Energy xxx (2014) 1e16

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

A new algorithm for combined dynamic economic emission dispatch

with security constraints
R. Arul a, *, S. Velusami a, G. Ravi b
a
Department of Electrical Engineering, Annamalai University, Annamalainagar, Tamilnadu 608 002, India
b
Department of EEE, Pondicherry Engineering College, Puducherry, India

a r t i c l e i n f o a b s t r a c t

Article history: The primary objective of CDEED (combined dynamic economic emission dispatch) problem is to
Received 13 January 2014 determine the optimal power generation schedule for the online generating units over a time horizon
Received in revised form considered and simultaneously minimizing the emission level and satisfying the generators and system
10 August 2014
constraints. The CDEED problem is bi-objective optimization problem, where generation cost and
Accepted 15 November 2014
Available online xxx
emission are considered as two competing objective functions. This bi-objective CDEED problem is
represented as a single objective optimization problem by assigning different weights for each objective
functions. The weights are varied in steps and for each variation one compromise solution are generated
Keywords:
Combined dynamic economic emission
and finally fuzzy based selection method is used to select the best compromise solution from the set of
dispatch compromise solutions obtained. In order to reflect the test systems considered as real power system
Chaotic differential harmony search model, the security constraints are also taken into account. Three new versions of DHS (differential
algorithm harmony search) algorithms have been proposed to solve the CDEED problems. The feasibility of the
Self-adaptive differential harmony search proposed algorithms is demonstrated on IEEE-26 and IEEE-39 bus systems. The result obtained by the
algorithm proposed CSADHS (chaotic self-adaptive differential harmony search) algorithm is found to be better
Chaotic self-adaptive differential harmony than EP (evolutionary programming), DHS, and the other proposed algorithms in terms of solution
search algorithm
quality, convergence speed and computation time.
Evolutionary programming
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction
The primary objective of CDEED (combined dynamic economic
emission dispatch) problem is to determine the optimal power
generation schedule for the online generating units over a time
horizon and at the same time minimizing the emission level and
satisfying the security constraints (transmission line flow limit and
load bus voltage limits), generator constraints and spinning reserve
capacity constraints simultaneously. The US Clean Air Act Amend-
ments of 1990 mandates a significant reduction of NOx and SO2
emissions from 1980 levels [1]. These environmental constraints
forced the utilities to modify their design or operational method-
ology to decrease the atmospheric emissions from the electrical
power plants. The methods available to control the emissions, are
installing emission control equipments in power plant, switching to
lower sulfur coal, fuel switching, replacing the aged fuel-burners
* Corresponding author. Tel. þ91 9865003003.
E-mail address: arulphd@yahoo.co.in (R. Arul).
with newer ones and emission dispatch. Out of these methods,
the emission dispatch method is an attractive method, in which
both the objectives (cost and emission) are minimized simulta-
neously. In the past decades, many conventional techniques such as
LP (linear programming) [2], NLP (non-linear programming) [3], QP
(quadratic programming) [3], and LR (Lagrange relaxation) [4] have
been proposed to solve the ELD (economic load dispatch) problems.
These techniques often use approximations to limit complexity of
the problem. The solution obtained from these methods is normally
inferior in solution quality since they usually get stuck at the local
optimal solution. To overcome this difficulty, in the recent past,
stochastic search optimization algorithms such as PSO (particle
swarm optimization) [5], SA (simulated annealing) [6], DE (differ-
ential evolution) [7], AIS (artificial immune system) [8], ECE
(enhanced cross-entropy method) [9], HHS (hybrid swarm intelli-
gence based harmony search) [10], EAPSO (enhanced adaptive
particle swarm optimization) [11], BCO-SQP (bee colony optimiza-
tion and sequential quadratic programming) [12], SOA-SQP (seeker
optimization algorithm-sequential quadratic programming) [13],
improved PSO method [14] and ICA (imperialist competitive

http://dx.doi.org/10.1016/j.energy.2014.11.037
0360-5442/© 2014 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
2 R. Arul et al. / Energy xxx (2014) 1e16

algorithm) [15] have been used to solve the DED (dynamic eco- metrics tests are also executed to ascertain the quality of the so-
nomic dispatch) problems without emission and security con- lution obtained in all case studies considered.
straints. NPAHS (Harmony search method with a new pitch
adjustment rule) [16], CQGA (chaotic quantum genetic algorithm)
[17], EFA (enhanced firefly algorithm) [18], SALCSSA (self-adaptive 2. Bi-objective CDEED problem
learning charged system search algorithm) [19] and FSALPSO (fuzzy
self-adaptive learning particle swarm optimization) [20] methods The bi-objective CDEED problem is represented as a single
have been used to solve DED problems with emission constraints. objective optimization problem by assigning different weights for
Vaisakh et al. [21] recently have applied bacterial foraging PSO-DE each objective functions. The mathematical expression for CDEED
algorithm to solve DED with security and spinning reserve capacity problem is given in (1) [29,30]
constraints. In the literature, no research works were done to solve
the DED problems with both emission and security constraints. So X
N
Ft ¼ ðw*F1 þ h*ð1  wÞ*F2 Þ (1)
in this paper, an attempt is made to solve the DED problems with i¼1
both emission and security constraints along with other conven-
tional constraints. Heuristic methods normally do not always The letter ‘h’ is called penalty factor or scaling factor. The scaling
guarantee the global optimal solution but they generally provide a factor is multiplied with the emission function to get an equivalent
reasonable solution, which is sub-optimal or a value near to global cost value in $/hr. The calculation of h value is given in Ref. [31]. The
optimal solution. HS (harmony search) algorithm was proposed by value of weighing factor w indicates which objective function is
Geem et al. [22] which is a new meta-heuristic optimization algo- given more importance. The problem becomes classical DED
rithm. It imitates the music improvisation process of the music problem, when w ¼ 1 that minimizes only the fuel cost. The
players to improvise the pitches of their instruments so as to obtain problem becomes pure DEmD (dynamic emission dispatch) prob-
better harmony. The uniqueness of HS algorithm is, it will consider lem when w ¼ 0, which minimizes only the emission level. In
all the existing solution vectors in the HM (harmony memory) CDEED problem, the value of w is decreased in steps from 1 to 0, for
matrix to generate a new solution vector. This property increases each w value, a compromise solution will be generated. The weight
the exploration power of the HS algorithm to produce better so- assigned for each objective functions in (1) are such that the
lutions. It is also proved that the performance of HS algorithm is summation of weights should be equal to one [32,33]. Finally fuzzy
better than GA (genetic algorithm), EP (evolutionary program- based selection method is utilized to select the BCS (best compro-
ming), IFEP (improved fast evolutionary programming), PSO (par- mise solution) from the set of compromise solutions obtained for
ticle swarm optimization) and DE (differential evolution) each variation of w value. As the value of w decreases, the fuel cost
algorithms [23e25]. It has been observed that in the complex and value increases and the emission value decreases.
multi-modal fitness landscapes the performance of HS algorithm is
inferior because it may trap in the local optima or show faulty
3. Fuzzy based selection of BCS (best compromise solution)
convergence. In 2007, Mahdavi et al. [26] proposed an improved
IHS (improved HS algorithm), in which the control parameters PAR
Choosing a best compromise solution from the obtained Pareto
(pitch adjustment rate) and bw (band width) are dynamically
optimal set is important in decision making process. The fuzzy
varied with respect to the iteration number to generate better
membership approach is used to find a best compromise solution.
harmony vector than HS algorithm. Mahammed and Mahdavi [27]
Due to imprecise nature of the decision maker's judgment the ith
in 2008 proposed GHS (global-best harmony search) algorithm, in
objective function fi of individual k is represented by a membership
this algorithm some concepts borrowed from particle swarm in-
function mki defined as follows [20,34]:
telligence are used to enhance the performance of HS algorithm.
Prithwish Chakraborty et al. [28] in 2009 proposed a new hybrid 8
>
> 1 fi  fimin
algorithm called DHS (differential harmony search) algorithm. The >
>
>
< f max  f
DED (dynamic economic dispatch) problem itself is a hard problem i
to solve, when the objective function of emission is added further to mki ¼ i
fimin < fi < fimax (2)
>
> fimax  fimin
>
>
it, then the problem becomes more complicated than before and it >
:
needs a very efficient algorithm to generate accurate solution for 0 fi  fimax
this problem. To solve this problem, three versions of DHS (differ-
ential harmony search) algorithms are proposed. They are CDHS where fimin and fimax are the minimum and maximum value of ith
(chaotic differential harmony search) algorithm, SADHS (Self- objective function among all non-dominated solutions, respectively
adaptive differential harmony search) algorithm and CSADHS for each non-dominated solution k, the normalized membership
(chaotic self-adaptive differential harmony search) algorithm. In function mk is calculated as
the CDHS algorithm, the mutation constant F value is chaotic var- PN k
iables generated using logistic map and in every generation the i¼1 mi
mk ¼ Pp PN k (3)
value of F is changed. The F value in the proposed SADHS algorithm
k¼1 i¼1 mi
is generated using random variable generation method and it is
varied using a self-adaptive mechanism. In the CSADHS algorithm, where p is the total number of non-dominated solutions. The best
the F value is generated using logistic map and self-adaptive compromise solution is having maximum value of mk.
mechanism is introduced to change the value of F during the run
of the algorithm. In order to demonstrate the effectiveness of the
proposed algorithms, they are applied on IEEE-26 bus and IEEE-39 4. Problem formulation
bus system to solve the CDEED problem with generator, security,
spinning reserve capacity constraints. The result obtained by the 4.1. Objective functions
proposed CSADHS algorithm is found to be better than EP, DHS and
the other proposed algorithms in terms of solution quality, The objective functions considered in the CDEED problem are as
convergence speed and computation time. The performance follows:

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
 R. Arul et al. / Energy xxx (2014) 1e16 3

4.1.1. Fuel cost function

The simplified cost function with valve point effects for a Pi;min  Pi;t  Pi;max for i ¼ 1; 2; 3…N (9)
generating unit is given in (4) [21]
where Pi,min and Pi,max are the minimum and maximum real power
  XT X
N
2
   of generation limit of the ith unit, respectively.
Ft ¼ Fi;t Pi;t ¼ ai þ bi Pi;t þ ci Pi;t þ jei sin fi * Pi;min  Pi;t j
t¼1 i¼1 4.2.3. Prohibited operating zones
(4) The generating unit operation is normally avoided in the pro-
hibited operating zones, since it will cause vibration in the turbine
where Ft is the total fuel cost of the entire dispatch periods, T is the shaft. Mathematically the feasible operating ranges of a generating
total number of dispatch interval. N is the number of committed unit are given as below [17] and [21]:
generating units, Pi,t is the power output of ith unit at time period ‘t’,
Fi,t (Pi,t) is the fuel cost of ith unit at the output of Pi,t. Pi,min is the 2 LL 3
Pi;min  Pi;t  Pi;1 or
minimum generation level of ith generating unit, and ai, bi, ci, ei and 6 UL  P  P LL 7
fi are the fuel cost coefficients of the ith unit with valve point Pi;t ¼ 4 Pi;k1 i;t i;k or 5 (10)
effects. Pi;ULn  Pi;t  Pi;max ; k ¼ 2; 3; …; ni
i

4.1.2. Emission function The number of prohibited operating zones in the ith generating
The total emission produced from fossil-fueled thermal stations unit is represented by ni. k is the index of the prohibited operating
can be expressed as follows [35]: zones of the ith generating unit. Pi;k LL and P UL are the lower and
i;k
upper bounds of kth prohibited operating zones of unit i.
  X Ng    
F2 ¼ Ei Pgi ¼ 102 ai þ bi Pgi þ gi Pgi
2
þ di exp li Pgi (5) 4.2.4. Ramp rate constraint
i¼1 The ramp-up and ramp-down limits are represented as follows
[7]:
where ai, bi, gi, di and li are emission curve coefficients of the ith
generator. Pit1  Pi;t  DRi ; if power generation decreases

4.2. Operational constraints and security limits

Pi;t  Pit1  URi ; if power generation increases
4.2.1. Real power balance constraints
where Pit1 is previous hour power generation of ith unit. URi and
The power generation should meet the load demand and power
DRi are the upper and down ramp rate limits of the ith unit,
loss in the system for all the time period considered and the
respectively. The modified generator operating range constraints
equation is as follows [21]:
after combining the ramp rate limits are as follows:

X
N    
Pi;t  PD;t  PL;t ¼ 0 where i ¼ 1; 2; …; N and t ¼ 1; 2; …; T max Pi;min ; Pit1  DRi  Pi;t  min Pi;max ; Pit1 þ URi
i¼1
(11)
(6)

where PD,t is the load demand and PL,t is the real power loss at time such that
interval ‘t’. The power loss is calculated using the power flow
 
Equation (8).
Pit;min ¼ max Pi;min ; Pit1  DRi &
  (12)
Pit;max ¼ min Pi;max ; Pit1 þ URi
gðjVj; qÞ ¼ 0 (7)

where
8 Sp 4.2.5. Security constraints
< Pi;cal ðjVj; qÞ  Pi
>
gðjVj; qÞ ¼ Qi;cal ðjVj; qÞ  QiSp (8) The voltage at load buses and real power flow in the trans-
>
: Sp mission lines should be within the limits specified for secure
Pn;cal ðjVj; qÞ  Pn operation [21].

where Pi,cal and Pn,cal are the calculated real power at ith PQ bus and
nth PV bus, respectively. Qi,cal and QiSp are the calculated and LFkt  LFkmax ; k ¼ 1; …; NL (13)
specified reactive power at ith PQ bus.PnSp is specified real power at
nth PV bus. PiSp is the specified real power at the PQ bus i. jVj and q
Vi;min  Vi  Vi;max i ¼ 1; 2; 3…NB (14)
are voltage magnitude and phase angles of different buses.

4.2.2. Generation capacity constraint
The real power output of all the committed generating units are
limited by its lower and upper bounds for stable operation under all
the time intervals considered [17].
where Vi,min and Vi,max are the minimum and maximum limits for
the voltage at ith PQ bus. The total number of PQ bus is denoted by
NB. Total number of transmission lines are represented by NL. LFkt is
apparent power flow in the line k at time t in MVA. LFkmax is the
maximum rating of transmission line k.

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
4 R. Arul et al. / Energy xxx (2014) 1e16

4.2.6. Spinning reserve constraint 6. Chaotic variables using logistic map

The spinning reserve constraint equations are given in (15) and
(16) [21] The chaotic variables are deterministic, non-periodic, non-
converging and bounded. Generation of chaotic variables using
X
N logistic map is a very simple method. The generation of chaotic
SRi;t  SRDt (15) variable depends on the initial condition and parameters. The
i¼1 heuristic optimization algorithm may sometime trap at the local
optima. Chaotic variable method can be combined with the heu-
   ristic algorithm to increase the exploration power of the heuristic
SRi;t ¼ min Pi;t max  Pi;t ; SRi;max (16) algorithm. The mathematical equation for the logistic map is given
in (21) [36]:
where SRDt is the spinning reserve demand at tth interval, SRi,t is
the reserve contribution of the ith unit at tth interval. SRi,max is the
maximum reserve contribution offered by the ith unit. Pi,max is the Ynþ1 ¼ 4*Yn *ð1  Yn Þ (21)
maximum limit of power generation at ith unit and Pi,t is the real
power output of ith unit at tth interval. The initial value of chaotic variables Yn 2 (0, 1) and Yn ; {0, 0.25,
0.5, 0.75, 1}. The generation number is denoted by the letter ‘n’.
5. Constraints handling in CDEED problem

The fitness function value is calculated as follows [10] and [21]: 7. Mutation operator in DE (differential evolution) algorithm

 !  DE algorithm is a population-based heuristic algorithm which

XT X
N 
 
F1 ¼ Ft þ leq *  Pi;t  PD;t  PL;t  uses three operators; crossover, mutation and selection. The mu-
  gþ1
tation operator is used to generate a new mutant vector ðV i Þ
t¼1 i¼1
X
T X
N   X
T X
N   using the (22) [37]
þ lrr * Pi;t  P  þ lpoz * P 
rr lim i;t pz lim
t¼2 i¼1 t¼1 i¼1
 
X
T X
NB   X
T X
NL   V gþ1 ¼ xgr1 þ F xgr2  xgr3 (22)
þ lv * Vi;t  V  jLF j  LF max  i
i lim þ lLF * k k
t¼1 i¼1 t¼1 k¼1
 ! 
T  X
X N  The performance of DE depends on the value chosen for the
 
þ lsr *  SRi;t  SRDt  mutation factor (F). Normally the value of F is chosen from the
 
t¼1 i¼1 range of 0e2 to avoid stagnation in search process. The generation
(17) number is denoted by g. The indexes xr1, xr2 and xr3 are random and
mutually different vectors are taken from the current population of
size NP.
where leq, lpoz, lrr, lsr, lLF, lv and are the penalty factors corre-
sponding to equality, prohibited operating zones, ramp rate limits,
spinning reserve capacity, line flow, and voltage limits constraints.
8. Proposed CDHS, SADHS and CSADHS algorithms
The value assigned for the penalty factors are as follows:

2 3 8.1. CDHS (chaotic differential harmony search) algorithm

P t1  DRi ; if Pi;t < Pit1  DRi
6 i 7
Prr lim ¼ 4 Pit1 þ URi ; if Pi;t > Pit1 þ DRi 5 (18) The details about the HS (harmony search) algorithm and DHS
Pi;t otherwise (differential harmony search) algorithm are given in Refs. [37,28].
The F value in the proposed CDHS algorithm is generated using
(23). For every generation, the value of F is changed. The value
"   # of Ynþ1 in (23) is generated using (21). The initial starting
LL UL LL UL
min Pi;t  Pi;k ; Pi;k  Pi;t ; if Pi;k  Pi;t  Pi;k value in (21) is fixed as Y0 ¼ 0.91 (Y0 ¼ Yn). Improvisation of
Pi;t pz lim ¼
0 otherwise harmony vector in the proposed CDHS algorithm is executed
(19) using (24).

2 F ¼ FL þ Ynþ1 *ðFU  FL Þ;
Vi;min if Vi;t < Vi;min (23)
where FL and FU are the lower and upper limits of F:
Vi lim ¼ 4 Vi;max if Vi;t > Vi;max (20)
Vi;t otherwise
0 0
xnew ¼ xi þ F ðxr1  xr2 Þ (24)

0
where xi is an intermediate harmony vector generated after
Test system leq lpoz lrr lsr lLF lV considering memory consideration and random selection rules of
CDHS algorithm. xr1 and xr2 are random and different harmony
IEEE-26 bus 1000 1000 1000 500 500 500
IEEE-39 bus 1000 2000 e 500 500 500
vector selected from HM matrix. The improvisation process is
detailed in Sections 8.2 & 8.3.

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
 R. Arul et al. / Energy xxx (2014) 1e16 5

8.2. Computational steps of the proposed CDHS algorithm for
CDEED problem
The computational steps of the proposed CDHS algorithm are as
follows:
Step 1 Read data of generators cost curve, emission coefficients,
ramp rate limits, valve point effects, prohibited operating
zones, minimum and maximum generation bounds, load
demand pattern, the spinning reserve requirements, the
maximum line flow limit, maximum reserve contribution of
all the committed units and load bus voltage bounds,
optimal control parameters value of the proposed algo-
rithm, penalty factor, scaling factor and weight factor value.
The generation of design variables in the intermediate harmony
vector is as follows:
The generation of design variables in the intermediate harmony
vector is carried out by the parameter HMCR. The value of HMCR is
greater than or equal to a random number, then a value for the first
0
design variable x1;1 will be generated from the specified of HM (x11;1
HMS
to x1;1 ) by the memory consideration rule. If not, a value will be
produced using random selection rule from the possible range
(lower to upper limit) of the first decision variable. The same
method is followed to generate value for other design variables
from x12;1 to x1N;T of the intermediate harmony vector. The size of
00
new harmony vector ðxi;j new Þ is given in (27). The generation of
new harmony vector is given in Section 8.3 (Part-II). Check con-
00
straints violation for xi;j new as in Step 2.

00
h  00 00 00
  00 00 00
  00 00 00
i
xi;j new ¼ x1;1 ; x2;1 ; …; xN;1 ; x1;2 ; x2;2 ; …; xN;2 ; …; x1;T ; x2;T ; …; xN;T (27)

00
Begin w ¼ 1 to 0 in steps of 0.001. Step 5 Evaluate the fitness function value for the xi;j new using
(17).
Step 2 The harmony memory (HM) is randomly initialized ac- Step 6 HM update: This process is carried out by comparing the
00
cording to the specified bounds of each unit. fitness value of xi;j new with the worst fitness value in HM. If
00
the fitness value of xi;j new is better then replace the worst
The size of HM is given in (25).

2 h     i 3
x11;1 ; x12;1 ; …; x1N;1 ; x11;2 ; x12;2 ; …; x1N;2 ; …; x11;T ; x12;T ; …; x1N;T7
6
6 h     i 77
6 x21;1 ; x22;1 ; …; x2N;1 ; x21;2 ; x22;2 ; …; x2N;2 ; …; 2 2 2
x1;T ; x2;T ; …; xN;T 7
6 7
HM ¼ 6 7 (25)
6 « « « 7
6
4 h     i 7
5
x1;1 ; x2;1 ; …; xHMS
HMS HMS
N;1 ; xHMS HMS HMS
1;2 ; x2;2 ; …; xN;2 ; …; HMS HMS
x1;T ; x2;T ; …; xN;THMS

00
fitness value with the fitness value of xi;j new and go to Step
Here xki; j is an element of a harmony vector. It denotes the power
7. Otherwise, directly go to Step 7.
generation of ith unit at jth interval of the kth harmony vector in
Step 7 Check stopping condition: Verify whether maximum num-
HM. where i ¼ 1, 2, …, N, j ¼ 1, 2, …, T, and k ¼ 1, 2, …, HMS.
ber of iteration is reached. If it is yes, computation is
Check the equality and in-equality constraint violation. If any of
terminated and prints the results. Otherwise, go to Step 3
the design variables violates the bounds then it should be penalized
and repeat the process until the maximum number of iter-
by adding penalty constant. Evaluate the fitness function value for
ation is reached.
the harmony vectors in HM using (17).

End for weight w loop.

Step 3 As mentioned in Section 8.3, the F value is generated.
Step 4 As mentioned in Section 8.3, (Part-I), an intermediate har-
0 Step 8 The solution obtained for each value of w is called as a
mony vector ðxi;j Þ is generated using memory consider-
compromise solution. The best compromise solution is
ation and random selection rules. The size of intermediate
chosen using fuzzy logic method as detailed in Section 3.
harmony vector is given in (26).

0
h 0 0 0
  0 0 0
  0 0 0
i
xi;j ¼ x1;1 ; x2;1 ; …; xN;1 ; x1;2 ; x2;2 ; …; xN;2 ; …; x1;T ; x2;T ; …; xN;T (26)

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
6 R. Arul et al. / Energy xxx (2014) 1e16
8.3. Pseudo-code to generate F and improvisation process for the
proposed CDHS algorithm is as follows
8.4. SADHS (self-adaptive differential harmony search) algorithm
The random variable generation method (rand function) is used
to generate F value in the proposed SADHS algorithm. Self-adaptive
method is used to change the value of F with respect to a set of
fitness values generated in the previous generations. The compu-
tational sequences in the proposed SADHS algorithm are similar to
the proposed CDHS algorithm except steps 3 and 7. The steps 3 and
7 given below are used in the proposed SADHS algorithm:
Step 3 The equation given in (28) is the random variable generation
method which is used to produce F value. In Section 8.5, the
pseudo-code to generate F value is detailed.
F ¼ FL þ rand*ðFU  FL Þ;
(28)
where rand is random number between 0 and 1:
Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
Step 7 Check for stopping criterion: Ensure whether the
maximum number of generation is reached. If it is yes,
computation process is terminated and it prints the re-
sults. Otherwise, verify whether consecutive generations
(count (1)) reached the pre-specified value (fixed as 5). If
it is no, (set S ¼ 0) then go to Step 3, perform the next
generation without altering the value of F. If it is yes, then
verify whether the number of times the superior harmony
vector are produced (count (2)) reached the pre-specified
value (fixed as 2). If it is no, (set S ¼ 1, reset count (1) and
count (2)) then go to Step 3. Generate new F value for the
next generation. If it is yes, (set S ¼ 0, reset count (1) and
count (2)) then go to Step 3. Perform the next generation
without altering the value of F.
 R. Arul et al. / Energy xxx (2014) 1e16
8.5. Pseudo-code to generate F value in the proposed SADHS
algorithm is as follows
The improvisation process and generation of new harmony
vector in the proposed SADHS algorithm is similar as given in
Section 8.3.
8.6. CSADHS (chaotic self-adaptive differential harmony search)
algorithm
The proposed CSADHS algorithm uses the logistic map method
as mentioned in Section 8.1 to generate F value and self-adaptive
method as mentioned in Section 8.4 to change the F value with
respect to a set of fitness values generated in the previous gener-
ations. The computational procedures in the proposed CSADHS
algorithm are identical as that of the proposed SADHS algorithm
except Step 3. The modifications in Step 3 for the proposed CSADHS
algorithm are detailed in Section 8.7. Fig. 1 shows the flow chart for
the proposed CDHS/SADHS/CSADHS algorithm.
8.7. Pseudo-code to generate F value in the proposed CSADHS
algorithm is as follows
The improvisation process and generation of new harmony
vector in the proposed CSADHS algorithm is the same as presented
in Section 8.3.
9. Simulation results and discussions
The effectiveness of the proposed algorithms are tested on two
test systems. They are
Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
7
Test system-1: IEEE-26 bus test system with valve point effects,
ramp rate limits, prohibited operating zones, security con-
straints and spinning reserve capacity and
Test system-2: IEEE-39 bus test system with valve point effects,
ramp rate limits, security constraints and spinning reserve
capacity.
The best optimum power generation schedule obtained for 24 h
after conducting 50 trial runs is presented for the two test systems
which are considered. MatLab codes have been written for the
proposed algorithms, DHS and EP separately and executed in
1.6 GHz, 2 GB RAM, Pentium-IV, IBM personal computer for com-
parison. A detailed investigation was made by varying the algo-
rithm parameters and the obtained optimal control parameters
which are given in the description of test systems.
9.1. Test system-1 (IEEE-26 bus system)
This test system consists of 6 generators, 26 buses and 46
lines. The generators coefficient, emission coefficient, load de-
mand, value-point effects, ramp rate limits, prohibited operating
zones and security constraints data are available in
Refs. [21,38,39]. The power flow in all the lines is limited to a
maximum of 500 MVA. The spinning reserve requirement at
every dispatch interval should be greater than or equal to 5% of
the load demand. The load demand pattern-3 [21] is considered.
The optimal parameters used for simulation of EP algorithm is
Pop_no ¼ 40, Max_iter ¼ 2000, mutation operator ¼ 0.01 and
selection operator ¼ 0.08, for the DHS algorithm, HMS ¼ 40,
HMCR ¼ 0.96, FL ¼ 0.3, FU ¼ 1.2 and NI ¼ 2000, for the proposed
CDHS algorithm, HMS ¼ 40, HMCR ¼ 0.96, FL ¼ 0.3, FU ¼ 1.2,
Y0 ¼ 0.91 and NI ¼ 2000, for the proposed SADHS algorithm,
HMS ¼ 40, HMCR ¼ 0.96, FL ¼ 0.3, FU ¼ 1.2 and NI ¼ 2000 and for
the proposed CSADHS algorithm HMS ¼ 40, HMCR ¼ 0.96,
FL ¼ 0.3, FU ¼ 1.2, Y0 ¼ 0.91 and NI ¼ 2000. The highest reserve
contribution of all the committed units is fixed as SRi,max ¼ URi..
The lower and upper load voltage bounds are fixed as Vmin ¼ 0.95
and Vmax ¼ 1.1p.u, respectively. The slack bus voltage is fixed as
1.05 p.u. The DEED problem is carried out by substituting w equal
to 1 in (1). The generation schedule obtained by the proposed
CSADHS algorithm corresponding to the optimal fuel cost is
shown in Table 1. The comparative result of the optimal fuel cost
and their corresponding emission obtained using the DHS, EP
and the proposed algorithms are shown in Table 2. It is proved
from Table 2, that the fuel cost obtained by the proposed CSADHS
8 R. Arul et al. / Energy xxx (2014) 1e16

Fig. 1. Flow chart for CDEED problem using the proposed CDHS/SADHS/CSADHS algorithm.

algorithm is better than the other methods. The convergence
characteristic curve of the proposed CSADHS algorithm is better
compared to the EP, DHS and the other proposed algorithms
which are shown in Fig. 2. The DEmD problem is carried out by
substituting w equal to 0 in (1). The generation schedule
obtained corresponding to the best emission is given in Table 3.
The comparative result of best emission obtained by the pro-
posed algorithms, DHS and EP algorithm is presented in Table 4.
From Table 4, it is proved that the best emission obtained by the
proposed CSADHS algorithm is better compared to the other

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
 R. Arul et al. / Energy xxx (2014) 1e16 9

Table 1
Optimal generation schedule obtained using the proposed CSADHS algorithm for DEED (IEEE-26 bus system).

Hour P1 (MW) P2 (MW) P3 (MW) P4 (MW) P5 (MW) P6 (MW) PL loss Total cost ($/h)

1 447.1700 200.0000 256.9911 150.0000 164.2397 50.0000 8.4008 15398.8602

2 445.8634 164.2397 256.9911 142.7395 164.2397 50.0000 4.0734 14808.5225
3 425.8634 164.2397 256.9911 95.1353 164.2397 50.0000 6.4692 13932.4528
4 388.2195 164.2397 256.9911 98.5546 126.1598 50.0000 4.1647 13011.3302
5 445.8634 164.2397 256.9911 75.4035 164.2397 50.0000 6.7374 13934.8903
6 388.2195 164.2347 252.4743 120.0000 164.2397 50.0000 4.1682 13705.9778
7 388.2195 164.2397 256.9911 120.3049 164.2397 50.0000 3.9940 13745.3169
8 388.2195 164.2397 256.9911 120.5012 126.1598 50.0000 6.1113 13260.4774
9 388.2195 164.2397 256.9911 78.0492 126.1598 50.0000 3.6593 12729.5505
10 444.5203 126.1598 256.9911 50.0000 126.1598 50.0000 3.8310 12607.0297
11 388.2195 164.2397 256.9911 58.4475 126.1598 50.0000 4.0576 12499.8082
12 385.4912 164.2397 210.0000 50.0000 164.2397 50.0000 3.9707 12251.0841
13 388.2195 126.1598 256.9911 56.6881 126.1598 50.0000 4.2183 11986.5474
14 330.5756 126.1598 252.8984 50.0000 164.2397 50.0000 3.8735 11641.5344
15 388.2195 126.1598 256.9911 56.7424 126.1598 50.0000 4.2726 11987.4051
16 388.2195 164.2397 255.3755 50.0000 126.1598 50.0000 3.9945 12351.2508
17 444.7083 126.1598 256.9911 50.0000 126.1598 50.0000 4.0190 12608.4941
18 445.8634 164.2397 256.9911 75.9257 164.2397 50.0000 7.2596 13939.5769
19 447.8634 164.2397 256.9911 125.0423 164.2397 50.0000 8.3762 14600.4655
20 445.8634 164.2397 256.9911 141.5136 200.0000 50.0000 8.6078 15290.5337
21 445.8634 164.2397 256.9911 126.1768 200.0000 50.0000 8.2710 15081.1927
22 445.8634 164.2397 256.9911 123.3037 164.2397 50.0000 4.6376 14537.7764
23 445.8634 164.2397 256.9911 75.2296 164.2397 50.0000 6.5635 13933.3116
24 388.2195 126.1598 256.9911 120.9447 164.2397 50.0000 6.5548 13270.7444

methods. The convergence characteristic curve corresponding to
the best emission dispatch of the proposed CSADHS algorithm is
better than the other proposed algorithms, EP and DHS algo-
rithms and it is shown in Fig. 3. The CDEED problem, the w value
is decreased from 1 to 0 in steps of 0.001, for each w value one
compromise solution is generated. If the value of w is decreased
the fuel cost value will increase and the emission value will
decrease in each step simultaneously. Finally, the fuzzy logic
method detailed in Section 3 is used to select the best compro-
mise solution (BCS) from a set of compromise solutions. The
generation schedule obtained by the proposed CSADHS algo-
rithm for the BCS obtained is given in Table 5. The comparative
result of the best compromise solution obtained by the proposed
algorithms, DHS and EP is presented in Table 6. From Table 6, it is
proved that the BCS obtained by the proposed CSADHS algorithm
is better compared to other methods. The emission-cost trade-off
curve of the proposed CSADHS algorithm to select the BCS is
shown in Fig. 4. In all the above cases, the generator and security
constraints for the optimal solution obtained are examined and
found to be well within the specified bounds. Due to lack of
space, only the voltages at various load buses corresponding to
the 1st hour of the proposed CSADHS algorithm in DEED problem
is shown in Fig. 5. The load buses voltages at the 1st hour is
chosen because the load demand at this hour is maximum. The
spinning reserve capacity curve corresponding to the proposed
CSADHS algorithm in the DEED problem is shown in Fig. 6. From
hour and PD denotes the percentage of load demand at each
hour. The transmission loss obtained by the proposed CSADHS
algorithm in all cases is less compared to other proposed algo-
rithms, DHS and EP algorithm. The performance of the proposed
CSADHS algorithm in all cases is better compared to the other
algorithms considered.
9.2. Test system-2 (IEEE-39 bus system)
In order to demonstrate the efficiency further, the proposed
algorithm is applied on IEEE-39 bus test system, which consists
of 10 units with valve point effects, ramp rate limits, spinning
reserve capacity and security constraints. The load demand
pattern-1 [21] is considered in this test system. The generator
cost coefficients, emission coefficients, ramp rate limits, mini-
mum and maximum generation bounds, the spinning reserve
5
x 10
4
3.9
3.8
Fuel Cost ($/hr)

the Fig. 6, it is proved that the spinning reserve requirements for 3.7
every hour are greater than 5% of the load demand. The term SR
in the Fig. 6, denotes spinning reserve capacity available at each 3.6 CSADHS

3.5
SADHS
Table 2 3.4
CDHS EP
Comparison of result for DEED (IEEE-26 bus system). DHS
Method Fuel cost ($/h) Emission (ton/h) PL (MW) CPU time (min) 3.3

EP 326615.8245 23085.4357 130.6271 2.956

DHS 325460.1159 23185.8153 130.3745 2.745 3.2
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Proposed CDHS 324131.5450 23205.5924 130.3072 2.742
Iteration
Proposed SADHS 323926.6223 23245.7015 130.2956 2.742
Proposed CSADHS 323120.6453 23263.1908 130.2892 2.740
Fig. 2. Convergence characteristic curve for DEED (IEEE-26 bus system).

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
10 R. Arul et al. / Energy xxx (2014) 1e16

Table 3
Optimal generation schedule obtained using the proposed CSADHS algorithm for DEmD (IEEE-26 bus system).

Hour P1 (MW) P2 (MW) P3 (MW) P4 (MW) P5 (MW) P6 (MW) PL loss Fuel cost ($/h) Emission (ton/h)

1 458.4009 200.0000 210.0000 150.0000 200.0000 50.0000 8.4009 15487.1188 288.0720

2 417.7850 200.0000 206.7560 150.0000 200.0000 50.0000 4.5410 14999.2806 200.2706
3 380.0000 200.0000 187.9051 150.0000 188.8283 50.0000 6.7334 14194.9225 141.3373
4 309.0257 200.0000 187.5929 150.0000 187.5929 50.0000 4.2115 13382.2589 104.2204
5 380.0000 200.0000 188.2111 150.0000 188.4371 50.0000 6.6482 14193.9432 141.3012
6 350.0000 200.0000 194.6731 150.0000 194.5373 50.0000 4.2104 14003.7351 129.1981
7 350.0000 200.0000 197.3503 150.0000 196.9603 50.0000 4.3106 14051.0137 132.4107
8 326.7454 200.0000 189.6383 150.0000 189.6383 50.0000 6.0220 13561.5106 113.1180
9 292.5336 200.0000 185.8408 150.0000 185.8408 50.0000 4.2152 13148.8545 96.9013
10 284.1868 200.0000 184.9815 150.0000 184.9815 50.0000 4.1498 13004.1425 93.4748
11 275.8961 200.0000 184.1331 150.0000 184.1331 50.0000 4.1623 12849.2586 90.2285
12 259.3178 200.0000 182.4149 150.0000 182.4149 50.0000 4.1476 12674.3701 84.1467
13 242.8497 200.0000 180.6216 150.0000 180.6216 50.0000 4.0929 12480.4144 78.5743
14 210.0000 200.0000 182.0087 150.0000 181.9921 50.0000 4.0008 12115.8028 71.3756
15 242.8441 200.0000 180.6210 150.0000 180.6210 50.0000 4.0861 12480.3358 78.5725
16 267.5442 200.0000 183.2740 150.0000 183.2740 50.0000 4.0922 12748.0904 87.1005
17 284.1748 200.0000 184.9803 150.0000 184.9803 50.0000 4.1354 13003.9254 93.4701
18 380.0000 200.0000 188.8530 150.0000 188.7796 50.0000 6.6326 14193.7639 141.2946
19 404.7000 200.0000 203.4178 150.0000 200.0000 50.0000 8.1178 14786.1437 180.4545
20 448.7941 200.0000 210.0000 150.0000 200.0000 50.0000 8.7941 15309.8510 262.7126
21 433.4034 200.0000 210.0000 150.0000 200.0000 50.0000 8.4034 15185.7903 230.1303
22 401.7993 200.0000 202.7181 150.0000 200.0000 50.0000 4.5174 14733.2464 176.6956
23 380.0000 200.0000 189.7543 150.0000 188.8313 50.0000 6.7396 14194.9922 141.3398
24 326.7639 200.0000 189.6406 150.0000 189.6406 50.0000 6.0451 13561.6765 113.1280

capacity, the maximum line flow limit, maximum reserve
contribution and lower and upper bounds of the load bus volt-
ages are taken from Refs. [21,40,41] for comparison purpose. The
requirement of spinning reserve capacity is greater than or equal
to 5% of the load demand at every dispatch interval. The
maximum line flow in all the transmission lines is limited to
500 MVA. The maximum reserve contribution of all the
committed units is fixed as SRi,max ¼ Pi,min. The lower and upper
bounds load bus voltages are assumed as Vmin ¼ 0.95 and
Vmax ¼ 1.1 p.u, respectively. The slack bus voltage is fixed as
1.05 p.u. The optimal parameters used for simulation of EP al-
gorithm is Pop_no ¼ 40, Max_iter ¼ 2000, mutation
operator ¼ 0.01 and selection operator ¼ 0.09, for the DHS al-
gorithm, HMS ¼ 40, HMCR ¼ 0.95, FL ¼ 0.3, FU ¼ 1.5 and
NI ¼ 2000, for the proposed CDHS algorithm, HMS ¼ 40,
HMCR ¼ 0.95, FL ¼ 0.3, FU ¼ 1.5, Y0 ¼ 0.91 and NI ¼ 2000, for the
proposed SADHS algorithm, HMS ¼ 40, HMCR ¼ 0.95, FL ¼ 0.3,
FU ¼ 1.5 and NI ¼ 2000 and for the proposed CSADHS algorithm
HMS ¼ 40, HMCR ¼ 0.95, FL ¼ 0.3, FU ¼ 1.5, Y0 ¼ 0.91 and
NI ¼ 2000. The optimal generation schedule obtained by the
proposed CSADHS algorithm for DEED problem is shown in
Table 7. The comparative result of the best fuel cost obtained by
the proposed algorithms, DHS and EP algorithm are shown in
Table 8. The convergence characteristic of the proposed algo-
rithms, EP and DHS algorithm for DEED problem are shown in
Fig. 7. From Fig. 7, it is derived that the proposed CSADHS algo-
best emission of the proposed CSADHS algorithm in DEmD
problem is shown in Table 9. The comparative result of the best
emission obtained by the proposed algorithms, DHS and EP al-
gorithm is shown in Table 10. The convergence characteristic
curve of the proposed CSADHS algorithm in DEmD problem is
fast and smooth compared to the other algorithms shown in
Fig. 8. The optimal generation schedule obtained corresponding
to the BCS of the proposed CSADHS algorithm is shown in
Table 11. The comparative result of the BCS obtained using the
proposed algorithms; DHS and EP algorithm are shown in
Table 12. The emission-cost trade-off curve of the proposed
CSADHS algorithm to fix the BCS is shown in Fig. 9. The result
obtained using the proposed CSADHS algorithm in DEED, DEmD
and CDEED problems are better compared to the result obtained
using the EP, DHS and the other proposed algorithms in terms of
solution quality, speed of convergence and computational time.
The generator and security constraints for the optimal solution
5000
4800
4600
Emission (ton/hr)

4400
rithm has the ability of finding the optimal solution quickly and
steady compared to EP, DHS and the other proposed algorithms. 4200
DHS
The optimal generation schedule obtained corresponding to the
4000
CDHS
3800
SADHS
Table 4 EP
Comparison of result for DEmD (IEEE-26 bus system). 3600 CSADHS
Method Fuel cost ($/h) Emission (ton/h) PL (MW) CPU time (min) 3400
EP 330225.1037 3382.2515 132.5201 2.957
DHS 330248.5485 3362.3732 131.6739 2.741 3200
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Proposed CDHS 330253.7531 3356.0256 131.3209 2.741
Iteration
Proposed SADHS 330258.6713 3305.3502 129.3072 2.740
Proposed CSADHS 330267.8374 3257.6517 127.2932 2.739
Fig. 3. Convergence characteristic curve for DEmD (IEEE-26 bus system).

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
 R. Arul et al. / Energy xxx (2014) 1e16 11

Table 5
Optimal generation schedule obtained using the proposed CSADHS algorithm for CDEED (IEEE-26 bus system).

Hour P1 (MW) P2 (MW) P3 (MW) P4 (MW) P5 (MW) P6 (MW) PL loss Fuel cost ($/h) Emission (ton/h)

1 466.5936 200.0000 209.7748 138.6788 199.2204 50.0000 4.2676 15534.6895 321.4244

2 428.2909 194.0355 206.2524 145.9685 200.0000 50.0000 4.5473 15029.3227 214.1750
3 411.4572 200.0000 176.6107 128.0000 190.6891 50.0000 6.7570 14121.5136 146.5467
4 387.0792 160.0392 208.6173 78.5347 200.0000 50.0000 4.2704 13090.2952 161.8120
5 390.2214 200.0000 204.4795 150.0000 162.5301 50.0000 7.2310 14038.8504 155.0872
6 343.6522 199.1183 203.5654 147.3088 195.2067 50.0000 3.8514 13959.7034 133.2268
7 336.6130 199.4212 208.3133 150.0000 200.0000 50.0000 4.3475 13941.3777 139.6421
8 332.1615 200.0000 173.8365 150.0000 200.0000 50.0000 5.9980 13463.1837 117.2311
9 380.0000 160.0000 179.2271 120.5000 174.7000 50.0000 4.4271 12843.6729 144.8582
10 387.7654 170.7643 209.2161 73.6677 162.5268 50.0000 3.9403 12704.6003 151.8772
11 388.2508 125.1197 195.8889 121.2848 163.5520 50.0000 4.0962 12572.8152 134.9298
12 386.5501 127.2480 170.7934 93.5185 195.8428 50.0000 3.9528 12361.0633 133.7726
13 328.8663 200.0000 205.2096 50.0000 170.4134 50.0000 4.4893 12149.6395 117.1869
14 393.3527 165.4571 209.6855 70.5290 121.1674 50.0000 4.1917 11709.9289 151.3777
15 390.5453 200.0000 184.1929 50.0000 129.0121 50.0000 3.7503 12119.8351 136.1414
16 388.0202 197.3032 183.2395 50.0000 165.4343 50.0000 3.9972 12500.2916 136.8505
17 389.3849 200.0000 209.6977 50.0000 155.2060 50.0000 4.2886 12721.8341 157.5602
18 380.8169 199.4581 192.6029 140.0000 194.3733 50.0000 7.2512 14202.1347 146.3563
19 401.6441 200.0000 206.5690 150.0000 200.0000 50.0000 8.2131 14759.2803 181.0492
20 455.6923 196.2263 207.6372 150.0000 199.2158 50.0000 8.7716 15419.3813 279.2410
21 467.0682 199.7862 178.3411 148.2893 199.9579 50.0000 8.4427 15302.0848 302.6854
22 395.9133 199.9996 208.6481 149.9623 200.0000 50.0000 4.5233 14675.1459 178.5001
23 387.6242 200.0000 175.5634 146.8387 196.7048 50.0000 6.7311 14100.8219 147.9675
24 334.0570 199.8731 175.4981 150.0000 196.7179 50.0000 6.1461 13498.0528 115.9253

Table 6 confirmed from the Fig. 10, that load bus voltages are within their
Comparison of result for CDEED (IEEE-26 bus system). bounds. The load bus voltages in the 1st hour are selected since
Method Fuel cost ($/h) Emission (ton/h) PL (MW) CPU time (min) load demand at this hour is maximum. The spinning reserve
EP 329943.7453 8439.6402 135.8453 2.956
capacity curve corresponding to the proposed CSADHS algorithm
DHS 326876.9617 7244.7322 130.9442 2.741 in the DEED problem is shown in Fig. 11. From the Fig. 11, it is
Proposed CDHS 326835.3160 5838.9681 130.0286 2.740 proved that the spinning reserve requirements for every hour are
Proposed SADHS 326825.1035 5325.1207 129.5138 2.739 greater than 5% of the load demand. The transmission loss ob-
Proposed CSADHS 326819.5187 4005.4249 128.4825 2.738
tained by the proposed CSADHS algorithm in all cases is less
compared to other proposed algorithms and the DHS and EP
algorithm. Further, in order to ensure that the proposed CSADHS
obtained are examined for all cases and found to be well within algorithm has the best searching ability to find best solution in
the bounds specified for all the constraints considered. Due to all test systems considered, the minimum, mean and maximum
lack of space only the 1st hour load bus voltages of the proposed value obtained for 50 trial runs are compared with other
CSADHS algorithm is shown in Fig. 10 for DEED problem. It is methods and presented in Tables 13 and 14. From Tables 13 and
14, it is shown that, in all test systems the mean value obtained
by the proposed CSADHS algorithm is very close to the best
5
x 10
3.31

3.3 1.1
CSADHS
3.29
Fuel Cost ($/hr)

3.28
Load bus voltage (p.u)

X=4005.4249, 1.05
3.27
Y=326819.5187

3.26

3.25 1

3.24

3.23
0 0.5 1 1.5 2 2.5
Emission (ton/hr) x 10
4 0.95
6 8 10 12 14 16 18 20 22 24
Load bus No
Fig. 4. Emission-cost trade-off curve obtained using the proposed CSADHS algorithm
for IEEE-26 bus system. Fig. 5. Load bus voltage at the 1st interval of the proposed CSADHS algorithm for DEED
(IEEE-26 bus system).

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
12 R. Arul et al. / Energy xxx (2014) 1e16

450 Table 8
Comparison of result for DEED (IEEE-39 bus system).
400 Method Fuel cost ($/h) Emission PL (MW) CPU time
SR/hr (ton/h) (min)
350
EP 826851.3292 38105.1257 150.5253 3.645
Spinning reserve (MW)

DHS 824363.6313 38268.2554 147.2628 3.476

300
Proposed CDHS 819817.9275 38337.4287 144.3832 3.475
Proposed SADHS 813017.3657 38362.6055 142.7427 3.474
250 Proposed CSADHS 812759.6141 38371.8289 140.4369 3.473

200
5
150 x 10
5% PD 9.2

100 9.1

50 9

8.9
0
0 5 10 15 20 25 DHS

Fuel Cost ( $/hr)

8.8
Time (hrs)
8.7
Fig. 6. Spinning reserve capacity curve obtained by the proposed CSADHS algorithm
for the DEED problem (IEEE-26 bus system). 8.6 CSADHS
8.5 EP
SADHS CDHS
solution obtained. Moreover, the proposed CSADHS algorithm 8.4
has repeated generated the best solution for more number of
8.3
times than other methods. This proves the consistency charac-
teristic of the proposed CSADHS algorithm. 8.2

8.1
0 200 400 600 800 1000 1200 1400 1600 1800 2000
10. Analysis of solution quality
Iteration
In order to analyze the solution quality further, three different Fig. 7. Convergence characteristic curve for DEED (IEEE-39bus system).
performance metrics are performed. They are convergence metric
(g), spread (D) and minimum spacing (SM). The description of these Table 15. It can be seen from Table 15, the mean and variance value
metrics is available in Ref. [42]. These metrics will support to obtained using the proposed CSADHS algorithm is smaller than
evaluating closeness to the true or reference Pareto-optimal front other algorithms, which means that for different performance
and also for evaluating diversity of solutions in the Pareto-optimal metrics the proposed CSADHS algorithm gives better convergence
front. The statistical value obtained by performing the performance consistently.
metrics test for all the algorithms, are compared and presented in

Table 7
Optimal generation schedule obtained using the proposed CSADHS algorithm for DEED (IEEE-39 bus system).

Hour P1 (MW) P2 (MW) P3 (MW) P4 (MW) P5 (MW) P6 (MW) P7 (MW) P8 (MW) P9 (MW) P10 (MW) PL (MW) Total cost ($/h) Emission (ton/h)

1 456.4968 396.7994 320.7154 300.0000 222.5997 160.0000 129.5904 47.0000 20.0000 55.0000 8.2017 52294.3790 1627.6443
2 379.8726 316.7994 262.7774 250.0000 172.7331 122.4498 129.5904 47.0000 20.0000 55.0000 6.2227 44702.8051 1616.0404
3 379.8726 309.5329 182.7774 236.8128 122.8666 122.4498 129.5904 47.0000 20.0000 55.0000 5.9025 41006.6001 1611.7226
4 379.8726 309.5329 182.5018 186.8128 73.0000 122.4498 129.5904 47.0000 20.0000 55.0000 5.7603 38727.8112 1609.4896
5 456.4968 389.5329 262.5018 180.8305 122.8666 142.7331 129.5904 47.0000 20.0000 55.0000 6.5521 45688.1832 1619.5675
6 379.8726 309.5329 188.9075 180.8305 172.7331 122.4498 129.5904 47.0000 20.0000 55.0000 5.9168 40822.5421 1609.8482
7 379.8726 309.5329 188.6535 130.8305 122.8666 122.4498 129.5904 47.0000 20.0000 55.0000 5.7963 38653.3174 1607.6401
8 379.8726 309.5329 185.1998 84.1505 73.0000 122.4498 129.5904 47.0000 20.0000 55.0000 5.7960 36454.7156 1605.4596
9 303.2484 309.5329 185.5305 60.0000 73.0000 122.4498 129.5904 47.0000 20.0000 55.0000 5.3520 33960.1759 1601.2072
10 226.6242 229.5329 191.8045 60.0000 122.8666 122.4498 129.5904 47.0000 20.0000 55.0000 4.8684 32004.8327 1595.0150
11 150.0000 222.2665 187.8924 60.0000 73.0000 160.0000 129.5904 47.0000 20.0000 55.0000 4.7493 29803.2653 1591.3179
12 150.0000 142.2665 205.5891 60.0000 73.0000 122.4498 129.5904 47.0000 20.0000 55.0000 4.8958 27797.1869 1588.6111
13 150.0000 135.0000 142.8966 60.0000 73.0000 122.4498 99.5904 47.0000 20.0000 55.0000 4.9368 25948.0807 1585.5935
14 150.0000 135.0000 73.0000 60.0000 73.0000 98.9707 93.0603 47.0000 20.0000 55.0000 5.0310 23778.4880 1582.6762
15 150.0000 135.0000 74.9453 60.0000 73.0000 57.0000 93.0603 47.0000 20.0000 55.0000 5.0056 22802.8498 1582.6117
16 150.0000 135.0000 74.9236 60.0000 73.0000 57.0000 63.0603 47.0000 20.0000 55.0000 4.9839 22456.7813 1582.4428
17 150.0000 135.0000 73.0000 60.0000 73.0000 57.0000 34.9048 47.0000 20.0000 55.0000 4.9048 22056.5468 1582.2099
18 226.6242 135.0000 75.1858 60.0000 73.0000 57.0000 56.5301 47.0000 20.0000 55.0000 5.3401 23862.8715 1585.6110
19 226.6242 135.0000 75.2827 60.0000 122.8666 107.0000 56.5301 47.0000 20.0000 55.0000 5.3036 26059.0771 1585.9534
20 303.2484 135.0000 73.0000 60.0000 172.7331 152.6857 86.5301 47.0000 20.0000 55.0000 5.1973 30338.8636 1589.6089
21 379.8726 135.0000 73.0000 60.0000 219.9548 122.4498 93.0603 47.0000 20.0000 55.0000 5.3375 32196.2387 1593.0130
22 456.4968 215.0000 84.8514 60.0000 222.5997 122.4498 123.0603 47.0000 20.0000 55.0000 6.4580 36767.7054 1600.3222
23 456.4968 295.0000 164.8514 60.0000 222.5997 156.9455 129.5904 47.0000 20.0000 55.0000 7.4838 41245.2275 1607.0525
24 456.4968 309.5329 244.8514 65.3680 222.5997 160.0000 129.5904 47.0000 20.0000 55.0000 10.4393 43328.6533 1611.1705

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
 R. Arul et al. / Energy xxx (2014) 1e16 13

Table 9
Optimal generation schedule obtained using the proposed CSADHS algorithm for DEmD (IEEE-39 bus system).

Hour P1 (MW) P2 (MW) P3 (MW) P4 (MW) P5 (MW) P6 (MW) P7 (MW) P8 (MW) P9 (MW) P10 (MW) PL (MW) Total cost ($/h) Emission (ton/h)

1 321.1461 360.3372 340.0000 300.0000 243.0000 160.0000 130.0000 120.0000 80.0000 55.0000 9.4833 54455.2669 1621.7779
2 241.1690 280.3491 260.0000 250.0000 240.1918 160.0000 100.0064 90.0000 80.0000 55.0000 6.7163 46927.4640 1609.1865
3 161.1826 200.3491 227.9564 228.7734 243.0000 160.0000 130.0000 120.0000 80.0000 55.0000 6.2615 43336.1129 1600.6437
4 170.0776 153.2387 207.2690 207.4085 223.0000 160.0000 130.0000 120.0000 79.9997 55.0000 5.9935 40869.9459 1596.5140
5 241.1329 233.2387 287.2690 257.4085 243.0000 160.0000 130.0000 120.0000 80.0000 55.0000 7.0491 47404.8800 1608.9984
6 161.1410 183.1170 237.0177 236.7992 243.0000 160.0000 130.0000 120.0000 79.9999 55.0000 6.0748 43440.9671 1600.6340
7 150.0304 153.0045 208.0899 206.8368 243.0000 159.9999 130.0000 120.0000 80.0000 55.0000 5.9615 40866.8270 1596.5127
8 150.0093 135.0000 165.9619 166.9719 243.0000 160.0000 130.0000 120.0000 79.9994 55.0000 5.9425 38166.0049 1592.4290
9 150.0001 135.0002 115.4909 116.9745 243.0000 160.0000 130.0000 120.0000 80.0000 55.0000 5.4657 35912.5322 1588.3715
10 150.0541 135.0000 73.0000 66.9745 243.0000 160.0000 123.7270 118.1889 80.0000 55.0000 4.9445 33605.1407 1584.6350
11 150.0412 135.0000 73.1599 60.0000 243.0000 149.0742 99.7596 88.1889 51.6485 55.0000 4.8723 30908.1651 1583.8745
12 150.0122 135.0272 73.0007 60.0396 227.1862 155.1458 69.7596 58.2855 21.6485 55.0000 5.1053 28775.6539 1583.3439
13 150.0367 135.0283 73.0000 60.0000 182.6628 142.6560 39.7596 47.0000 20.0000 55.0000 5.1434 26503.0735 1582.9063
14 150.0441 135.0000 73.0051 60.0000 140.8172 103.4890 20.0000 47.6855 20.1781 55.0000 5.2190 24378.1357 1582.5224
15 150.1632 135.0000 73.0000 60.0000 107.7255 97.2399 20.0428 47.0000 20.0000 55.0000 5.1714 23526.2763 1582.3881
16 150.1952 135.0000 73.0000 60.0000 90.9139 84.0496 20.0000 47.0000 20.0000 55.0000 5.1587 22955.7890 1582.2874
17 150.0226 135.0169 73.0035 60.0000 74.3143 70.3261 20.0000 47.1666 20.2877 55.0000 5.1377 22021.8776 1582.1809
18 150.0033 135.0000 73.0000 60.0000 124.3143 120.3261 20.5626 47.0000 20.4400 55.0000 5.6463 23879.7444 1582.5220
19 150.0299 135.0099 73.0091 60.0424 174.3143 160.0000 30.7194 47.0000 20.4806 55.0000 5.6056 26358.9591 1582.8905
20 150.0097 135.0279 97.9594 94.9293 224.3143 159.9997 60.7190 77.0000 50.4806 55.0000 5.4399 31419.1167 1585.9366
21 150.0120 135.0000 91.9791 92.8401 243.0000 160.0000 90.7183 106.9997 80.0000 55.0000 5.5492 34217.4636 1586.1693
22 150.0000 163.1693 171.9791 142.8401 243.0000 160.0000 120.7183 120.0000 80.0000 55.0000 6.7068 38874.9814 1592.7967
23 168.6446 206.4753 251.9791 192.8401 243.0000 160.0000 130.0000 120.0000 80.0000 55.0000 7.9391 43276.2192 1600.7250
24 183.7789 221.7496 274.9776 242.8401 243.0000 160.0000 130.0000 120.0000 80.0000 55.0000 11.3462 45195.2484 1604.9995

11. Discussions and conclusions
The dynamic economic and emission dispatch problem is an
extension of the conventional ELD problem, in which the ramp rate
limits of the generators are taken into account for the entire time
periods considered. DEED problem is formulated as a highly non
linear, constrained, bi-objective optimization problem. It is very
difficult to solve this problem using classical optimization tech-
niques that make use of derivatives and gradients, in general, are
not able to find the best optimal solution. So, in order to solve this
problem, three new versions of DHS algorithms have been pro-
posed. In order to demonstrate the effectiveness of the proposed
algorithms, they are successfully applied and tested on IEEE-26 bus
and IEEE-39 bus system to solve the DEED, DEmD and CDEED
problems. The security constraints are taken into account in all the
problems considered along with other conventional constraints.
Normally the chaotic variables are embedded with heuristic algo-
rithms to improve the effectiveness of the local search. Therefore in
the proposed CDHS algorithm, the advantage of the chaotic vari-
ables is utilized to improve the performance of algorithm and good
result has been reported compared to DHS algorithm. In the pro-
posed SADHS algorithm, self-adaptive method is employed to
change the value of mutation constant during the run of the algo-
rithm, since the solution accuracy depends on the mutation con-
stant. The merit of chaotic variables and self-adaptive method are
combined and used in the CSADHS algorithm to generate high
the optimal generation schedules obtained, the generator and se-
curity constraints are examined and found to be well within the
bounds specified for all the constraints. In all case studies, the result
obtained by the proposed CSADHS algorithm is better than the
result obtained using EP, DHS and the other proposed algorithms in
terms of solution quality, speed of convergence and CPU time. In
order to analyze the performance of the proposed CSADHS algo-
rithm, EP, DHS and the other proposed algorithms, minimum, mean
and maximum value obtained for 50 trial runs are compared. From
the comparison of result, it is shown that the mean value obtained
by the proposed CSADHS algorithm is very close to the best solution
obtained for all test systems considered. It is also noted that the
proposed CSADHS algorithm has repeated generated the best so-
lution for more number of times than the other methods. This
shows the consistency characteristic of the proposed CSADHS al-
gorithm. In addition, the solution quality was also analyzed by
conducting three different performance metrics test and in each
4
x 10
5
4.8
CSADHS
Emission (ton/hr)

4.6
quality solution. The fuzzy decision making method is applied to
SADHS
choose the best compromise solution from a set of compromise
solutions obtained by varying the value of weight w from 1 to 0. For 4.4

CDHS
4.2
Table 10 DHS EP
Comparison of result for DEmD (IEEE-39 bus system).
4
Method Fuel cost ($/h) Emission (ton/h) PL (MW) CPU time (min)

EP 847032.7524 38228.6353 151.2517 3.718

DHS 847229.6669 38215.6943 148.2612 3.476 3.8
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Proposed CDHS 847251.2314 38215.4210 148.0573 3.476
Iteration
Proposed SADHS 847267.0531 38215.3572 147.9952 3.475
Proposed CSADHS 847275.8457 38215.2457 147.9338 3.474
Fig. 8. Convergence characteristic curve for DEmD (IEEE-39bus system).

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
14 R. Arul et al. / Energy xxx (2014) 1e16

Table 11
Optimal generation schedule obtained using the proposed CSADHS algorithm for CDEED (IEEE-39 bus system).

Hour P1 (MW) P2 (MW) P3 (MW) P4 (MW) P5 (MW) P6 (MW) P7 (MW) P8 (MW) P9 (MW) P10 (MW) PL (MW) Total cost ($/h) Emission (ton/h)

1 549.0627 388.3921 339.9698 110.0000 242.9905 159.8654 130.0000 77.0000 50.0537 55.0000 2.3342 53162.0799 1624.8407
2 469.8948 308.3921 302.1224 60.0000 192.9905 118.5407 130.0000 85.2387 30.6702 55.0000 2.8494 44646.4070 1613.9119
3 390.4402 230.2386 222.1224 60.4501 220.8570 157.0953 129.2688 115.2387 20.9091 55.0000 1.6202 41686.0672 1604.1664
4 311.6361 223.1271 165.8252 60.0000 225.6881 159.4243 130.0000 120.0000 50.9091 55.0000 1.6099 39322.9384 1598.3546
5 379.9620 303.1271 234.1350 109.8491 232.2791 159.8331 129.6653 120.0000 79.7408 55.0000 3.5915 46711.9244 1609.6241
6 303.0652 309.9715 181.9362 60.0995 231.6085 157.3970 130.0000 119.9330 52.7441 55.0000 1.7550 41279.7906 1602.3233
7 225.3124 301.3327 184.0158 60.0000 217.7342 158.2679 129.7703 119.7902 50.5470 55.0000 1.7705 39192.1338 1598.6807
8 226.2105 222.2135 185.6965 60.2986 223.4140 160.0000 129.8787 118.0484 20.5470 55.0000 1.3072 36613.2337 1595.3355
9 154.1712 142.2135 183.3997 63.8532 243.0000 160.0000 129.1418 120.0000 50.5470 55.0000 1.3264 35180.3835 1589.3082
10 152.0084 135.8434 184.4866 60.0000 221.0674 151.5776 128.9211 90.0000 22.4523 55.0000 1.3568 32372.7877 1588.4278
11 153.0368 135.0000 106.3254 62.3456 218.2630 131.2824 130.0000 87.1206 23.1365 55.0000 1.5103 30423.6540 1585.2834
12 154.0319 135.9949 73.0000 61.5004 168.2630 117.8475 130.0000 86.1357 20.0000 55.0000 1.7734 27917.0367 1583.7600
13 150.5987 135.0000 73.0000 60.0000 124.0425 67.8475 130.0000 86.5126 20.0000 55.0000 2.0013 25864.3580 1583.1982
14 151.9239 135.0000 73.0000 60.6541 75.2235 57.0000 117.9652 56.5126 20.0000 55.0000 2.2793 24073.5577 1582.8406
15 151.4035 135.1167 73.6401 60.4232 73.0000 57.0000 89.7538 47.0015 20.0000 55.0000 2.3388 22869.5543 1582.6207
16 150.0000 135.0277 74.0373 60.0000 73.3963 57.9616 59.8205 47.0000 20.1341 55.0000 2.3775 22355.8368 1582.3959
17 150.0157 135.0000 73.2510 60.0000 73.0000 58.3618 30.3639 47.0432 20.3222 55.0000 2.3578 21998.0220 1582.2018
18 150.6660 135.7811 77.5606 60.0000 123.0000 74.8874 58.6627 47.0000 20.0000 55.0000 2.5578 24063.4716 1582.8127
19 152.8126 135.0000 73.0000 60.0000 173.0000 124.8874 60.9266 47.0000 20.8035 55.0000 2.4301 25910.4521 1583.0472
20 229.2803 135.0000 82.3337 60.2269 223.0000 160.0000 90.2478 47.0000 20.0000 55.0000 2.0887 30269.4370 1587.0724
21 303.1490 135.0000 73.0000 60.0000 228.3488 160.0000 120.2478 47.0000 20.2142 55.0000 1.9598 32422.0494 1590.0154
22 378.6853 214.9424 73.0000 60.1178 233.5268 159.7687 129.7869 47.0000 50.2142 55.0000 2.0421 36958.5160 1596.7997
23 457.1350 218.4778 153.0000 60.0000 243.0000 160.0000 130.0000 76.7498 49.8368 55.0000 3.1994 41828.4135 1603.7934
24 457.8128 298.4778 173.9942 60.0000 230.0683 160.0000 129.5875 89.7061 52.1321 55.0000 6.7788 43786.2985 1608.0713

1.1
Table 12
Comparison of result for CDEED (IEEE-39 bus system).

Method Fuel cost ($/h) Emission (ton/h) PL (MW) CPU time (min)

EP 825372.6483 42754.2513 57.6539 3.734

Load bus voltage (p.u)

DHS 823866.4745 40654.8475 55.9243 3.479 1.05

Proposed CDHS 821764.7539 39765.9375 55.7543 3.473
Proposed SADHS 821375.5245 39029.6401 55.6079 3.474
Proposed CSADHS 820908.4036 38258.8860 55.2156 3.472

1
metrics tests the mean and variance obtained by the proposed
CSADHS algorithm is smaller compared to other algorithms. From
these analyses, it is proved that the proposed CSADHS algorithm
has the ability to generate quality solutions even in the presence of
non-linearity. It is suggested that an electrical power company can
0.95
15 20 25 30 35
Load bus No
5
x 10 Fig. 10. Load bus voltage at the 1st interval of the proposed CSADHS algorithm for
8.5
DEED (IEEE-39 bus system).

8.45 1800

8.4 1600

CSADHS 1400
Fuel Cost ($/hr)

SR/hr
Spinning reserve (MW)

8.35
1200
X=38258.8860
8.3
Y=820908.4036 1000

8.25 800

600
8.2
400
8.15 5% PD
200

8.1 0
3.82 3.822 3.824 3.826 3.828 3.83 3.832 3.834 3.836 3.838 0 5 10 15 20 25
Emission (ton/hr) 4 Time (hrs)
x 10
Fig. 11. Spinning reserve capacity curve obtained by the proposed CSADHS algorithm
Fig. 9. Emission-cost trade-off curve obtained using the proposed CSADHS algorithm
for the DEED problem (IEEE-39 bus system).
for IEEE-39 bus system.

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
 R. Arul et al. / Energy xxx (2014) 1e16 15

Table 13
Comparison of the fuel cost and emission for the test system-1.

Method Minimum fuel cost ($/h) Mean fuel cost ($/h) Maximum fuel cost ($/h)

EP 326615.8245 326671.8285 326732.1557

DHS 325460.1159 325491.1891 325528.3756
Proposed CDHS 324131.5450 324150.1060 324165.2279
Proposed SADHS 323926.6223 323935.5632 323945.6423
Proposed CSADHS 323120.6453 323121.8476 323122.9408

Method Minimum emission (ton/h) Mean emission (ton/h) Maximum emission (ton/h)

EP 3382.2515 3465.2166 3530.6480

DHS 3362.3732 3399.9154 3443.3906
Proposed CDHS 3356.0256 3375.3129 3394.2939
Proposed SADHS 3305.3502 3315.1955 3325.3264
Proposed CSADHS 3257.6517 3258.1677 3258.6432

Table 14
Comparison of the fuel cost and emission for the test system-2.

Method Minimum fuel cost ($/h) Mean fuel cost ($/h) Maximum fuel cost ($/h)

EP 826851.3292 826920.3366 826973.7748

DHS 824363.6313 824400.2117 824434.5005
Proposed CDHS 819817.9275 819840.8002 819864.6120
Proposed SADHS 813017.3657 813030.6239 813042.3642
Proposed CSADHS 812759.6141 812761.1932 812763.0215

Method Minimum emission (ton/h) Mean emission (ton/h) Maximum emission (ton/h)

EP 38228.6353 38289.7230 38351.6591

DHS 38215.6943 38260.2530 38299.4317
Proposed CDHS 38215.4210 38253.8735 38284.6585
Proposed SADHS 38215.3572 38230.2933 38244.7863
Proposed CSADHS 38215.2457 38216.2440 38217.2249

Table 15
Comparison of the statistical values obtained by the performance metrics test.

Performance metrics Parameters Test system-1 Test system-2

EP DHS Proposed Proposed Proposed EP DHS Proposed Proposed Proposed

CDHS SADHS CSADHS CDHS SADHS CSADHS

Convergence metric (g) Mean 0.5798 0.3965 0.3523 0.2956 0.1654 0.7857 0.4760 0.2976 0.2658 0.1264
Variance 0.6434 0.2451 0.0176 0.0145 0.0025 0.0935 0.0858 0.0697 0.0674 0.0054
Spread (D) Mean 0.9464 0.8564 0.7453 0.5656 0.3752 0.7959 0.7695 0.5960 0.5674 0.3349
Variance 0.4737 0.2476 0.0854 0.0838 0.0252 0.0883 0.0658 0.0564 0.0386 0.0135
Minimum spacing (SM) Mean 0.8975 0.7956 0.7585 0.5863 0.3743 0.8953 0.8564 0.6831 0.6684 0.2348
Variance 0.9453 0.4658 0.0785 0.0642 0.0034 0.0975 0.0749 0.0568 0.0507 0.0026

make use of the proposed CSADHS algorithm to solve security
constrained economic dispatch problems to improve the company
revenue and to reduce the emission level from their plants.
Acknowledgments
The authors sincerely acknowledge the Authorities of Anna-
malai University, Annamalainagar, Tamilnadu, India for their sup-
port, encouragement and the extensive facilities extended to carry
out this research work. The suggestions and timely help extended
by Prof. Dr. T.S. Sampath Kumar, Department of Metallurgical and
Materials Engineering, Indian Institute of technology-Madras,
Chennai, India is also acknowledged.
References
[10] Ravikumar Pandi V, Panigrahi Bijaya Ketan. Dynamic economic load dispatch
[1] Ramanathan R. Emission constrained economic dispatch. IEEE Trans Power
Syst 1994;9:1994e2000.
[2] Farag A, AI-Baiyat S, Cheng TC. Economic load dispatch multi-objective opti-
mization procedures using linear programming techniques. IEEE Trans PWRS
1995;10:731e8.
[3] Momoh JA, EI-Hawary ME, Adapa R. A review of selected optimal power flow
literature to 1993 part-I: nonlinear and quadratic programming approaches.
IEEE Trans Power Syst 1999;14:96e104.
[4] Hindi KS, Ab Ghani MR. Dynamic economic dispatch for large power system: a
Lagrangian relaxation approach. Electr Power Comp Syst 1991;13:51e6.
[5] Basu M. Particle swarm optimization based goal-attainment method for dy-
namic economic emission dispatch. Electr Power Comp Syst 2006;34:1015e25.
[6] Panigrahi CK, Chattopadhyay PK, Chakrabarti RN, Basu M. Simulated anneal-
ing technique for dynamic economic dispatch. Electr Power Comp Syst
2006;34:577e86.
[7] Balamurugan R, Subramanian S. Differential evolution-based dynamic eco-
nomic dispatch of generating units with valve-point effects. Electr Power
Comp Syst 2008;36:828e43.
[8] Basu M. Artificial immune system for dynamic economic dispatch. Electr Po-
wer Energy Syst 2011;33:131e6.
[9] Immanuel Selvakumar A. Enhanced cross-entropy method for dynamic eco-
nomic dispatch with valve-point effects. Electr Power Energy Syst 2011;33:
783e90.
using hybrid swarm intelligence based harmony search algorithm. Expert Syst
Appl 2011;38:8509e14.

Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
16 R. Arul et al. / Energy xxx (2014) 1e16
[11] Niknam T, Golestaneh F. Enhanced adaptive particle swarm optimization al-
gorithm for dynamic economic dispatch of units considering valve-point ef-
fects and ramp rates. IET Gener Transm Distrib 2012;6:424e35.
[12] Basu M. Hybridization of bee colony optimization and sequential quadratic
programming for dynamic economic dispatch. Electr Power Energy Syst
2013;44:591e6.
[13] Sivasubramani S, Swarup KS. Hybrid SOA-SQP algorithm for dynamic eco-
nomic dispatch with valve-point effect. Energy 2010;35:5031e6.
[14] Yuan Xiaohui, Su Anjun, Yuan Yanbin, Nie Hao, Wang Liang. An improved PSO
for dynamic load dispatch of generators with valve-point effects. Energy
2009;34:67e74.
[15] Mohammadi-Ivatloo Behnam, Rabiee Abbas, Soroudi Alireza, Ehsan Mehdi.
Imperialist competitive algorithm for solving non-convex dynamic economic
power dispatch. Energy 2012;44:228e40.
[16] Niu Qun, Zhang Hongyun, Li Kang, Irwin George W. An efficient harmony
search with new pitch adjustment for dynamic economic dispatch. Energy
2014;65:25e43.
[17] Lio Gwo-Ching. A novel evolutionary algorithm for dynamic economic
dispatch with energy saving and emission reduction in power system inte-
grated wind power. Energy 2011;36:1018e29.
[18] Niknam Taher, Azizipanah-Abarghooee Rasoul, Roosta Abireza, Amiri Babak.
A new multi-objective reserve constrained combined heat and power dy-
namic economic emission dispatch. Energy 2012;42:530e45.
[19] Bahmani-Firouzi Bahman, Farjah Ebrahim, Seifi Alireza. A new algorithm for
combined heat and power dynamic economic dispatch considering valve-
point effects. Energy 2013;52:320e32.
[20] Bahmani-Firouzi Bahman, Farjah Ebrahim, Azizipanah-Abarghooee Rasoul. An
efficient scenario-based and fuzzy self-adaptive learning particle swarm
optimization approach for dynamic economic emission dispatch considering
load and wind power uncertainties. Energy 2013;50:232e44.
[21] Vaisakh K, Praveena P, Rama Mohana Rao S, Meah Kala. Solving dynamic
economic dispatch problem with security constraints using bacterial foraging
PSO-DE algorithm. Elect Power Energy Syst 2012;39:56e67.
[22] Geem ZW, Kim JH, Loganathan GV. A new heuristic optimization algorithm:
harmony search. Simulation 2001;76:60e8.
[23] Khorram Esmaile, Jaberipour Majid. Harmony search algorithm for solving
combined heat and power economic dispatch problems. Energy Convers
Manag 2011;52:1550e4.
[24] Khazali AH, Kalantar M. Optimal reactive power dispatch based on harmony
search algorithm. Elect Power Energy Syst 2011;33:684e92.
[25] Arul R, Ravi G, Velusami S. Non-convex economic dispatch with heuristic load
patterns, valve point loading effect, prohibited operating zones, ramp-rate
limits and spinning reserve constrains using harmony search algorithm.
Electr Eng 2013;95:53e61.
[26] Mahdavi M, Fesanghary M, Damangir E. An improved harmony search algorithm
for solving optimization problems. Appl Math Comput 2007;188:1567e79.
Please cite this article in press as: Arul R, et al., A new algorithm for combined dynamic economic emission dispatch with security constraints,
Energy (2014), http://dx.doi.org/10.1016/j.energy.2014.11.037
[27] Omran Mahamed GH, Mahdavi Mehrdad. Global-best harmony search. Appl
Math Comput 2008;198:643e56.
[28] Chakraborty Prithwish, Roy Gourab Ghosh, Das Swagatam, Jain Dhaval,
Abraham Ajith. An improved harmony search algorithm with differential
mutation operator. Fundam Inf 2009;95:1e26. http://dx.doi.org/10.3233/F1-
2009-181. IOS press.
[29] Kulkarni PS, Kothari AG, Kothari DP. Combined economic and emission
dispatch using improved back propagation neural network. Electr Mach Po-
wer Syst 2000;28:31e44.
[30] Lazzaretto A, Toffolo A. Energy, economy and environment as objectives in
multi-criterion optimization of thermal systems design. Energy 2004;29:
1139e57.
[31] Venkatesh P, Gnanadass R, Padhy Narayana Prasad. Comparison and appli-
cation of evolutionary programming techniques to combined economic
emission dispatch with line flow constraints. IEEE Trans Power Syst 2003;18:
688e97.
[32] Kothari DP, Dhillon JS. Power system optimization. Prentice-Hall of India;
2004. p. 321e86.
[33] Deb Kalyanmoy. Multi-objective optimization using evolutionary algorithms.
India: John Wiley & Sons, Ltd; 2001. p. 49e79.
[34] Vahidinasab V, Jadid S. Joint economic and emission dispatch in energy
markets: a multi-objective mathematical programming approach. Energy
2010;35:1497e504.
[35] Panigrahi Bk, Ravikumar Pandi V, Das Sanjoy, Das Swagatam. Multi objective
fuzzy dominance based bacterial foraging algorithm to solve economic
emission dispatch. Energy 2010;35:4761e70.
[36] Alatas B. Chaotic bee colony algorithms for global numerical optimization.
Expert Syst Appl 2010;37:5682e7.
[37] Arul R, Ravi G, Velusami S. Solving optimal power flow problems using chaotic
self-adaptive differential harmony search algorithm. Electr Power Comp Syst
2013;41:782e805.
[38] Gaing Zwe-Lee. Constrained dynamic economic dispatch solution using par-
ticle swarm optimization. In: IEEE Power Eng Soc Gen Meet, vol. 1; 2004.
p. 153e8.
[39] Guerrero RP. Differential evolution based power dispatch algorithms. M.S.
Thesis. Mayaguez campus: Electrical Engineering. University of Puerto Rico;
2004. p. 58e9.
[40] Aruldoss, Albert, Victoire T, Ebenezer Jeyakumar A. Reserve constrained dy-
namic dispatch of units with valve-point effects. IEEE Trans Power Syst
2005;20:1273e82.
[41] Basu M. Economic environmental dispatch using multi-objective differential
evolution. Appl Soft Comp 2011;11:2845e53.
[42] Dhanalakshmi S, Kannan S, Mahadevan K, Baskar S. Application of modified
NSGA-II algorithm to combine economic and emission dispatch problem.
Electr Power Energy Syst 2011;33:992e1002.