.7. DG placement for profit maximization

The present discussion encompasses the placement of the


same DG characteristics as the one considered for social welfare
maximization.

4.8. Placement of DG6

Fig. 10 shows the corresponding maximum profit at each load


bus after the placement of DG6. Fig. 11 shows the optimal DG
size corresponding to the maximum profit at each of the load bus.
The maximum profit of 75.135 $/h is found for the placement at
node 14. The corresponding optimal DG size is 21.76 MW which
Fig. 8. Optimal DG size at respective nodes with DG7. is less than the value obtained for social welfare maximization
shown in Fig. 5.
Interestingly, it is observed that for the placement of DG1 The variation of profit with the penetration of DG6 at load
and DG2, social welfare is maximized when placement is made bus 14 is shown in Fig. 12. The maximum profit is found for the
at node 4. In other words, the placement is found to track first optimal size as shown in Fig. 11.
candidate node of consumer payment based ranking rather than As the penetration increases, LMP at a node will reduce.
LMP ranking. Furthermore, as shown in Table 4 the penetration If the LMP reduces to a value making the consumer payment
of DG1 and DG2 is higher compared to that of DG6 and DG7. lower than the operating cost of DG, profit for DG owner would
The higher penetration can be attributed to the lower incremental be negative. This is apparent from Fig. 12 which shows that
cost as is apparent from Fig. 3. The cheaper the unit, the higher beyond the optimal DG size, profit will decrease and can even
be negative if the penetration reaches a higher value.

Fig. 9. Social welfare vs. DG size for the placement of DG7 at node 14. Fig. 10. Maximum profit at respective nodes with DG6.
1634 D. Gautam, N. Mithulananthan / Electric Power Systems Research 77 (2007) 1627–1636

Table 5
Result summary for the placement of DG with different cost characteristic
DG Best location Optimal DG Profit ($/h) Remarks
size (MW)

DG1 Bus 4 119.43 2766.58 CP based ranking


DG2 Bus 4 119.43 2260.33 CP based ranking
DG3 Bus 9 105.38 1592.49 –
DG4 Bus 11 37.29 470.72 –
DG5 Bus 4 50.46 323.53 –
DG6 Bus 14 21.76 75.14 LMP based ranking
DG7 Bus 14 12.55 29.25 LMP based ranking

Fig. 11. Optimal DG size at respective nodes with DG6.

Fig. 14. Optimal DG size at respective nodes with DG7.

Profit maximization results reveal that there is no profit for


DG owner when the placement is carried out at bus 5.
Results show that even for the DG with same cost character-
istic, profit maximization comes up with the lower optimal size
compared to social welfare maximization as can be seen from
Fig. 12. Profit vs. DG size for placement of DG6 at node 14. Tables 4 and 5.

4.9. Placement of DG7 4.10. Comparison between social welfare and profit
maximization
The profit that can be achieved to DG owner with the place-
ment of DG7 and corresponding optimal sizes at each of the load Tables 6 and 7 show the comparative study of results obtained
buses is shown in Figs. 13 and 14, respectively. The variation from two placement techniques. The placement of DG6 and DG7
of profit with the penetration of DG7 at load bus 14 is shown in
Fig. 15.

Fig. 13. Maximum profit at respective nodes with DG7. Fig. 15. Profit vs. DG size for placement of DG7 at node 14.
D. Gautam, N. Mithulananthan / Electric Power Systems Research 77 (2007) 1627–1636 1635

Table 6
Comparison of results for placement of DG6
DG Bus Social welfare maximization Profit maximization

Social welfare ($/h) LMP ($/MWh) PDG (MW) Profit ($/h) LMP ($/MWh) PDG (MW)

14 4577.18 40.86 42.84 75.135 43.67 21.76


11 4563.29 40.97 48.69 70.818 43 25.87
10 4537.51 40.87 43.32 56.611 42.79 22.08
9 4524.67 40.94 46.88 52.136 42.87 19.53
13 4508.79 40.86 42.8 46.013 42.19 23.6
7 4455.76 40.59 29.59 9.864 44.39 2.26
4 4519.41 41.02 50.98 20.618 43.97 5.26
12 4475.75 40.56 28.14 26.905 41.95 14.99
5 4441.02 40.45 22.44 10.629 41.3 8.8

Table 7
Comparison of results for placement of DG7
DG Bus Social welfare maximization Profit maximization
Social welfare ($/h) LMP ($/MWh) PDG (MW) Profit ($/h) LMP ($/MWh) PDG (MW)

14 4483.04 43.15 25.33 29.249 43.112 12.55


11 4462.79 43.15 24.32 19.109 44.553 12.61
10 4452.27 43.11 19.02 13.655 44.386 10.07
9 4443.48 43.10 17.28 9.131 44.085 8.63
13 4431.32 43.07 11.82 3.09 43.527 6.07
7 4428.72 43.01 2.26 3.127 44.392 2.26
4 4431.57 43.03 5.26 5.033 43.973 5.26
12 4427.19 43.03 5.47 0.959 43.354 2.78
5 4425.31 41.65 0.00 0 41.646 0.00

is observed for social welfare as well as profit maximization system, and similarly, the one with higher incremental cost, the
problem. The corresponding values of LMP at each of the nodes lower penetration. Considerable reduction in central generation
after placing the optimal size of DG are tabulated. dispatch is observed with high DG penetration.
LMP and consumer payment have been identified as tools
5. Conclusions for screening candidate nodes for DG placement. The DGs with
lower incremental cost compared to central generating stations
The paper proposes two new methodologies of DG placement have a higher penetration and is found to follow the ranking
in an OPF based wholesale electricity market. Optimal place- made on the basis of consumer payment. On the other hand, the
ment and size is identified for social welfare as well as profit DGs with higher incremental cost have lower penetration and is
maximization problem. For each DG cost characteristics, there found to follow the ranking made on the basis of LMP.
is an optimal location and size for which the net social welfare is It has also been observed that a high penetration of DG can
becoming maximum. The same condition is found to hold true also lead to negative profit for the DG owner. The situation is
for profit maximization, as well. found to prevail when LMP reduces considerably due to high DG
For the DG placement at a node, social welfare maximization penetration. If the LMP reduces to a value making the consumer
ends up with lower LMP value compared to profit maximiza- payment lower than the operating cost of DG, profit for DG
tion. Accordingly, optimal DG size for profit maximization is owner would be negative. Under such scenario, DG owner will
lower than that for social welfare maximization. This is due find no incentive for placement.
to the fact that social welfare is concerned with consumer as
well as producers surpluses; however, profit is concerned only
References
with the surplus to producers which will acquire high value
as the price increases. The high LMP results in higher con- [1] P.P. Barker, Determining the impact of distributed generation on power sys-
sumer payment with a consequent increase in the revenue to DG tems: Part 1—radial distribution systems, in: Proceedings of IEEE Power
owner. Engineering Society Summer Meeting, 2000, pp. 1645–1656.
DG penetration is found to reduce the dispatch of central [2] W.E. Khattam, K. Bhattacharya, Y. Hegazy, M.M.A. Salama, Optimal
generation. The optimal placement and penetration is found to investment planning for distributed generation in a competitive electricity
market, IEEE Trans. Power Syst. 19 (August (3)) (2004) 1674–1684.
depend on the cost characteristics of DG as well as those of [3] M. Pipattanasomporn, M. Willingham, S. Rahman, Implications of on-
central generations. The DG with incremental cost lower than site distributed generation for commercial/industrial facilities, IEEE Trans.
the central generation is found to have a higher penetration in the Power Syst. 20 (February (1)) (2005) 206–212.
1636 D. Gautam, N. Mithulananthan / Electric Power Systems Research 77 (2007) 1627–1636

[4] G. Celli, F. Pilo, Optimal distributed generation allocation in MV distri- [9] C. Wang, M. Hashem Nehrir, Analytical approaches for optimal placement
bution networks, in: 22nd IEEE PES International Conference on Power of distributed generation sources in power systems, IEEE Trans. Power
Industry Computer Applications PICA 2001, Sydney, Australia, May 2001, Syst. 19 (November (4)) (2004) 2068–2076.
pp. 81–86. [10] P. Agalgaonkar, S.V. Kulkarni, S.A. Khaparde, S.A. Soman, placement and
[5] G. Celli, F. Pilo, Penetration level assessment of distributed generation penetration of distributed generation under standard market design, Int. J.
by means of genetic algorithms, in: IEEE Proceedings of Power System Emerg. Electric Power Syst. 1 (1) (2004).
Conference, Clemson, SC, 2002. [11] W. Rosehart, Optimal placement of distributed generation, in: Now-
[6] N. Mithulananthan, Than Oo, Le Van Phu, Distributed generator placement icki (Ed.), 14th PSCC, Sevilla, 24–28 June 2002. Available online:
technique in power distribution system using genetic algorithm to reduce http://www.pscc02.org/papers/s11p02.pdf.
losses, Thammasat Int. J. Sci. Tech. 9 (September (3)) (2004) 56–62. [12] Electricity economics regulation and deregulation, in: G. Rothwell, T.
[7] N.S. Rau, Y.H. Wan, Optimum location of resources in distributed planning, Gomez (Eds.), IEEE Press Power Engineering Series, John Wiley & Sons,
IEEE Trans. Power Syst. 9 (4) (1994) 2014–2020. 2003.
[8] T. Griffin, K. Tomsovic, D. Secrest, A. Law, Placement of dispersed gen- [13] M. Shahidehpour, H. Yamin, Z. Li, Market Operations in Electric Power
erations systems for reduced losses, in: Proceedings of the 33rd Hawaii Systems, John Wiley & Sons, Inc., 2002.
International Conference on System Sciences, Hawaii, 2000. Available
online: http://tomsovic.eecs.wsu.edu/Vitae/Publications/GRIF00.pdf.

You might also like