You are on page 1of 15

Combined Dynamic Environment Economic Dispatch

using Quadratic Programming method

4th International Conference on ELECTRICAL ENGINEERING


CIMGLE 2012
Oran 10-11 NouvEMBRE 2012

Authors

R. Belhachem, F. Benhamida , S. Souag, M. Ali-Dahmane and Y. Ramdani

1
Abstract
The objective of this paper is to solve the Combined Dynamic
Economic/Emission Dispatch CDEED problem for different dispatch
intervals with practical constraints using Quadratic Programming
method (QP). This problem must satisfy the following constrained:

 the system load demand,


the spinning reserve capacity,

the ramping rate limits

The emission of gas such as NOX .

2
Objective Function of DEED
The objective of DEED is to simultaneously minimize the generation
cost F(Pi) and the emission of polluting gases E(Pi) in order to meet the
load demand of a power system over some appropriate period (t) while
satisfying various constraints.

min T ( P)   FT ( P)   h  ET ( P) 
 t 1 i1 Fi ( Pi ) hi  E ( Pi ) 
T N t t t t
i

FT ( Pi max )
Where hi 
ET ( Pi max )
3
Subject to the constraints
1)Power balance

i 1 i
N t
P  D t
 Lt where Lt  N N Pt B Pt  N B Pt  B
i 1 j 1 i ij i i 1 0i i 00
2) Spinning reserve constraints


N 

i 1 
minPi
max
 Pi
t up 
, Ri 

SR t
, t  1,2,..., T

3) Generator operation constraints with ramp rate limit

max( Pimin , Pit 1  Ridown )  Pit  min( Pimax , Pit 1  Riup )

where Ridown and Riup are the ramp rate limits of generator i

4
Combined Economic Emission Dispatch
For controlling a balance between economic cost and emission
level of the pollutant, the objective function may be modified as
follows: N N
min  t  w1  Fi  h w2  Ei
i 1 i 1

T N
The general form return min T   ai'  bi' Pi t  ci' ( Pi t ) 2
t 1 i 1

where ai'  w1ai  hi w2i bi'  w1bi  hi w2 i ci'  w1ci  hi w2 i

w1 , w2 : weight factors.
ai’,bi’, ci’: the combined cost &emission function coefficients.
hi : the price penalty factor in ($/kg) for unit i
5
Mapping of Emission DED in QP Model
The typical QP problem can be presented in the matrix form as

1 t x  [ P1 , P2 ,..., PN ]T
min J  X *H * X  f t * X
X 2  ai' 
subject to H  2* diag  
i 1:N  1  2 Bij Pij  B0 j
Aeq * X  beq j 1:N  
A* X  b
T
 b '1 b 'N 
LB  X  UB f  ,..., 
 1  2 B P
11 1  B01 1  2 B P
NN NN  B0N 

where  B11 ... B1N 


Aeq  [1, 1,...,1]  [ P1 , P2 ,..., PN ]   B , B00 B00 B
  01 B0 N   [ ,
P1 P2
,..., 00 ]
PN
 BN1 BNN 
beq  Dt  2* Lt lb  [ P1min , P2min ,..., PNmin ] ub  [ P1max , P2max ,..., PNmax ]

6
QP method applied to CEED Problem
The proposed method is based on the development in Taylor
series of the global objective function and we considered only the
first three terms
dFi (Pi ) 1 dFi 2 (Pi )
Fi (Pi ) ≈ Fi (Pi ) +
0
ΔPi + ΔPi 2
dPi Pi0
2 dPi Pi0

= ai ΔPi 2 + bi ΔPi
Pi  Pi  Pi 0 is the general form of the QP model
Afterwards linearization can be written as below
N
min Fi (Pi )   (ai Pi 2  bi Pi )
i 1
Subject to
 PL 
i11  P  Pi  0 Pi min  Pi 0  Pi  Pi max  Pi 0
N
and
 i  P0
i

7
Computational Procedures
To map the ED to QP in MATLAB, we propose the following
program:
for i=1:10
Pl=P'*B*P+B01*P+B00;
Aeq
=ones(1,n)+(P'*B+B01+B00/P); The “quadprog” matlab
beq=Pd+2*Pl; function formulated as:
ll=diag(1-2*B*P-B01');
A1=inv(ll)*a;
f=inv(ll)*b; x=quadprog (H, f, A, b, Aeq, beq, lb, ub)
H=2*diag(A1); % solves the the quadratic programming
P=quadprog(H,f,[],[],Aeq,beq,l,u); problem:
pln=P'*B*P+B01*P+B00; min 0.5*x'*H*x + f'*x
acu=(Pd+pln)-sum(P); % while satisfying the constraints
end A*x ≤ b
Aeq*x = beq
lb <= x <= ub

8
Data of theNumerical Example
The proposed strategy has been applied to six generating units systems.
The characteristics of the unit system are shown in Table 1.

Table 1. Six Unit Test System Characteristics


bi αi(lb/MW2 βi(lb/MWh
Unit Pimin Pimax ai($/MW2h) ci ($/h) γi(lb/h)
($/MWh) h) )
1 100 500 0.007 7 240 0.0042 0.3277 13.8593

2 50 200 0.0095 10 200 0.0042 0.3277 13.8593

3 80 300 0.009 8 220 0.0068 -0.5455 40.2669


4 50 150 0.009 11 200 0.0068 -0.5455 40.2669

5 50 200 0.008 10.5 220 0.0046 -0.5112 42.8955


6 50 120 0.0075 12 190 0.0046 -0.5112 42.8955

9
Numerical Example Results
The CEED Optimal Solution of the power system are given as in Table 2.
CEED Generation
T Pg1 Pg2 Pg3 Pg4 Pg5 Pg6
1 331.09 123.18 211.63 101.92 135.55 97.89
2 326.13 120.51 208.67 100.63 133.94 96.95
3 330.33 122.77 211.17 101.72 135.30 97.74
4 321.55 118.04 205.94 99.46 132.45 96.10
5 323.46 119.07 207.07 99.95 133.07 96.45
6 334.15 124.83 213.46 102.71 136.55 98.46
7 344.04 130.18 219.44 105.29 139.81 100.34
8 356.96 137.20 227.32 108.70 144.11 102.81
9 395.82 158.55 251.65 119.25 157.43 110.44
10 404.82 163.55 257.42 121.75 160.59 112.25
11 423.86 174.19 269.81 127.14 167.39 116.13
12 436.48 181.30 278.18 130.79 171.98 118.75
13 419.76 171.89 267.13 125.97 165.91 115.29
14 442.40 184.66 282.14 132.52 174.16 119.99
15 447.20 187.37 285.33 133.92 175.92 120.00
16 442.03 184.45 281.89 132.41 174.02 119.91
17 431.29 178.37 274.72 129.28 170.08 117.67
18 424.23 174.40 270.06 127.25 167.52 116.21
19 408.19 165.43 259.59 122.70 161.78 112.93
20 383.04 151.49 243.54 115.73 152.98 107.90
21 356.96 137.20 227.32 108.70 144.11 102.81
22 342.14 129.15 218.28 104.79 139.18 99.98
23 338.72 127.30 216.21 103.89 138.05 99.32
24 333.00 124.21 212.77 102.41 136.18 98.24

10
CED with/without emission constraints
We will compare the implementation of this algorithm in the CED
with/without emission constraints. These figures show the power
generation produced for 24 times intervals with accuracy less than 10-4.

Fig .1 Optimal dynamic dispatch without Emission Fig .2 Optimal dynamic dispatch with Emission
(w1=1,w2=0) (w1=1,w2=1)

11
Numerical Example Results
The CEED Optimal Solution of the power system are given as in Table 2.
Combined DEED (W1=1,W2=1) DED (W1=1,W2=0)
T D ∑P L Fcost Emission Accuracy × 10-3 ∑P L Fcost Emission
1 955 1001.3 46.2880 20898 1081.2 0.0526 999.2 44.1896 11798 1186.9
2 942 986.9 44.8615 20512 1052.8 0.0471 985.2 43.2378 11624 1162.1
3 953 999.1 46.0671 20839 1076.8 0.0517 997 44.0424 11771 1183.1
4 930 973.6 43.5651 20159 1027.0 0.0424 972.4 42.3705 11465 1139.5
5 935 979.1 44.1029 20305 1037.7 0.0443 977.7 42.7306 11531 1148.9
6 963 1010.2 47.1772 21139 1098.9 0.0563 1007.7 44.7495 11905 1201.0
7 989 1039.1 50.1268 21934 1157.7 0.0697 1035.5 46.4989 12254 1242.9
8 1023 1077.1 54.1215 23003 1237.3 0.0909 1071.9 48.8579 12715 1300.0
9 1126 1193.2 67.1692 26458 1498.4 0.1891 1182.5 56.4987 14140 1488.4
10 1150 1220.4 70.4125 27311 1563.6 0.2211 1208.4 58.3863 14478 1535.7
11 1201 1278.6 77.5573 29183 1707.5 0.3035 1264 62.9663 15209 1628.8
12 1235 1317.5 82.5102 30478 1807.5 0.3708 1303.2 68.1690 15718 1697.4
13 1190 1266.0 75.9873 28772 1675.8 0.2840 1251.8 61.8403 15049 1609.7
14 1251 1335.9 84.8933 31100 1855.8 0.4063 1321.8 70.8491 15962 1732.5
15 1263 1349.8 86.7684 31573 1894.7 0.5456 1336.7 73.7017 16153 1773.2
16 1250 1334.7 84.7433 31061 1852.7 0.4040 1320.6 70.6450 15947 1729.7
17 1221 1301.5 80.4525 29940 1765.9 0.3418 1286.9 65.9283 15506 1668.3
18 1202 1279.7 77.7008 29221 1710.3 0.3054 1265.1 63.0794 15223 1630.5
19 1159 1230.6 71.6484 27635 1588.4 0.2342 1218.1 59.1046 14606 1553.8
20 1092 1154.7 62.7055 25282 1408.9 0.1502 1145.9 53.8942 13664 1423.6
21 1023 1077.1 54.1215 23003 1237.3 0.0909 1071.9 48.8579 12715 1300.0
22 984 1033.6 49.5525 21779 1146.2 0.0669 1030.2 46.1588 12187 1234.8
23 975 1023.5 48.5272 21503 1125.8 0.0622 1020.5 45.5511 12066 1220.2
24 960 1006.8 46.8428 21048 1092.2 0.0549 1004.5 44.5509 11865 1196.2

Total 27467 27470 1497.9 604140 33701 27279 1306.9 325550 33987

12
Comments
We see that the production increases when we take
account the emissions of polluting gases.
That means the increment of production cost compared to
the case when the emissions are ignored. This change in
powers produced follows the quantity of gases emitted.
The plant less the pollutant is more the producer.

That means the increment of production cost compared to


the case when the emissions are ignored. This change in
powers produced follows the quantity of gases emitted. The
plant less the pollutant is more the producer.

13
Conclusion

This paper presents an optimal solution of dynamic


Thebasic
convergence and robustness of isthe QPitalgorithms
economic dispatch using quadratic programming QPcan
The advantage of this method that very
method
effectively be converged
are demonstrated through towards the solution
the application of QP toofthese
the
and taking into
optimization consideration
problem. the QP
We applied emission of gaseous
approach to the
with a six-unit
periodic system. of the optimal solutions of these
implementation
pollutants from fossil-fueled plants. Both the demand and
problems.
energy price are assumed to be periodic.

14
Thank You

You might also like