Documente Academic
Documente Profesional
Documente Cultură
(P
)
ng
=1
(1)
F
(P
) = o
+ b
+ c
2
(2)
With
F
1
= total generation cost;
Journal of Electrical and Control Engineering JECE
JECE Vol. 2 No. 2, 2012 PP. 8-14
C
2011-2012 World Academic Publishing
9
F
; b
; c
= P
Ioud
+P
Ioss
(S)
ng
=1
The loss formula used is the loss formula that developed by
Kron and adopted by Kirchmayer. It is called B-coefficient
methods [24]as shown in (4)
P
Ioss
= P
B
]
P
]
+Bu
ng
=1
P
+Bu (4)
ng
]=1
ng
=1
2) Minimum and Maximum Power Limits
Each generator should be within the limits. The equation
usually used to limit generation is [9-10, 18]:
P
,mn
P
P
,mux
(S)
The equation (5) is easy to implement in power flow, because
the equation (5) usually is represented by straight line. Actually
the limit is not exactly straight line but a curve (GCC). To
account the GCC in power flow the proposed method used the
security check algorithm based on NN (described in section
IV.A-B). The algorithm is very simple and flexible especially
for representing non linear generation operation limit near
steady state stability limit and under excitation operation area.
3) Minimum and Maximum Magnitude Voltage Limits
The magnitude voltage in each bus should be within the
limit.
|I
,mn
| |I
| |I
,mux
| (6)
To maintain the magnitude voltage within the limit, the tap-
transformer is changing according to the voltage. The reactive
power generation also regulated together with active power
generation using security check algorithm based on NN
(described in section IV.A-B).
III. OVERVIEW OF PSO
A. Particle Swarm Optimization (PSO)
PSO is a population-based optimization method developed
by Kennedy and Eberhart [14]. In the optimal power flow
problem, the optimal solution was founded by comparing the
value of objective function of all possible generation
combination by iterations. The operating cost which is
dominated by fuel cost was used as the objective function. The
value of objective function of new individual should be
compared with the old one each iteration. The lower operating
condition cost was used as local best condition for the next
iteration. Individual having the lowest cost among community
should be used as global best condition for the next iteration.
The PSO algorithm to update the generators operating point
described as follow [1, 5, 15, 25]:
X
k+1
= X
k
+I
k+1
(7)
I
k+1
= I
k
+c
1
ronJ
1
(Pbcst
-X
k
)
+c
2
ronJ
2
(0bcst
-X
k
) (8)
With:
I
k
= individu velocity i at iteration k
= weight parameter
c
1
,
c
2
= acceleration coefisien
ronJ
1
, ronJ
2
= random value between 0 and 1
X
k
= individu position i at iteration k
Pbcst
ncw
= X
ncw
+u. (p
modcI
-X
ncw
) 9)
X
ncw
=the new particles updated position;
X
ncw
=the new particles original random position;
p
modcI
=the model particles position;
u =a uniformly distributed random number in the range [0,
1);
The topology model ( p
modcI
) used in this paper is global best
model. So the updating X
ncw
each iteration (k) can be
described as (10). X
ncw
can be added anytime (at schedule
time), Once X
ncw
is added to the population, it is also need to
update each iterations using (10).
X
ncw
k+1
= X
ncw
k
+u. (0
bcst
k
-X
ncw
k
) (1u)
For initialization of adding particle (individual), the previous
best position of X
ncw
is initialized to the point of X
ncw
and its
velocity is set to zero. To ensure new particles updated
position will lie somewhere between p
modcI
onJ X
ncw
so U is
set equal to .
IV. METHODOLOGY
A. Develop Neural Network (NN) Models for Generator
Capability Curves
The proposed NN model for a generator capability curve is
very straightforward as it only has one input, one output and
one hidden layer, as shown in Fig. 1. The number of neuron in
hidden layer is constructed automaticaly by using constructive
backpropagation method [26].
The input data used in the training process are the sampling
point data along the GCC line curves provided by generator
manufactures data sheet. Since the GCC is spanned over a
plane, it has two directions x and y directions.
Journal of Electrical and Control Engineering JECE
JECE Vol. 2 No. 2, 2012 PP. 8-14
C
2011-2012 World Academic Publishing
10
Fig.1 NN model for generator capability curve
Fig. 2 The illustration of the input and output
This makes the computation extremely difficult. To
simplify the computation process, we convert all the (P, Q)
pairs into the polar coordinates, (R,) pairs as shown in Fig. 2.
Once is chosen, we only need to compute the length, R.
Therefore, will be our input for the training process and R
will be the output of the NN. The proper weighting and number
of neurons in the hidden layer are then determined to construct
the complete GCC curves. The reconstructed GCC curves are
set as the constraints in the optimal power flow. The one of
results Comparison between GCC data sheet and GCC based
on NN can be seen in Fig.3.
Fig. 3 Comparison between GCC data sheet and GCC based on NN
B. Develop a Security Check Routine.
The converged P,Q values obtained in the load flow need to
be checked against these GCC constraints. The checking
process can be accomplished in the following three steps:
1. The converged P, Q values are first converted into polar
pairs (R, ).
2. The value of can be used to determine the distance from
the origin to the GCC curve, R
ref
, as shown in Fig. 4.
3. The generator security can be checked by comparing the
value of R and R
ref
. If R Rref, the converged P, Q are
within the safety limits; otherwise, they are set to the
values converted from (R
ref
, ).
Fig. 4 Relationship between P,Q, , R and Rref
C. Overall algorithm of OPF-IPSO.
Fig.5 Flowchart of the proposed OPF-IPSO
0 200 400 600 800 1000
-600
-400
-200
0
200
400
600
Active Power(MWatt)
R
e
a
c
t
i
v
e
P
o
w
e
r
(
M
V
a
r
)
PAITON CAPABILITY CURVE
GCC data sheet
GCC based on NN
Journal of Electrical and Control Engineering JECE
JECE Vol. 2 No. 2, 2012 PP. 8-14
C
2011-2012 World Academic Publishing
11
Fig.5 shows the overall simulation flowchart of OPF-IPSO.
As input data were network impedances and loads while the
generator power will either be created randomly or regulated as
initial population. A regulated widespread initial population is
preferable to avoid local optimum solution Load flow
calculation was conducted to compute the total losses and
power generation of swing generator. Then each generator
power output will be checked using NN based security check
algorithm developed before for generator safety. If there is one
or more unsecure generators, IPSO algorithm -equation (7-10)
- will update power generator combination except the swing
generator and repeat the process until all generators are secure.
The other constraint such as system voltage level, equality and
inequality of power were also processed at this step.
The optimal solution was founded by comparing the value
of objective function of all possible generation combination by
iterations. The operating cost which is dominated by fuel cost
was used as the objective function. At each iterations the value
of objective function of new individual should be compared
with the old one and the lower operating condition cost was
used as local best condition for the next iteration. Individual
that having the lowest cost among community should be used
as global best condition for the next iteration.
V. SIMULATION AND ANALYSIS
A. Plant Data
The Plant used for simulation is the 500 kV Java-Bali
Power System with 8 generators as shown in Figure 6. The
characteristics data of generator, cost, and line impedances are
shown at Table I-II. An operating condition (load) can be seen
in Table IV. Base MVA used in this simulation is 5000MVA
TABLE I
GENERATOR DATA
Unit
Characteristic function of
generator
Production Cost
(Rp/Kwh)
Suralaya (bus 1)
6S.94P
1
2
+S9S668. uSP
1
+S16Su.21
0.138
Muara Tawar
(bus 8)
69u.98P
2
2
+2478u64.47P
2
+1u7892S72.17
1.450
Cirata (bus 10) u + 6uuu.uuP
3
+u 1.000
Saguling (bus
11)
u +SSu2. uuP
4
+u 0.917
Tanjung Jati
(bus 15)
21.88P
5
2
+197191. 76P
5
+16S6484.18
0.077
Gresik (bus 17)
1S2.1SP
6
2
+ 777148. 77P
6
+1S6u877u.9
0.378
Paiton (bus 22)
S2.19P
7
2
+S7S7u.67P
8
+822u76S.S8
0.030
Grati (bus 23)
SSS.92P
8
2
+2uu496u.6SP
8
+S16Su.21
1.067
The software used for simulation is Matlab and Newton
Raphson is chosen as load flow algorithm. The neural network
used is constructive back propagation and the IPSO used is
IPSO that developed by Marco A. et al[1]. As many as 64
populations were selected as initial population and some
particles (individuals) also were added to the population in
iteration process by IPSO algorithm.
Fig.6 Single line of 500 kv Java Bali power system
TABLE II
NETWORK DATA
From
Bus.
To
Bus
R
(pu)
X
(pu)
B
(pu)
1 2 0.0006264960000 0.0070087680000 0
1 4 0.0065132730000 0.0625763240000 0.005989820
2 5 0.0131333240000 0.1469257920000 0.003530571
3 4 0.0015131790000 0.0169283090000 0
4 5 0.0012464220000 0.0119750100000 0
4 18 0.0006941760000 0.0066692980000 0
5 7 0.0044418800000 0.0426754000000 0
5 8 0.0062116000000 0.0596780000000 0
5 11 0.0041113800000 0.0459950400000 0.004420973
6 7 0.0019736480000 0.0189618400000 0
6 8 0.0056256000000 0.0540480000000 0
8 9 0.0028220590000 0.0271129540000 0
9 10 0.0027399600000 0.0263241910000 0
10 11 0.0014747280000 0.0141684580000 0
11 12 0.0019578000000 0.0219024000000 0
12 13 0.0069909800000 0.0671659000000 0.006429135
13 14 0.0134780000000 0.1294900000000 0.012394812
14 15 0.0135339200000 0.1514073600000 0.003638261
14 16 0.0157985600000 0.1517848000000 0.003632219
14 20 0.0090361200000 0.0868146000000 0
15 16 0.0375396290000 0.3606623040000 0.008630669
16 17 0.0013946800000 0.0133994000000 0
16 23 0.0039863820000 0.0445966560000 0
18 19 0.0140560000000 0.1572480000000 0.015114437
19 20 0.0153110000000 0.1712880000000 0.016463941
20 21 0.0102910000000 0.1151280000000 0.011065927
21 22 0.0102910000000 0.1151280000000 0.011065927
22 23 0.0044358230000 0.0496246610000 0.004769846
Journal of Electrical and Control Engineering JECE
JECE Vol. 2 No. 2, 2012 PP. 8-14
C
2011-2012 World Academic Publishing
12
TABLE III
COMPARISON DATA OPF
IPSO with Rectangular Constraint
PSO with GCC
Constraint
IPSO with GCC
Constraint
MW Mvar MW Mvar MW Mvar
Suralaya (bus 1) 1414.13 1912.87 1317.15 1924.08 1317.15 1924.08
Muara tawar (bus 8) 0.00 -21.28 0.00 -19.65 0.00 -19.65
Cirata (bus 10) 800.00 391.26 800.00 391.79 800.00 391.79
Saguling (bus 11) 800.00 91.86 800.00 100.31 800.00 100.31
Tanjung jati (bus 15) 1600.00 9.89 1600.00 14.16 1600.00 14.16
Gresik (bus 17) 0.00 441.18 0.00 444.70 0.00 444.70
Paiton (bus 22) 4700.00 68.38 4800.00 67.73 4800.00 67.73
Grati (bus 23) 0.00 301.70 0.00 304.84 0.00 304.84
Total Generation 9314.13 3195.86 9317.15 3227.97 9317.15 3227.97
Total Load 9253.8 3628.8 9253.8 3628.8 9253.8 3628.8
Total Loss 60.33 63.35 63.35
Total Cost (Rp/Kwh) 2618579062.67 2616056926.55 2616056926.55
B. Result and Analysis
The result of OPF can be seen in table III. The total load
used in simulation is 9253.8 MW for active power and 3628.8
MVar for reactive power the composition load each bus can be
seen in table IV. The cost operation resulted by OPF-IPSO
with rectangular constraint is 2618579062.67 (Rp/Kwh) and
OPF-IPSO with GCC constraint is 2616056926.55 (Rp/Kwh).
Its clear that using OPF-IPSO with GCC constraint can reduce
cost operation 0.0964% or 2522136.12 (Rp/Kwh). The total
active power generation of generator Cirata (bus 10), generator
Saguling (bus 11) and generator Tanjungjati (bus 15) are the
same for both methods. Generator Muaratawar (bus 8) and
generator Gresik (bus 17) are operated as a condenser. The
generator Suralaya is slack bus. The reducing operating cost is
due to generator Paiton in OPF-IPSO with GCC constraint is
operated more marginal than OPF-IPSO with rectangular
constraint. This phenomenon can be seen in Fig. 7.
Fig.7 Comparison OPF-IPSO with rectangular constraint and GCC constraint
at the generator Paiton (bus 22)
Fig.7 shows that the maximum limit of generator Paiton has
a curvature. If the maximum limit is straight line it makes some
area is ignored (area outside rectangular inside curve). But
using the proposed security check algorithm this still can
utilize and can reduce the operating cost.
Table III Also shows that the operating cost between OPF-
IPSO with GCC constraint and OPF-PSO with GCC constraint
are the same. But the performance in getting global optimal
solution is different as shown in Fig.8.
Fig.8 Comparison of finding global optimal solution between OPF-IPSO and
OPF-PSO (wit GCC constraint)
Fig.8 shows that in this state of operation, the OPF-IPSO
only need 4 iterations to get the global optimum solution, but
the OPF-PSO need 6 iteration to get the global solution.
To check the magnitude voltage each bus. The data
generation resulted by proposed method can be run using
Newton Rhapson power flow. The results can be seen in Table
IV.
0 1000 2000 3000 4000 5000 6000
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
5000
Active Power(MWatt)
R
e
a
c
t
i
v
e
P
o
w
e
r
(
M
V
a
r
)
PAITON
GCC Constraint
OPF-IPSO GCC Const
Rectangular Constraint
OPF-IPSO Rect Const
0 2 4 6 8 10 12 14 16 18 20
2.5
3
3.5
4
4.5
5
x 10
9
Trend of finding global optimal solution
Number of iterations
C
o
s
t
O
p
e
r
a
t
i
o
n
(
R
p
/
K
w
h
)
OPF-PSO with GCC Constraint
OPF-IPSO with GCC Constraint
Journal of Electrical and Control Engineering JECE
JECE Vol. 2 No. 2, 2012 PP. 8-14
C
2011-2012 World Academic Publishing
13
TABLE IV
NEWTON RAPHSON POWER FLOW OPF-IPSO WITH GCC CONSTRAINT
Number of Iteration = 5;
Maximum error mismatch = 0.000000002
Bus Voltage Generation Load
Num
Mag.
(pu)
Angle
(deg)
Active
Power
(MW)
Reactive
Power
(Mvar)
Active
Power
(MW)
Reactive
Power
(Mvar)
1 1.02 0 1317.148 1924.08 137.7 40.5
2 1.019 -0.056 0 0 632.7 204.3
3 1.004 -0.317 0 0 684 234.9
4 1.005 -0.189 0 0 489.6 162.9
5 1.003 -0.228 0 0 627.3 193.5
6 0.999 -0.613 0 0 684 162.9
7 1 -0.579 0 0 581.4 153
8 1 -0.297 0 -19.65 0 0
9 0.999 -0.17 0 0 740.7 285.3
10 1 0.168 800 391.79 612 220.5
11 1 0.322 800 100.31 0 0
12 0.998 0.625 0 0 531 315.9
13 0.995 1.955 0 0 357.3 122.4
14 0.995 5.059 0 0 296.1 326.7
15 1 7.623 1600 14.16 0 0
16 0.999 7.112 0 0 775.8 285.3
17 1 7.077 0 444.7 189 81.9
18 1.005 -0.102 0 0 0 0
19 1.001 1.965 0 0 249.3 15.3
20 0.995 4.757 0 0 471.6 219.6
21 0.995 6.862 0 0 322.2 185.4
22 1 9.376 4800 67.73 755.1 244.8
23 1 8.148 0 304.84 117 173.7
The Total Loss is :
P = 63.348 (MW)
Q = -400.833 (MVar)
Table V is the result of Newton Rhapson power flow using
data resulted by OPF-IPSO with rectangular constrain. From
table IV and V can be seen that three are no voltages out of the
limits. The limit of voltage magnitude used in this simulation is
u.9S pu |I