Professional Documents
Culture Documents
net/publication/274403946
CITATION READS
1 849
4 authors, including:
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Unit Commitment with Considering Energy Efficiency in the Smart Grid View project
A fuzzy Analysis to connect the large-scale distant wind farm to grid in probabilistic multi objective transmission expansion planning View project
All content following this page was uploaded by Mohammad Mahdi Hosseini Biyouki on 04 April 2015.
the restructured power market, beside the secure price-sensitive loads to establish the market clearing
operation of power system as the main purpose, the price and determinegeneration of each generatoras
participating entities like GenCo and TransCo profit well as amount of dispatchable loads according to
maximization must be considered. On the other the system parameters.In the proposed scheme, as a
hand, ISO’s aim to maximize social welfare is contingency occurs, under the smart market
obtained through minimizing power market entities environment, the minimum load shed that satisfies
cost as well ascustomer payment minimization the optimization objective function to maximize
considering operational constraints and supply GenCo profit and maximizing loadability limit as
limits.In a competitive environment, the locational an index ofVSM is obtained.
marginal prices (LMPs) are determined according
to the GenCo offers and lines congestions. When a 2. Loadability limit as a voltage stability index
generator contingency occurs, the GenCos are Loadability limit of a power system can be
obliged to pay the penalty of load shedding which is considered as the voltage stability index of the
in term of expected energy not supplied (EENS) or power system [21]. One of the commonly used
in a one hour snapshot of time domain the expected methods to evaluate voltage stability of a system is
power not supplied (EPNS). GenCos also must continuation power flow (CPF). At the voltage
provide the generation costs and the congestion stability limit of a power system P-V curve, the
charges [18]. Jacobian matrix of power flow equations becomes
Congestion in the power system is a consequence of singular. CPF implements successive power flow
occurrence of a generator or line outage and solutionsconsidering anincremental load profile.
insufficient transmission lines capacity to transmit The CPF procedure consists of prediction and
required power from generators to load demands at correction steps. Considering a determined solution,
a specific location. This leads to locational marginal a tangent predictor is utilized to estimate the next
price (LMP) difference in the system buses. The solution for a specified pattern of load increase. In
LMP increase in the congested line buses causes a the next step, the corrector step determines the
change in the market equilibrium point where the precise solution using Newton-Raphson method
customers’ payments increase in comparison with through CPF. Once again, a new prediction is made
the local generation cost. Congestion charge (rent) for a specified load increment considering the new
is the charge collected by the ISO for investment in tangent vector. Then the corrector step is applied
network reinforcement and is defined as difference again. This process terminates when the critical
of LMPs in the buses of a congested line multiplied point of P-V curve is reached. A load increment
by the power flow from that line. The congestion factor for active and reactive power of demand is
charge is formulated as follows [19, 20]: applied to execute CPF procedure.
nB
C cong ,ij = ∑ (λ
i≠j
i − λ j )Pij , Pij ≥ 0 (1) 3. Problem Formulation
In a restructured power system, the ISO tries to
maximize the social welfare through minimizing
where C cong ,ij is the congestion charge of branch market participants’ costs. So when a contingency
ij in $ / h , n B is the number of buses, λi and occurs in this market-based power system, while
λ j are the LMPs of buses i and j in $ / MW h , executing optimal shedding of fixed loads, the
objective function considered is maximization of
and Pij is the power flow of line ij in MW . GenCo profit considering non-linear constraints and
When a line outage occur the access and usage the voltage stability of the system. This multi-
revenue that was obtained by the TransCo objective function is comprised of two sub-
decreases, while the congestion charges collected optimization problems. The first one is the GenCo
maximum profit which is obtained from subtracting
by ISO increases. The EPNS expenses due to a line
generation costs, EPNS expenses, and congestion
outage may be partially recovered through the
charges from sold power revenues, which is
increment of congestion charges.However, this expressed as in (2).The second part is the maximum
paper proposes a steady state load shedding scheme loadability limit of the system which is calculated
where a multi-objective function tomaximize through integration of CPF, as expressed in(3):
technical and economic issues of restructured power nG nG nB
system is considered. The objective function
considers two sub-optimization problems: The
f1 = ∑i =1
RGenCo − ∑
i =1
C GenCo − ∑ EPNS −
i =1
GenCo profit maximization and as a consequence nB (2)
social welfare maximizationand the loadability limit
maximization of the system as an index for voltage
∑
i =1
C cong ,ij
i≠j
security margin (VSM) as a power system
component outage occurs.In the GenCo profit f 2 = Psys (3)
maximization sub-optimization problem, the smart where nG is the number of buses with generator,
market procedure is utilized where the ISO as the
RGenCo is the revenue of the GenCo in $ / h ,
smart market authority receives GenCo’s generators
offers blocks and dispatchable loads bids blocks as C GenCo is the cost of the generator in $ / h ,
6th Global Conference on Power Control and Optimization, Las Vegas, USA, 6-8 August 2012
( )
is called a swarm. Each agent’s position is
QGc ,i − Q Dc ,i − ∆Q Dc ,i = represented by its x, y axis position and also its
nB (6) velocity is expressed by the x axis and y axis
− ∑ (
V ic . V jc . Y ij .Sin θ ij − δ ic + δ cj ) velocity (vx and vy). The position and velocity
i =1 information of each agent represents the
modification of each agent’s position. Each agent
,i ≤ PG ,i ≤ PG ,i
PGmin c max
(7)
knows its best value so far (pbest) and its x,y
QGmin
,i ≤ QGc ,i ≤ QGmax
,i (8) position. Moreover, each agent knows the best
value so far in the group (gbest) among pbests.
,i ≤ PD ,i ≤ PD ,i
PDmin c n
(9)
Each agent tries to modify its position using the
S ic, j ≤ S imax
,j (10) following information: The current positions (x,y),
the current velocities (vx, vy), the distance between
V i min ≤ V ic ≤ V i max (11) the current position and pbest, and the distance
where superscript n and c are indexes of normal between the current position and gbest. This
modification can be represented by the concept of
andcontingency states, PGc ,i is active power
velocity (modified value for the current positions)
generation in contingency state, PDc ,i is load active [22]. In a D-dimensional search space, the position
of i-th agent can be represented by
powerdemand in contingency state, ∆PDc ,i is
vector s m = ( s m 1 ,..., s md ,..., s mn ) and the velocity
amount of loadactive power shed, V ic and V jc are of i-th agent can be proposed by
buses voltages in contingency state, Y ij is v m = (v m 1 ,...,v md ,...,v mn ) . The best previous
admittance of line ij , θij is admittance angle of experience of agent m is proposed by
pbest m = ( pbest m 1 ,..., pbest md ,..., pbest mn ) and
line ij , δ ic and δ cj are voltage angle in buses i and
the best value among all the experiences of all
j in contingency state, QGc ,i is reactive power agents in the group is stored in
gbest m = ( gbest m 1 ,..., gbest md ,..., gbest mn ) .
generation in contingency state, Q Dc ,i is load
Velocity of each agent can be modified by (12):
reactive power in contingency state, ∆Q Dc ,i is k +1
v md = ω v md
k
+ c1 rand 1 × ( pbest md − s md
k
)+
shed, PGmin (12)
c 2 rand 2 × ( gbest d − s md )
amount of load reactive power ,i and k
PGmax
,i are minimum and maximum active power
where v mk is velocity of agent m at iteration k, ω is
generation, QGmin max
,i and QG ,i are minimum and
weighting function, c1 and c 2 are weighting
maximum reactive power generation, PDmin
,i is coefficients, rand is a random number between 0
minimum amount of load to be supplied, PDn ,i is and 1, s mk is current position of agent m at iteration
k, pbest m is pbest of agent m, and gbest is gbest of
load active power demand in normal state, S ic, j is
the group. c1 and c 2 try to pull each agent toward
apparent power flow of line ij in contingency
pbest and gbest position[5]. According to the early
6th Global Conference on Power Control and Optimization, Las Vegas, USA, 6-8 August 2012
examinations c1 and c 2 are often set to 2.0 which is importance coefficients of 2 ( γ i = 2 ), area 2 buses
also applicable to power system optimization load importance coefficients are equal to 1.7
problems.Velocity of an agent can be changed using ( γ i = 1.7 ), and area 3 buses load importance
three vectors mentioned in (12). The velocity is coefficients are considered to be 1.3 ( γ i = 1.3 ). The
usually limited to a certain maximum value.
The new position of agent m considering the current modified system has three 30 MW dispatchable
position of it (searching point in the solution space) loads in buses 7, 15, and 30. As a contingency in
the system occurs, these price-sensitive loads may
can be obtained using the following equation:
k +1 k +1
be removed or remain in the system according to
s md = s md
k
+ v md (13) the settled LMP of each bus. According to the pre-
The weighting function ω is obtained from (14)and contingency operating conditions and after
is utilized in PSO algorithm to improve its executing OPF and smart market, generator 2
performance through providing a balance between generates 48.78 MW of the total generation of
global and local explorations. This leads to 202.16 MW among 6 generators.Therefore, it seems
averagely fewer iterations to obtain a sufficiently that outage of thisgenerator may have a
optimal solution. considerable effect on the stability of the system
and causes ISO to shed an amount of fixed and
ω max − ω min
ω k +1 = ω max − × iter (14) dispatchable loads to prevent widespread blackout
itermax in the system.After executing the smart market
procedure, the dispatchable loads amount is
where ω max is initial weight, ω min is final weight,
determined to be 47.3 MW which is added to 161.6
itermax is maximum iteration number, and iter is MW fixed loads of the system and constitute the
current iteration number. PSO using (12)and (14) is total load of the system.If no dispatchable load is
called inertia weights approach (IWA). As considered in the smart market procedure, the fixed
generally developed, ω max and ω min are set to 0.9 loads demands are increased to1.4 times of the
and 0.4. The advantages of PSO over other initial demands to balance loads and generation. In
the normal operation of the system, the GenCo’s
algorithms is lower sensitivity to the nature of the
revenue from the sold energy is 11303 $/h and the
optimization problem objective function,
generation cost of the generators belonging to
derivative-free characteristic unlike many GenCo is 2123.4 $/h. The congestion charges are
conventional techniques, easy implementation for 566.2926 $/h which must be paid by the GenCo. So
all kinds of optimization problems, etc. In the in normal operation condition of the system, the
proposed algorithm, the system initial data, profit of GenCo is 8613.5 $/h. The maximum
generators offers, and dispatchable loads bids to loadabilitylimit of the system in the normal state is
participate in the smart market procedure are 561.6439 MW. In the normal state, if no
provided. Then, the generation quantity and price of dispatchable load is considered in the system,
each generator and the quantity and price of executing smart market results in a total profit of
dispatchable loads are obtained through the smart 5652.2 $/h. The revenue of the GenCo in this case
market procedure. In the PSO algorithm, the control is 10929$/h and the generation cost of the
variables which define the dimensions of an agent generators is 5025.5 $/h. The congestion charges
include the load shedding of each bus. It should be are decreased to a value of 251.3259 $/h. The
noted that in the load shedding procedure, after a maximum loadability limit of the system at no
component outage, dispatchable loads are cut dispatchable load decreases to 491.6439 MW. It
before the fixed loads if their LSMP (Load seems that considering dispatchable loads in normal
Shedding Marginal Pricing) is higher than the fixed state of system leads to increase in GenCo revenue
loads and vice versa. Also, the amount of load shed and profit as well as increment in congestion
in each bus is limited to 50% of the load demand in charges but a decrease in generation cost and
maximum loadability limit of system is
that bus, which implies that PDmin ,i = 0.5PD ,i . It is
n
noticeable.As contingency of G2 occurs,
also supposed that the decrement or increment of performing single objective optimization problem
the active power generation of all generators is using PSO to maximize GenCo profit ( f 1 ) results in
limited to 20% of their initial generation expressed a total profit of 1626.383 $/h. In this case, the
,i = 0.8 × PG ,i and PG ,i = 1.2 × PG ,i .
by PGmin n max n revenue of GenCo is 10998 $/h, the generation cost
is 5039.5 $/h, and the congestion charges
are174.917 $/h. The maximum loadability limit of
5. Simulation Results the system in the single objective optimization
The modified IEEE 30 bus test system shown in problem ( f 2 ) is 562.7143 MW. In the multi-
Fig. 1 is considered as the case study to evaluate the
effectiveness of the proposed method. Initial objective optimization problem, considering the
operating conditions of the system are shown in normalizing coefficients w 1 and w 2 to be 0.75 and
Table 1. The system is composed of three areas. In 0.25, the PSO algorithm integrating smart market
order to show the importance of loads in the load procedure reveals the minimum load shedding that
shedding procedure, area 1 buses are given load maximizes GenCo’s profit and social welfare as
well as maintaining system voltage stability. As the
6th Global Conference on Power Control and Optimization, Las Vegas, USA, 6-8 August 2012
LSMPs of dispatchable loads buses are high in the Table 1. Initial operating condition
case of G2 contingency and results in a significant Voltage Demand Generation
EPNS expenses, no dispatchable load is dispatched. magnitude (with dispatchable loads) power
Bus Vi PD,i QD,i PG,i(M QG,i
In the multi-objective optimization problem, the No. (pu) (MW) (MVAr) W) (MVAr)
obtained GenCo revenue is 10954 $/h. The 1 1.041 0 0 28.25 2.85
generation cost of GenCo is 5039.5 $/h and the 2 1.039 5.0448 2.952 48.78 7.57
congestion charges are 190.4377 $/h. The minimum 3 1.021 22.318 11.159 0.00 0.00
total load shed that satisfies the multi-objective 4 1.023 8.834 1.859 0.00 0.00
5 1.027 0 0 0.00 0.00
optimization problem is 59.843 MW which is 6 1.025 0 0 0.00 0.00
presented in details inTable 2. In this situation the 7 1.014 0 0 0.00 0.00
profit of GenCo is 1575.9 $/h. The obtained 8 1.020 13.948 13.948 0.00 0.00
maximum loadability limit in this situation is 9 1.037 0 0 0.00 0.00
532.458MW. As it is shown in Table 2, after the 10 1.043 6.741 2.324 0.00 0.00
11 1.037 0 0 0.00 0.00
contingency of G2, LMPs of the buses with less
12 1.023 13.018 8.718 0.00 0.00
than 0.5PDn ,i show an increase of about 12 $/MWh. 13 1.059 0 0 43.89 23.72
14 1.008 7.2069 1.859 0.00 0.00
LMPs of the buses with about 50 percent of load to 15 1.005 0 0 0.00 0.00
be shed, is however almost constant. This 16 0.991 4.068 2.092 0.00 0.00
demonstrates the effect of applying appropriate 17 0.969 10.461 6.741 0.00 0.00
optimal load shedding on holding the LMPs of 18 0.971 3.7197 1.046 0.00 0.00
buses to be constant. 19 0.956 11.042 3.952 0.00 0.00
20 0.955 2.557 0.813 0.00 0.00
21 1.05 3.390 2.169 0.00 0.00
6. Conclusions 22 1.052 0 0 25.91 17.85
Market-driven power system load shedding 23 1.028 22.318 11.159 23.04 20.85
considering voltage stability using PSO is discussed 24 1.032 10.112 7.788 0.00 0.00
in this paper. Restructured power system concerns 25 1.049 0 0 0.00 0.00
26 1.030 4.068 2.673 0.00 0.00
including social welfare maximization through cost 27 1.070 0 0 32.29 19.33
minimization of entities and customers of the power 28 1.030 0 0 0.00 0.00
system beside the voltage stability of the power 29 1.05 2.789 1.046 0.00 0.00
system are taken into consideration when a power 30 1.039 0 0 0.00 0.00
Total - 151.643 82.306 202.16 97.16
system component outage occurs. Effect of a
contingency in the power system on different
Table 2. Load curtailment in the case of G2
economic aspects of power system is analyzed. A
contingency
multi-objective function to provide intelligent-based Load data Loadshed
optimal power system load shedding is presented (no dispatchable load) ( G2 contingency)
including both the technical and economic aspects Load
Bus PD,i LMPb LMPa
shed Load %
of load shedding. The proposed method is executed No. (MW) ($/MWh)
(MW)
($/MWh)
under the smart market environment of 1 0 57.877 0 - 70.782
MATPOWER 4.1 and is applied to a modified 2 7.062 58.029 0.729 10.3 71.154
IEEE 30 bus test system with satisfactory results. 3 31.245 59.044 5.993 19.1 71.717
4 12.368 59.027 1.593 12.8 71.565
The obtained results demonstrates the capability of 5 0 58.918 0 - 71.209
the proposed method to procure an optimal load 6 0 59.352 0 - 71.303
shedding with minimum load cuts considering 7 0 59.782 0 - 71.268
GenCo maximum profit while maintaining voltage 8 19.528 59.492 6.324 32.3 71.387
9 0 60.307 0 - 71.442
stability of the system. 10 9.438 60.799 1.086 11.5 71.514
11 0 60.307 0 - 71.442
12 18.226 72.000 3.549 19.4 72.000
13 0 72.000 0 72.000
14 10.089 72.830 1.731 17.1 73.310
15 0 72.591 0 72.968
16 5.695 73.992 1.671 29.3 73.424
17 14.646 75.385 7.317 49.9 74.288
18 5.207 75.708 0.767 14.7 75.292
19 15.460 77.151 7.683 49.7 76.215
20 3.580 77.290 1.510 42.1 76.314
21 4.746 60.911 0.946 19.9 71.160
22 0 60.931 0 - 70.992
23 31.245 68.978 11.754 37.6 72.294
24 14.158 64.426 4.905 34.6 72.048
25 0 62.059 0 - 70.934
26 5.695 63.279 1.672 29.3 72.329
27 0 60.191 0 - 69.767
28 0 59.356 0 - 70.857
29 3.905 61.558 0.603 15.4 70.328
30 0 62.542 0 - 69.929
Fig. 1. Modified IEEE 30 bus test system
Total 212.3 - 59.843 28.18 -
6th Global Conference on Power Control and Optimization, Las Vegas, USA, 6-8 August 2012