Professional Documents
Culture Documents
Title: Solving combined heat and power economic dispatch problem using real
coded genetic algorithm with improved mühlenbein mutation
PII: S1359-4311(16)00027-2
DOI: http://dx.doi.org/doi: 10.1016/j.applthermaleng.2015.12.136
Reference: ATE 7550
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will
undergo copyediting, typesetting, and review of the resulting proof before it is published in its
final form. Please note that during the production process errors may be discovered which could
affect the content, and all legal disclaimers that apply to the journal pertain.
Solving combined heat and power economic dispatch
problem using real coded genetic algorithm with
improved Mühlenbein mutation
A. Haghraha, M. Nazari-Herisa, B. Mohammadi-ivatlooa,*
a
Department of Electrical Engineering, University of Tabriz, Tabriz, Iran
*Corresponding author: 109, ECE Department, University of Tabriz, 29 Bahman Blvd., Tabriz,
Iran Tel: +98-41-33393744, Fax: +98-41-33300829 (Attention: Mohammadi)
Email addresses: arslan.haghrah@gmail.com (A. Haghrah), mnazari_heris@yahoo.com (M.
Nazari-Heris), bmohammadi@tabrizu.ac.ir (B. Mohammadi-ivatloo)
Highlights
An improved Muhlenbein mutation is proposed for GA algorithm
Proposed algorithm is evaluated using different benchmark functions.
The proposed algorithm has shown better convergence and constraint handling capability.
Proposed algorithm found lower cost for CHPED problem in comparison with other
algorithms.
Abstract
The combined heat and power economic dispatch (CHPED) is a complicated optimization
problem which determines the production of heat and power units to obtain the minimum
production costs of the system, satisfying the heat and power demands and considering
operational constraints. This paper presents a real coded genetic algorithm with improved
Mühlenbein mutation (RCGA-IMM) for solving CHPED optimization task. Mühlenbein
mutation is implemented on basic RCGA for speeding up the convergence and improving the
optimization problem results. To evaluate the performance features, the proposed RCGA-IMM
procedure is employed on six benchmark functions. The effect of valve-point and transmission
losses are considered in cost function and four test systems are presented to demonstrate the
effectiveness and superiority of the proposed method. In all test cases the obtained solutions
utilizing RCGA-IMM optimization method are feasible and in most instances express a marked
improvement over the provided results by recent works in this area.
Keywords: Combined heat and power (CHP), economic dispatch, real code genetic algorithm
Page 1 of 29
(RCGA), non-convex optimization problem.
Nomenclature
C Total production cost.
N p
Number of conventional thermal units.
N h
Number of heat only units.
a j
Cost coefficient of jth co-generation unit.
d j
Cost coefficient of jth co-generation unit.
f j
Cost coefficient of jth co-generation unit.
Page 2 of 29
Ploss Transmission loss.
H d
Thermal power demand of system.
pmin
Pi Minimum power output of the ith thermal unit in MW.
pmax
Pi Maximum power output of the ith thermal unit in MW.
cmin
Pj Minimum power output of the jth co-generation unit in MW.
cmax
Pj Maximum power output of the jth co-generation unit in MW.
cmin
H j
Minimum heat output of the jth co-generation unit in MWth.
cmax
H j
Maximum heat output of the jth co-generation unit in MWth.
hmin
H k
Minimum heat output of the kth heat only unit in MWth.
hmax
H k
Maximum heat output of the kth heat only unit in MWth.
Page 3 of 29
years, in prior approaches, non-linear optimization algorithm such as dual and quadratic
programming [6], and gradient decent methods, such as Lagrangian relaxation [7] have been
employed. However non-convex fuel cost function of the generating units were not considered
for solving the problem.
In [8], differential evolution with Gaussian mutation (DEGM) is introduced for solving
CHPED problem considering valve-point loading and prohibited operating zones of conventional
thermal generators. Implementation of Gaussian mutation to DE optimization method resulted to
better search efficiency and providing the global optimal solution with high probability. The
performance of Lagrangian relaxation in solution of CHPED problem is improved in [9] by
utilization of surrogate subgradient multiplier updating procedure. An optimization method
based on benders decomposition (BD) has been employed in [10] for solving the CHPED
problem, where non-convex feasible operation region of co-generation units has been taken into
account. The CHPED problem is solved by proposing a hybrid optimization tool based on
harmony search (HS) and genetic algorithm (GA) in [11]. The authors recommended to utilize
HSGA which encompass the advantages of adaption and parallelism of GA and inferior
individuals identification of HS, in order to obtain the global optimum with high probability. The
authors utilized time varying acceleration coefficients PSO (TVAC-PSO) in [12] for the solution
of CHPED problem, considering valve-point loading, system losses and capacity limits. This
paper introduced a new large test system, considering valve-point loading and the proposed
algorithm which has capability to be applied in large systems, obtains the optimal feasible
solution. Self adaptive real-coded genetic algorithm has been employed for solving the CHPED
problem in [13], considering the optimization problem with equality and inequality constraints.
Simulated binary crossover (SBX) is applied for achieving self adaptation. In [14] HS algorithm
as a new optimization technique has been implemented for obtaining the optimal solution of the
CHPED problem. Optimal solution of CHPED problem by applying invasive weed optimization
(IWO) procedure is presented in [15]. A solution for CHPED problem in large scale power
systems has been introduced in [16] by proposing an improved group search optimization
procedure (IGSO) but the obtained results for system 4 are not feasible in which the minimum
obtained cost for this test instance is 58049.019 $.
In this paper a novel real coded genetic algorithm (RCGA) with an upgraded mutation
process is employed for solving CHPED optimization problem. Valve point effects and system
Page 4 of 29
transmission losses are taken into account for the solution of the problem. Benchmark test cases
and test systems have been utilized to prove the effectiveness of the proposed method. The
proposed RCGA-IMM has the capability for dealing with CHPED problem considering
valve-point loading effect and transmission losses. The obtained solutions for generation of
system units by implementing the proposed RCGA-IMM show feasibility and better solution in
terms of total cost, compared with reported studies in this area.
The rest of this paper is organized as follows: Section 2 represents the mathematical
formulation of the CHPED problem, in which valve point effects and transmission losses are
taken into account. Section 3 provides the brief description and basic aspects of GA and a
detailed description of the proposed RCGA-IMM. Section 4 expresses the implementation of the
proposed procedure to four test instances and provides a comparison of the obtained optimal
results with the recent researches in the area of CHPED problem. The paper conclusions are
presented in Section 5.
2. Formulation of the CHPED Problem
The CHPED is stated to obtain the minimum operation cost of heat and power units,
satisfying the heat and power demands. The objective function of the CHPED problem
considering conventional thermal units, combined heat and power units and heat-only units is
formulated as (1):
N N
p c
C ( Pi ) C
p c c
m in i j
( Pj , H j )
i =1 j =1
N
h
C
h
k
(H k
) ($/ h ) (1)
k =1
conventional thermal units, co-generation units, and heat-only units. The heat and power output
of the unit are defined by H and P, respectively. i, j and k are utilized for indicatiing the above
mentioned units. The production cost of different unit types can be stated as follows:
C i ( Pi ) = i ( Pi ) i Pi
p p 2 p
i
($ / h ) (2)
C j ( Pj , H j ) = a j ( Pj ) b j Pj c j d j ( H j )
c c c 2 c c 2
e jH f j H j P j ($/ h )
c c c
j
(3)
Page 5 of 29
) bk H c k ($ / h )
h h 2 h
Ck (H k
) = ak (H k k
(4)
Where C i Pi p
is the respective fuel cost of conventional thermal unit i for producing
p
Pi MW for 1 hour period. The cost function of conventional thermal units are modelled by
MWth heat. ak , bk , and ck are the cost coefficients of kth heat-only unit.
In order to obtain the optimal solution of the objective function (1), the following constraints
should be taken into account:
• Power production and demand balance
N N
p c
P
p c
Pi j
= Pd (5)
i =1 j =1
H
c h
H j k
= H d
(6)
j =1 k =1
j = 1, , N c (8)
( Pj ) H H
c m in c c cm ax c
H j j j
( Pj )
j = 1, , N c (9)
cmin c cmax c c
Where Pj (H j ) and Pj (H j ) which are functions of generated heat H j
represent
minimum and maximum power limits of jth CHP unit respectively. Heat generation limits are
cmin c cmax c c
identified by H j
( Pj ) and H j
( Pj ) which are functions of generated power Pj . It should
Page 6 of 29
be mentioned that there are dependency between limitations of the CHP units power production
and unit heat production plus limitations of the heat production and unit power production.
• Production limits of heat-only units
H H
h m in h hm ax
H k k k
k = 1, ,Nh (10)
2.1. Valve point impact consideration
Most of the reported studies have implemented quadratic and cubic cost function [19,
17]. When steam admission valve starts to open, because of the wire drawing impacts, a ripple is
created in the production cost. A sinusoid term has been added to the production cost of the
generation units for modeling this impact [20, 21]. Valve-point effects is utilized to express this
ripple in the production cost, which is taken into account in the proposed work, making the
optimization problem non-convex and non-differentiable. The fuel cost function with the
consideration of valve-point effects can be stated as:
C i ( Pi ) = i ( Pi ) i Pi
p p 2 p
i
| i sin ( i ( Pi Pi )) |
pmin p
(11)
In which i and i are the valve-point effects cost coefficients. The unit fuel cost by
i = 300 i = 0.035
pmin pmax
, , Pi = 0 and Pi = 680 .
2.2. Transmission loss consideration
Transmission loss is a function of power production of all units. Two approaches have
been introduced for calculating transmission loss including load flow approach [22] and Krons
loss formula which is known as B-matrix coefficient loss procedure [23]. Krons loss formula is
utilized in proposed work. The transmission loss Ploss utilizing B-coefficient formula can be
represented as follows:
N N N N
p p p c
Pi B im Pm P
p p p c
Plo s s = i
B ij P j
i =1 m =1 i =1 j =1
N N
c c
P
c c
j
B jn
Pn (12)
j =1 n =1
By consideration of transmission loss, power production and demand balance expressed in (5)
needs to be modified as follows:
Page 7 of 29
N N
p c
P = Pd Plo s s
p c
Pi j
(13)
i =1 j =1
Where range i
represents the mutation range which is normally set to 0.1( b i a i ) . The sign
k
= k
2 (15)
k=0
1
k 0 , 1 is randomly generated with P ( k
= 1) = .
16
Page 8 of 29
b
k
= k2 (16)
k = a 1
In which
C u r r e n t I te r a tio n
a = 1 (17)
M a x I te r a tio n
b = a 2
(18)
k a
and k 0 , 1 is randomly generated with P ( k
= 1) = . The minimum possible
2 2
( )
proximity in improved version of this mutation is produced with a precision of range i
.2 1 2
.
1 and 2 in the formula above are algorithm parameters and must be determined by the
implementer.
3.3. Implementation of RCGA-IMM for CHPED problem
Steps of implementing RCGA-IMM for CHPED problem and some of its intricacies are
discussed in this section.
3.3.1. Generating initial population
For generating initial population, the upper and lower limitations of all of the power and
heat production of generation units except a power only one and a heat only one should be taken
into account. The production of the excepted units will be determined considering the power and
heat demand equality constraints. In order to calculate values of these two units, already
allocated power and heat must be calculated. Excluded heat unit generation can be calculated by
using (6) as below:
N N
c h
h c h
H b
= H d
H j
H k
(19)
j =1 k =1,k b
Where b is the index of excluded heat unit. The remaining power requirement to be supported
p
by Pb , can be calculated using (13) and steps described below:
1. Ploss ,temp
= 0
N N
Pb = Pd Ploss Pi
p p c
2. , temp
p
i = 1, i b
c
j =1
Pj
3. Ploss , temp
= Ploss , current
N N
Pd Ploss Pi Pj
p c
4. if , current i =1
p c
j =1
then finish process else go to 2
Page 9 of 29
Where Ploss , temp
is a temporary variable and Ploss , current
is loss calculated for current state of the
individual.
This process for satisfying power and heat equality constraints are also applied after crossover
and mutation operators.
It must be noticed that in the case of combined heat and power units, which have
feasibility region, upper and lower limits of generated heat and power are defined virtually as
below:
m ax P = c
m ax P
H F R
max H = H
c
max
P FR
min P = P
c
min
H FR
min H = H
c
max (20)
P FR
P e n a lty Q u a d ( X i ) = k 1 ( X i X i ) k2 | X i X i | k3
d e s ir e d 2 d e s ir e d
0 if X is in th e fe a s ib le r e g io n ,
P e n a lty C o n s ta n t ( X ) =
k if X is n o t in th e fe a s ib le r e g io n .
3.3.3. Crossover
Weighted averaging of both parents parameters is used as the crossover operator in the
proposed algorithm. Corresponding weights are generated randomly in such a way that parent
Page 10 of 29
with better fitness is more likely to get higher weights. If the individual obtained from crossover
operator is more appropriate than its infirm parent (parent with weaker fitness), it will be
replaced in new generation.
The prerequisite of applying crossover operator is selection of two individuals as parents.
Selection method may affect the final solution of optimization process, so it is important how it
is implemented in the algorithm. In the proposed algorithm a sequential selection is used for
choosing parents, in which crossover operator is applied to ith index and N pop
ith index in the
N pop
1 i < (21)
2
As in generating initial population, crossover is not applied to two units that will satisfy the
shortage in power and heat demand equality constraint. Calculating excluded heat and power
units generation are as discussed previously. After finishing crossover application on population
in every generation, population is sorted considering fitness value.
3.3.4. Mutation
As the success factor of proposed algorithm, Mühlenbein mutation in an improved form
which is described formerly is implemented on conventional GA. Mutation operator is applied to
all parameters of individuals except two excepted units defined before generating initial
population. These two units will satisfy power and heat demand equalities by means of
aforementioned calculation method.
In the proposed algorithm, mutation operator is only applied to a pre-determined top percent of
individuals from the aspect of fitness. Also population is sorted after ending application of
mutation operator.
3.3.5. Pseudo-code of the algorithm
Pseudo-code of the algorithm is presented here and also the flowchart of the algorithm is
provided in Fig. 3 for clarification.
For generation =1 to Iteration num do
population size
for i =1 to do
2
Page 11 of 29
then
replace it by offspring
end if
end for
sort population by fitness
For j =1 to determined mutation num do
Select an individual from determined top percent of population and mutate
it
if m u ta te d in d iv id u a l is fitte r th a n p o p u la tio n [ p o p u la tio n s iz e ] then
replace it by mutated individual
end if
end for
sort population by fitness
end for
4. Case studies
The proposed RCGA-IMM is implemented on six benchmark functions and four test
systems. A comparison between proposed optimization method and conventional GA is done in
terms of convergence characteristics. It should be noted that the steps of the optimization process
and parameters for conventional GA and proposed RCGA-IMM are all similar, except mutation
process. Algorithm parameters used for test systems are represented in Table 1. The data
provided in this paper is achieved by independently running algorithm for 100 times. It should be
mentioned that the obtained results for CHPED are rounded up to four decimal digits.
4.1. Benchmark functions
Six benchmark functions are studied in this section in order to evaluate the performance
of the proposed RCGA-IMM algorithm. Data of benchmark functions are adopted from [27] and
are shown in Table 2. Proposed RCGA-IMM is applied to mentioned benchmark functions and
mean and standard deviation of the results are presented in Table 3.
4.2. Test system I
The first system tested, contains a power-only unit and a heat-only unit which has been
taken from [7]. The linear cost functions and capacity limits of power-only unit (unit 1) and
heat-only unit (unit 4) are shown in Eqs. (22) and (23), respectively.
Page 12 of 29
C 1 ( P1 ) = 5 0 P1 0 P1 1 5 0 M W (22)
C 4 ( H 4 ) = 2 3 .4 H 4 0 H 4 2 6 9 5 .2 M W th (23)
The system power and heat demand are 200 MW and 115 MWth, respectively. The parameters
of cost functions of CHP units are given in Table 4. Figs. 1 and 4 shows the heat-power feasible
operation regions of the cogeneration units. The performance of the proposed RCGA-IMM
method on CHPED optimization problem is validated by comparing the obtained results with
eighteen references. From Table 5, it is obvious that by implementing RCGA-IMM procedure,
total cost provided is 9257.075 which is equal with the results of recent studies.
4.3. Test system II
This test contains a conventional power unit, three cogeneration units and a heat-only
units, is proposed by [2]. Equations (24) and (25) represents the cost functions of power-only
unit (unit 1) and heat-only unit (unit 5), respectively.
C 1 ( P1 ) = 0 .0 0 0 1 1 5 P1 0 .0 0 1 7 2 P1 7 .6 9 9 7 P1 2 5 4 .8 8 6 3
3 2
35 P1 135 MW (24)
2 .0 1 0 9 H 950
2
C 5 ( H 5 ) = 0 .0 3 8 H 5 5
0 H 5 60 MWth (25)
Three different load profiles (LPs) are considered for solving the CHPED problem. The power
and heat demand for first load profile (LP1) are 300 MW and 150 MWth, respectively. For
second load profile (LP2), the power and heat demand are 250 MW and 175 MWth and third
load profile (LP3) is considered the power and heat demand 160 MW and 220 MWth,
respectively.
The provided results for test system 2 considering three different load profiles are shown in
Table 6, comparing the obtained results with the results of recent applied methods. The
convergence characteristics of RCGA-IMM and conventional GA for LP1 are presented in Fig.
5. The obtained optimal results for LP1 have been compared with the results of HS [2], GA [2],
EDHS [14], CPSO [12], TVAC-PSO [12], FA [28], IWO [15] and BD [10]. As it can be
observed in Table VI, total cost obtained for LP1 is 13660.5322 $/h and the obtained results are
feasible. The total cost reported for EDHS is 13613 $/h in which the obtained results are not
feasible, since the power output of unit 4 is out of feasible region. A comparison of the obtained
results for LP2 is done with respect to HS [2], GA [2], EDHS [14], CPSO [12], TVAC-PSO [12],
Page 13 of 29
FA [28], IWO [15] and BD [10]. The results provided for LP2 are feasible and the total cost
obtained for this profile is 12104.8682 $/h. The prepared optimal results for LP3 have been
compared with the results of HS [2], GA [2], EDHS [14], CPSO [12], TVAC-PSO [12] and BD
[10]. The obtained results for LP2 and LP3 utilizing EDHS [14] are not feasible too. The total
cost obtained for LP3 using RCGA-IMM is 11758.6349 $/h.
4.4. Test system III
A test system which consists of 7 units including 4 power-only units (units 1-4), 2 CHP
units (units 5 and 6) and a heat-only unit (unit 7), considering valve-point effects and
transmission losses is taken in to account in this section. Unit data has been taken from [29].
Table 7 contains the cost function parameters of this test instance with the feasible operation
region coordinates of CHP units. B-matrix is utilized to show the coefficients of the network
loss.
49 14 15 15 20 25
14 45 16 20 18 19
15 16 39 10 12 15 7
2B = 10 (26)
15 20 10 40 14 11
20 18 12 14 35 17
25 19 15 11 17 39
The unit of the mentioned matrix elements are 1/MW. Table 8 compares the prepared optimal
results with the results of PSO [18], EP [30], DE [20], RCGA [29], BCO [29], CPSO [12],
TVAC-PSO [12] and KH [31]. Total cost provided utilizing RCGA-IMM method is 10094.0552
$/h lower than the total cost of compared methods in which the optimum result is related to
TVAC-PSO [12] which is 10100.3164 $/h. The convergence characteristics of the proposed
method in comparison with conventional GA for this test system are depicted in Fig. 6.
4.5. Test system IV
This test system consists of 13 power only units, 6 CHP units, and 5 heat-only units. Data
for each unit of this test instance considered has been adopted from [12], which are shown in
Table 9. The power and heat demands of the system are 2350 MW and 1250 MWth,
respectively. Table 10 includes the optimal solution of this case utilizing RCGA-IMM, which is
compared with results prepared utilizing CPSO [12], TVAC-PSO [12], GSO[16], IGSO[16],
OTLBO[32] and GWO[33]. Fig. 7 presents the convergence characteristics of the proposed
method in comparison with conventional GA for this test system.
Page 14 of 29
5. Conclusion
In this paper RCGA-IMM as a meta-heuristic method is proposed for the solution of
CHPED optimization problem. Six benchmark functions and four test instances are utilized to
show the efficiency of this method and a marked improvement and feasibility in providing the
optimal solution in all cases. The proposed RCGA-IMM shows a great capability for handling
different constraints including valve-point loading, system transmission losses, capacity limits,
and heat-power dependency. The highlights of the proposed method can be categorized into two
pivotal fields: superiority of performance and ease of implementation.
Better convergence, capability of handling several constraints in non-convex and complex search
spaces and the most important, achieving feasible solutions with lesser cost function values is
observed by comprehensive simulation results. Outcomes of the simulation results, indicate the
superiority of the proposed method compared to primitive and recently developed methods. The
other advantage of the proposed algorithm is the ease of implementation as it is developed based
on the conventional and well known algorithm, GA.
References
[1] Alipour M, Mohammadi-Ivatloo B, Zare K. Stochastic risk-constrained short-term scheduling
of industrial cogeneration systems in the presence of demand response programs. Applied
Energy 2014;136:393–404.
[2] Vasebi A, Fesanghary M, Bathaee S. Combined heat and power economic dispatch by
harmony search algorithm. International Journal of Electrical Power & Energy Systems
2007;29(10):713–9.
[3] Dong L, Liu H, Riffat S. Development of small-scale and micro-scale biomass-fuelled chp
systems–a literature review. Applied thermal engineering 2009;29(11):2119–26.
[4] Wang H, Lahdelma R,Wang X, JiaoW, Zhu C, Zou P. Analysis of the location for peak
heating in chp based combined district heating systems. Applied Thermal Engineering 2015.
[5] Alipour M, Zare K, Mohammadi-Ivatloo B. Short-term scheduling of combined heat and
power generation units in the presence of demand response programs. Energy 2014;71:289–301.
[6] Rooijers FJ, van Amerongen RA. Static economic dispatch for co-generation systems. Power
Systems, IEEE Transactions on 1994;9(3):1392–8.
[7] Guo T, Henwood MI, van Ooijen M. An algorithm for combined heat and power economic
dispatch. Power Systems, IEEE Transactions on 1996;11(4):1778–84.
Page 15 of 29
[8] Jena C, Basu M, Panigrahi C. Differential evolution with gaussian mutation for combined
heat and power economic dispatch. Soft Computing 2014;1–8.
[9] Sashirekha A, Pasupuleti J, Moin N, Tan C. Combined heat and power (chp) economic
dispatch solved using lagrangian relaxation with surrogate subgradient multiplier updates.
International Journal of Electrical Power & Energy Systems 2013;44(1):421–30.
[10] Abdolmohammadi HR, Kazemi A. A benders decomposition approach for a combined heat
and power economic dispatch. Energy Conversion and Management 2013;71:21–31.
[11] Huang SH, Lin PC. A harmony-genetic based heuristic approach toward economic
dispatching combined heat and power. International Journal of Electrical Power & Energy
Systems 2013;53:482–7.
[12] Mohammadi-Ivatloo B, Moradi-Dalvand M, Rabiee A. Combined heat and power economic
dispatch problem solution using particle swarm optimization with time varying acceleration
coefficients. Electric Power Systems Research 2013;95:9–18.
[13] Subbaraj P, Rengaraj R, Salivahanan S. Enhancement of combined heat and power
economic dispatch using self adaptive real-coded genetic algorithm. Applied Energy
2009;86(6):915–21.
[14] Khorram E, Jaberipour M. Harmony search algorithm for solving combined heat and power
economic dispatch problems. Energy Conversion and Management 2011;52(2):1550–4.
[15] Jayabarathi T, Yazdani A, Ramesh V, Raghunathan T. Combined heat and power economic
dispatch problem using the invasive weed optimization algorithm. Frontiers in Energy
2014;8(1):25–30.
[16] Hagh MT, Teimourzadeh S, Alipour M, Aliasghary P. Improved group search optimization
method for solving chped in large scale power systems. Energy Conversion and Management
2014;80:446–56.
[17] Song Y, Chou C, Stonham T. Combined heat and power economic dispatch by improved ant
colony search algorithm. Electric Power Systems Research 1999;52(2):115–21.
[18] Wang L, Singh C. Stochastic combined heat and power dispatch based on multi-objective
particle swarm optimization. International Journal of Electrical Power & Energy Systems
2008;30(3):226–34.
[19] Su CT, Chiang CL. An incorporated algorithm for combined heat and power economic
dispatch. Electric Power Systems Research 2004;69(2):187–95.
Page 16 of 29
[20] Basu M. Combined heat and power economic dispatch by using differential evolution.
Electric Power Components and Systems 2010;38(8):996–1004.
[21] Mohammadi-Ivatloo B, Rabiee A, Soroudi A. Nonconvex dynamic economic power
dispatch problems solution using hybrid immune-genetic algorithm. IEEE Systems Journal
2013;7(4):777–85.
[22] Abdelaziz A, Kamh M, Mekhamer S, Badr M. A hybrid hnn-qp approach for dynamic
economic dispatch problem. Electric Power Systems Research 2008;78(10):1784–8.
[23] Victoire TAA, Jeyakumar AE. Reserve constrained dynamic dispatch of units with
valve-point effects. Power Systems, IEEE Transactions on 2005;20(3):1273–82.
[24] Lee KY, El-Sharkawi MA. Modern heuristic optimization techniques: theory and
applications to power systems; vol. 39. John Wiley & Sons; 2008.
[25] Haghrah A, Mohammadi-Ivatloo B, Seyedmonir S. Real coded genetic algorithm approach
with random transfer vectors-based mutation for short-term hydro–thermal scheduling. IET
Generation, Transmission & Distribution 2014;9(1):75–89.
[26] Herrera F, Lozano M, Verdegay JL. Tackling real-coded genetic algorithms: Operators and
tools for behavioural analysis. Artificial intelligence review 1998;12(4):265–319.
[27] He S, Wu QH, Saunders J. Group search optimizer: an optimization algorithm inspired by
animal searching behavior. Evolutionary Computation, IEEE Transactions on
2009;13(5):973–90.
[28] Yazdani A, Jayabarathi T, Ramesh V, Raghunathan T. Combined heat and power economic
dispatch problem using firefly algorithm. Frontiers in Energy 2013;7(2):133–9.
[29] Basu M. Bee colony optimization for combined heat and power economic dispatch. Expert
Systems with Applications 2011;38(11):13527–31.
[30] Wong KP, Algie C. Evolutionary programming approach for combined heat and power
dispatch. Electric Power Systems Research 2002;61(3):227–32.
[31] Adhvaryyu PK, Chattopadhyay PK, Bhattacharjya A. Application of bio-inspired krill herd
algorithm to combined heat and power economic dispatch. In: Innovative Smart Grid
Technologies-Asia (ISGT Asia), 2014 IEEE. IEEE; 2014, p. 338–43.
[32] Roy PK, Paul C, Sultana S. Oppositional teaching learning based optimization approach for
combined heat and power dispatch. International Journal of Electrical Power & Energy Systems
2014;57:392–403.
Page 17 of 29
[33] Jayakumar N, Subramanian S, Ganesan S, Elanchezhian E. Grey wolf optimization for
combined heat and power dispatch with cogeneration systems. International Journal of Electrical
Power & Energy Systems 2016;74:252–64.
[34] Song Y, Xuan Q. Combined heat and power economic dispatch using genetic algorithm
based penalty function method. Electric machines and power systems 1998;26(4):363–72.
[35] Ramesh V, Jayabarathi T, Shrivastava N, Baska A. A novel selective particle swarm
optimization approach for combined heat and power economic dispatch. Electric Power
Components and Systems 2009;37(11):1231–40.
[36] Rao PN. Combined heat and power economic dispatch: A direct solution. Electric Power
Components and Systems 2006;34(9):1043–56.
Figure Captions
• Fig. 1: The heat-power feasible regions for a combined heat and power unit (CHP unit 3 in case
I).
• Fig. 2: Illustration of unit fuel cost considering valve-point effects.
• Fig. 3: Flowchart of the algorithm.
• Fig. 4: The heat-power feasible regions for a combined heat and power unit (CHP unit 2 in case
I).
• Fig. 5: Convergence characteristics of the proposed algorithm in comparison with conventional
GA for test system 2 load profile 1.
• Fig. 6: Convergence characteristics of the proposed algorithm in comparison with conventional
GA for test system 3.
• Fig. 7: Convergence characteristics of the proposed algorithm in comparison with conventional
GA for test system 4.
Tables
Table 1: Algorithm parameters used for test systems.
Test Case Population Determined Iteration Determined 1 2
Page 18 of 29
2 3 2000 5000 1000 0.5 20 8
3 2000 4000 1000 0.5 20 8
4 3000 4000 1000 0.5 20 8
Table 2: Benchmark functions data.
Benchmark functions n Search space Global
minimum
30, 30
n
n
f1 ( x ) = (100( x i 1 x i ) ( x i 1))
2 2 2 30 0
i =1
5 .1 2 , 5 .1 2
n
n
f2 (x) = ( x i 10 cos (2 x i ) 10)
2 2 30 0
i =1
100, 100
2 n
n i 30 0
f3 ( x) = i =1 j =1
xj
1.0316285
5, 5
n
1 2
f 4 ( x ) = 4 x 1 2.1 x 1 x1 x1 x 2 4 x 2 4 x 2
2 4 6 2 4
f 5 ( x ) = max i | x i |,1 i n 30
100, 100
n
0
0
600, 600
n
1 n x i 100 30
( x i 100) 1
2 n
f6 (x) = i =1
cos
i =1
4000 i
Table 3: Comparison of different algorithm mean and standard deviation for benchmark
functions.[27]
Method f1 f2 f3 f4 f5 f6
an
Std. 3 2
361.497 0.3594 2594.9593 3 .1 3 1 4 1 0 1.5063 6 .7 5 4 5 1 0
an
Std. 3 2
32.1436 5.9400 2 .1 1 0 9 1 0 1 .2 7 8 6 1 0 0.2500 0.4434
an
Std. 2 2
30.1771 0.9509 3.6813 0 3 .9 9 8 1 1 0 3 .0 8 6 7 1 0
FEP Me 5.06 4 .6 1 0
2
1 .6 1 0
2
1.03 0.3 1 .6 1 0
2
Page 19 of 29
an
Std. 2 2 4 2
5.87 1 .2 1 0 1 .4 1 0 4 .9 1 0 0.5 2 .2 1 0
an
Std. 2 4
13.61 23.1 6 .6 1 0 4 .9 1 0 1.2 0.12
an
Std. 4 7 4 2
43.13 0.33 5 .3 1 0 6 .0 1 0 6 .5 1 0 5 .0 1 0
an
Std. 5 7
14.45 21.49 8 .5 1 0 6 .0 1 0 0.42 0.77
RCGA-I Me 8
5 .9 0 8 7 1 0 1 .2 6 2 2 1 0
31
2 .9 5 6 8 1 0
14
1.03162845 351 .3 6 1 9 1 0 41 .3 3 1 0 1 0 3
MM an
Std. 9
1 .4 8 7 0 1 0 7 .6 6 3 9 1 0
31
1 .7 1 3 7 1 0
13
1 .5 6 2 1 1 0
15 4
2 .0 6 9 3 1 0 3 .1 2 1 9 1 0
3
Table 4: Cost function parameters of the CHP units of cases I and II.
Unit a b c d e f Feasible region
coordinates
P c
,H
c
Case I
2 0.0345 14.5 2650 0.03 4.2 0.031 [98.8,0],
[81,104.8],
[215,180],
[247,0]
3 0.0435 36 1250 0.027 0.6 0.011 [44,0],
[44,15.9],
[40,75],
[110.2,135.6],
[125.8,32.4],
[125.8,0]
Case II
2 0.0435 36 1250 0.027 0.6 0.011 [44,0],
Page 20 of 29
[44,15.9],
[40,75],
[110.2,135.6],
[125.8,32.4],
[125.8,0]
3 0.1035 34.5 2650 0.025 2.203 0.051 [20,0], [10,40],
[45,55], [60,0]
4 0.072 20 1565 0.02 2.34 0.04 [35,0], [35,20],
[90,45],
[90,25],
[105,0]
Table 5: Comparison of simulation results for case I.
Method P1 P2 P3 H H 3 H TPa THb TCc
2 4
Page 21 of 29
2 8
BD[10] 0.00 160.00 40.00 40.00 75.00 0.00 200.00 115.00 9257.07
RCGA-IM 0.0000 160.0000 40.0000 40.0000 75.0000 0.0000 200.0000 115.0000 9257.0750
M
* Not feasible.
a Total power (MW).
b Total heat (MWth).
c Total cost ($).
Table 6: Comparison of simulation results for case II.
Lo Method P1 P2 P3 P4 H H 3 H H 5
TPa THb TCc
2 4
ad
LP HS[2] 134.7 48.2 16.2 100.8 81.0 23.9 6.29 38.7 300.0 150.0 13723.
1 400 000 300 500 900 200 00 000 200 000 2000
GA[2] 135.0 70.8 10.8 83.28 80.5 39.8 0.00 29.6 299.9 149.9 13779.
000 100 400 00 400 100 00 400 300 900 5000
EDHS*[ 135.0 18.1 13.0 133.7 84.0 37.7 0.00 28.1 300.0 149.9 13613.
14] 000 563 749 688 626 657 00 118 000 401 0000
CPSO[1 135.0 40.7 19.2 105.0 64.4 26.4 0.00 59.1 300.0 150.0 13692.
2] 000 309 728 000 003 119 00 955 037 076 5212
TVAC-P 135.0 41.4 18.5 105.0 73.3 37.4 0.00 39.2 300.0 150.0 13672.
SO[12] 000 019 981 000 562 295 00 143 000 000 8892
FA[28] 134.7 40.0 20.2 105.0 75.0 27.8 0.00 47.1 299.9 149.9 13683.
4 0 5 0 0 7 2 9 9 22
IWO[15] 134.7 40.0 20.8 104.4 75.0 37.6 0.00 37.4 300.0 150.0 13683.
3 0 6 1 0 0 0 65
BD[10] 135.0 40.7 19.2 105.0 73.5 36.7 0.00 39.6 300.0 150.0 13672.
000 687 313 000 957 759 00 284 000 000 83
RCGA-I 135.0000 40.7680 19.2320 105.0000 7 3.5960 36.7760 0.0000 39.6280 300.0000 150.0000 13660.5322
MM
LP HS[2] 134.6 52.9 10.1 52.23 85.6 39.7 4.18 45.4 250.0 175.0 12284.
2 700 900 100 00 900 300 00 000 000 000 4500
GA[2] 119.2 45.1 15.8 69.89 78.9 22.6 18.4 54.9 250.0 174.9 12327.
200 200 200 00 400 300 000 900 500 600 3700
Page 22 of 29
EDHS*[ 135.0 0.11 0.00 114.8 85.8 56.3 0.00 32.8 250.0 174.9 11836.
14] 000 12 00 888 178 198 00 135 000 511 0000
CPSO[1 135.0 40.3 10.0 64.60 70.9 39.9 4.07 60.0 250.0 175.0 12132.
2] 000 446 506 60 318 918 73 000 012 009 8579
TVAC-P 135.0 40.0 10.0 64.94 74.8 39.8 16.1 44.1 250.0 175.0 12117.
SO[12] 000 118 391 91 263 443 867 428 000 000 3895
FA[28] 134.8 40.0 10.0 65.18 75.0 40.0 16.9 43.0 249.9 174.9 12119.
1 0 0 0 0 7 2 9 9 86
IWO[15] 134.5 40.0 10.9 64.47 75.0 38.9 8.81 52.2 250.0 175.0 12134.
9 0 4 0 8 1 0 0 33
BD[10] 135.0 40.0 10.0 65.00 75.0 40.0 14.4 45.5 250.0 175.0 12116.
000 000 000 00 000 000 029 971 000 000 60
RCGA-I 135.0000 40.0000 10.0000 65.0000 75.0000 40.0000 14.0595 45.9405 250.0000 175.0000 12104.8682
MM
LP HS[2] 41.41 66.6 10.5 41.39 97.7 40.2 22.8 59.2 160.0 220.0 11810.
3 00 100 900 00 300 300 300 100 000 000 8800
GA[2] 37.98 76.3 10.4 35.03 106. 38.3 15.8 59.9 159.8 220.1 11837.
00 900 100 00 0000 700 400 700 100 800 4000
EDHS*[ 135.0 0.00 0.00 25.00 87.2 58.1 40.1 34.3 160.0 219.9 93181.
14] 000 00 00 00 560 586 823 703 000 672 0000
CPSO[1 35.59 57.3 10.0 57.05 89.9 40.0 30.0 60.0 160.0 220.0 11781.
2] 72 554 070 87 767 025 232 000 183 024 3690
TVAC-P 42.14 64.6 10.0 43.22 96.2 40.0 23.7 60.0 160.0 220.0 11758.
SO[12] 33 271 001 95 593 001 404 000 000 000 0625
BD[10] 42.14 64.6 10.0 43.22 96.2 40.0 23.7 60.0 160.0 220.0 11758.
54 296 000 50 614 000 386 000 000 000 06
RCGA-I 42.16 64.6 10.0 43.18 96.2 40.0 23.7 60.0 160.0 220.0 11758.
MM 60 523 000 17 810 000 190 000 000 000 6349
* Not feasible.
a Total power (MW).
b TOtal heat (MWth).
c Total cost ($).
Table 7: Cost function parameters of test system III.
Page 23 of 29
Unit min max
P P
Power
only units
1 0.008 2 25 100 0.042 10 75
2 0.003 1.8 60 140 0.04 20 125
3 0.0012 2.1 100 160 0.038 30 175
4 0.001 2 120 180 0.037 40 250
a b c d e f Feasible region
coordinates
P c
,H
c
CHP
units
5 0.0345 14.5 2650 0.03 4.2 0.031 [98.8,0],
[81,104.8],
[215,180],
[247,0]
6 0.0435 36 1250 0.027 0.6 0.11 [44,0],
[44,15.9],
[40,75],
[110.2,135.6],
[125.8,32.4],
[125.8,0]
a b c H
hmin
H
hmax
Heat only
units
7 0.038 2.0109 950 0 2695.20
Table 8: Comparison of the proposed algorithm with previous methods for case III.
Outp PSO[1 EP[30] DE[20 RCGA[ BCO[2 CPSO[1 TVAC-PSO KH[31] RCGA-I
ut 8] ] 29] 9] 2] [12] MM
P1 18.462 61.361 44.211 74.6834 43.945 75 47.3383 46.3835 45.6614
Page 24 of 29
6 8 7
P2 124.26 95.120 98.538 97.9578 98.588 112.380 98.5398 104.122 98.5398
02 5 3 8 0 3
P3 112.77 99.942 112.69 167.230 112.93 30 112.6735 64.3729 112.6735
94 7 13 8 2
P4 209.81 208.73 209.77 124.907 209.77 250 209.81582 246.185 209.8158
58 19 41 9 19 3
P5 98.814 98.8 98.821 98.8008 98.8 93.2701 92.3718 98.9736 93.9960
7
P6 44.010 44 44 44.0001 44 40.1585 40.0000 40.7401 40.0000
7
H 5
57.923 18.071 12.537 58.0965 12.097 32.5655 37.8467 0.0000 28.2842
6 3 9 4
H 6
32.760 77.554 78.348 32.4116 78.023 72.6738 74.9999 66.7100 75.0000
3 8 1 6
H 7
59.316 54.373 59.113 59.4919 59.879 44.7606 37.1532 83.2900 46.7158
1 9 9
Total 608.14 607.95 608.03 607.580 608.03 600.808 600.7392 600.777 600.6865
pow 27 61 72 8 84 6 7
er
Total 150 150 149.99 150 150 150.000 150 150.000 150.0000
heat 99 0 0
Total 10613 10390 10317 10667 10317 10325.3 10100.3164 10111.1 10094.0552
Power
only units
1 0.00028 8.1 550 300 0.035 0 680
2 0.00056 8.1 309 200 0.042 0 360
3 0.00056 8.1 309 200 0.042 0 360
4 0.00324 7.74 240 150 0.063 60 180
5 0.00324 7.74 240 150 0.063 60 180
Page 25 of 29
6 0.00324 7.74 240 150 0.063 60 180
7 0.00324 7.74 240 150 0.063 60 180
8 0.00324 7.74 240 150 0.063 60 180
9 0.00324 7.74 240 150 0.063 60 180
10 0.00284 8.6 126 100 0.084 40 120
11 0.00284 8.6 126 100 0.084 40 120
12 0.00284 8.6 126 100 0.084 55 120
13 0.00284 8.6 126 100 0.084 55 120
a b c d e f Feasible
region
coordinates
P c
,H
c
CHP
units
14 0.0345 14.5 2650 0.03 4.2 0.031 [98.8,0],
[81,104.8],
[215,180],
[247,0]
15 0.0435 36 1250 0.027 0.6 0.11 [44,0],
[44,15.9],
[40,75],
[110.2,135.6],
[125.8,32.4],
[125.8,0]
16 0.0345 14.5 2650 0.03 4.2 0.031 [98.8,0],
[81,104.8],
[215,180],
[247,0]
17 0.0435 36 1250 0.027 0.6 0.11 [44,0],
[44,15.9],
[40,75],
[110.2,135.6],
Page 26 of 29
[125.8,32.4],
[125.8,0]
18 0.1035 34.5 2650 0.025 2.203 0.051 [20,0],
[10,40],
[45,55],
[60,0]
19 0.072 20 1565 0.02 2.34 0.04 [35,0],
[35,20],
[90,45],
[90,25],
[105,0]
a b c H
hmin
H
hmax
Heat only
units
20 0.038 2.0109 950 0 2695.20
21 0.038 2.0109 950 0 60
22 0.038 2.0109 950 0 60
23 0.052 3.0651 480 0 120
24 0.052 3.0651 480 0 120
Table 10: Comparison of the proposed algorithm with previous methods for case IV.
Output CPSO[12] TVAC-PSO[1 GSO[16] IGSO*[16 OTLBO*[3 GWO*[3 RCGA-IM
2] ] 2] 3] M
P1 680 538.5587 627.7455 628.152 538.5656 538.8440 448.8000
Page 27 of 29
P9 180 109.8666 110.1816 159.992 109.8992 109.9653 159.7354
H 20
415.9815 458.7020 469.3368 466.2575 468.9043 469.7337 467.4871
H 23
120 120 119.6511 120 119.9854 120.0000 119.9991
cost ($) 8 43 92
Maximu 60076.69 58359.552 58318.87 58219.14 57913.7731 - 58301.9013
m cost 03 92 13
($)
Page 28 of 29
Minimu 59736.26 58122.7460 58225.74 58049.01 57856.2676 57846.84 57927.6919
m cost 35 50 97
($)
* Not feasible.
Page 29 of 29