Energy Production Cost016

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Energy Efficiency

DOI 10.1007/s12053-016-9439-6

ORIGINAL ARTICLE

Energy production cost minimization in a combined heat


and power generation systems using cuckoo optimization
algorithm
Mehdi Mehdinejad & Behnam Mohammadi-Ivatloo &
Reza Dadashzadeh-Bonab

Received: 16 November 2014 / Accepted: 15 March 2016


# Springer Science+Business Media Dordrecht 2016

Abstract In this paper, cuckoo optimization algorithm is optimization algorithm method in comparison with many
implemented to solve energy production cost minimiza- existing methodologies.
tion in a combined heat and power (CHP) generation
system. This problem is also known as combined heat Keywords Combined heat and power economic
and power economic dispatch problem, which looks for dispatch (CHPED) . Cuckoo optimization algorithm
optimal values of power and heat generation of each CHP (COA) . Valve-point effects
unit to minimize the total production cost. Cuckoo opti-
mization algorithm is a new metaheuristic algorithm. It is Nomenclature
inspired by the life of a bird family, called cuckoo, that A. indices
special lifestyle of these birds and their characteristics in i Index for conventional thermal units
egg laying and breeding has been the basic motivation for j Index for cogeneration units
development of this algorithm. Unlike of the some pre- k Index for heat-only units
vious approaches, the effect of valve point is considered
in the cost function and clearly formulated in the conven- B. parameters
tional polynomial cost function as absolute sinusoidal Np Number of conventional
term. The proposed method is applied to three small (with thermal units
three different test cases), medium, and large test systems Nc Number of cogeneration units
in order to evaluate its efficiency and feasibility. The Nh Number of heat-only units
obtained results demonstrated a higher quality solution αi, βi, and γi The cost coefficients of the i th
and superior performance of the proposed cuckoo conventional thermal units
λi and ρi The cost coefficients for
modeling valve-point effects
Pdemand and Hdemand System power and thermal
M. Mehdinejad : B. Mohammadi-Ivatloo (*) : demands
R. Dadashzadeh-Bonab aj, bj, cj, dj, ej and fj The cost coefficients of the j th
Smart Energy Systems Laboratory, Faculty of Electrical and
Computer Engineering, University of Tabriz, Tabriz, Iran
cogeneration unit
e-mail: bmohammadi@tabrizu.ac.ir ak, bk, and ck The cost coefficients of the
k th heat-only unit
M. Mehdinejad p p
Pi min and Pi max Minimum and maximum
e-mail: m.mehdinejad1369@gmail.com
power outputs of the i th con-
R. Dadashzadeh-Bonab ventional power-only unit in
e-mail: reza.dadashzadehbonab@gmail.com megawatt
Energy Efficiency

H hk min and Hhk max Minimum and maximum source of the non-convexity of CHPED problem is
thermal outputs of the k th arisen from modeling the effects of valve points. There
heat-only unit are several works in literature that focused on solution of
CHPED problem. Considering that the CHPED prob-
C. variables
    lem is a non-convex and non-linear optimization prob-
Minimum and maxi-
Pci min Hcj and Pci max Hcj lem, there is no mathematical or metaheuristic algorithm
mum power limit of
that can guarantee the global optimal solution of such
CHP unit j which are
problems. Many attempts are done to present a better
functions of generated
algorithm, which can result in a saving in total operation
    heat Hcj
costs of CHP-integrated systems. In this work, a similar
H cjmin Pcj and H ci max Pcj Heat generation limits
problem in the literature is used for the sake of the
which are functions of
comparison of the obtained results with the previously
generated power Pcj
applied methods to CHPED problem. The implemented
H Heat output of unit
method was able to obtain the feasible and better solu-
P Power output of unit
tions in comparison with all of the previously published
D. Functions works in a reasonable time.
Ci(Ppi ) Fuel cost of conventional thermal unit i
Cj(Pcj , Hcj ) Cost function of the cogeneration unit j
h
Ck(Hk ) Cost of heat-only unit k Literature review

In Su and Chiang (2004), improved genetic algorithm


with multiplier updating (IGA-MU) is used to solve
Introduction CHPED problem. Optimal solution of CHPED problem
using harmony search (HS) algorithm is presented in
Background and motivation Vasebi et al. (2007) and Khorram and Jaberipour (2011).
Cubic cost function for power-only units is considered
Combined heat and power (CHP) systems are consid- in Vasebi et al. (2007) and Khorram and Jaberipour
ered as the most efficient technologies for power and (2011), and valve-point effects are not taken into ac-
heat generation. In this system, the required heat and count. In Piperagkas et al. (2011), multi objective
power of the network are generated simultaneously with CHPED problem is solved taking into account wind
higher efficiency (Alipour et al. 2014). There are other power generation using particle swarm optimization
methods like as thermoelectric generation and phase (PSO) algorithm with considering quadratic cost func-
change material engine systems that generate electricity tions for simultaneously minimizing cost and emissions.
from excess heat in industries like as steel (Johansson In Song et al. (1999), CHPED problem is solved using
and Soderstrom 2014) and metallurgic sector (Ferreira ant colony search algorithm (ACSA), where the authors
et al. 2015). Determination of the optimal heat and recommended to use another heuristic algorithm with
power output of each CHP unit is an important problem ACSA in order to solve premature convergence problem
in operation of the CHP integrated systems, which is of ACSA. In Ramesh et al. (2009), a novel selective
known as CHP economic dispatch (CHPED). The ob- particle swarm optimization (SPSO) is presented to
jective of CHPED is determining the optimal outputs of improve the efficiency of PSO algorithm in solution of
the power and heat sources to satisfy heat and power CHPED problem. Other evolutionary computation tech-
demand of system with minimum fuel cost considering niques such as genetic algorithm (GA) (Subbaraj et al.
operational constraints. The CHPED problem has 2009), evolutionary programming (EP) (Wong and
attracted a lot of interests in recent years. Dual depen- Algie 2002), harmonic search algorithm (HSA)
dency of heat and power production in CHP units makes (Huang and Lin 2013), PSO (Behnam Mohammadi-
the CHPED problem a complicated optimization prob- Ivatloo et al. 2013a; Wang and Singh 2008), and teach-
lem, which needs powerful optimization techniques to ing learning-based optimization (TLBO) (Roy et al.
solve it. This dual dependency makes the search space 2014) are proposed in literature for solution of CHPED
of the problem non-convex in some cases. Another problem. A review of research works related to short
Energy Efficiency

term scheduling of CHP units can be found in Salgado host birds, they grow and become a mature cuckoo
and Pedrero (2008). (Rajabioun 2011). Appropriate penalty functions are
Recently, cuckoo inspired algorithms have attracted incorporated in fitness function for handling different
attention of the researchers due to their efficiency in equality and inequality constraints. The COA has been
handling large optimization problems. Two types of validated and compared with other algorithms such as
cuckoo inspired algorithms have been developed in GA and PSO (Rajabioun 2011). The main contributions
literature titled cuckoo search algorithm (CSA) of this work can be summarized as follows:
(Gandomi et al. 2013) and cuckoo optimization algo-
rithm (COA) (Rajabioun 2011). Differences of these 1. Implementation of COA algorithm to CHPED prob-
two cuckoo inspired algorithms are highlighted in Fister lem with non-convex objective functions and search
and Yang (2015). CSA has been introduced in Gandomi spaces.
et al. (2013) as a new metaheuristic optimization algo- 2. Demonstrating the effectiveness of the implemented
rithm for solving structural optimization problems. The algorithm using both standard and large-scale
interested reader can refer to Nguyen and Truong (2015) CHPED test systems with different sizes and
for application of CSA algorithm for solving distribu- characteristics.
tion network reconfiguration problem. 3. Obtaining feasible and better results for large-scale
COA was presented in 2011 by R. Rajabioun CHPED test systems in comparison with available
(Rajabioun 2011), about two years after the publication methods in literature.
of the original CSA. There is no direct link between 4. Better convergence characteristics and lower simu-
CSA and COA, except that both algorithms were in- lation times in comparison with other algorithms.
spired by cuckoo behavior (Fister and Yang 2015). Due
to the title similarity and the similar source of inspira-
tion, unfortunate confusion may occur. The main differ- Paper organization
ences of both algorithms are highlighted in Fister and
Yang (2015). In Mellal et al. (2013), the optimal selec- The remainder of this paper is organized as follows:
tion of obsolete tools in manufacturing systems is done BFormulation of CHPED problem^ provides the
using COA. The COA is used in Mellal et al. (2012) as a mathematical formulation of the CHPED problem
new approach to deal with technological obsolescence considering valve-point effects. BCuckoo optimization
of industrial plants. Two bio-inspired algorithms called algorithm^ gives a brief overview of the cuckoo opti-
the cuckoo optimization algorithm and the hoopoe mization algorithm. Implementation of COA to CHPED
heuristic are used in Mellal and Williams (2014) to problem is provided in BImplementation of cuckoo op-
optimize the machining control parameters of these timization algorithm for CHPED problem,^ and then,
processes. The COA method is proposed for solving several case studies and comparison with some other
optimal power flow (OPF) problem in Le Anh et al. optimization methods are presented in BSimulation re-
(2015). In Ameryan et al. (2014), four novel clustering sults and discussions.^ Conclusions are finally given in
methods based on cuckoo optimization algorithm are BConclusion^.
presented.

Contributions Formulation of CHPED problem

In this paper, COA is proposed for solving non-linear In this section, the CHPED problem is formulated as an
and non-convex CHPED problem. COA is a new evo- optimization problem, and the corresponding objective
lutionary optimization algorithm which is inspired by function and constraints are described.
lifestyle of a bird family called cuckoo (Rajabioun
2011). Specific egg laying and breeding of cuckoos are Objective function
the basis of this optimization algorithm. Cuckoos used
in this modeling exist in two forms: mature cuckoos and The objective function of CHPED problem is minimiz-
eggs. Mature cuckoos lay eggs in some other birds’ nest, ing the total power and heat production cost, while the
and if these eggs are not recognized and not killed by heat and power demands and other constraints are met.
Energy Efficiency

The objective function can be stated as follows (Behnam Constraints


Mohammadi-Ivatloo et al. 2013a):
X
Np X
Nc   XNh
   .  The CHPED problem should be solved considering the
min C i ðPpi Þ þ C j Pcj ; H cj þ C k H hk $ h ð1Þ following constraints:
i¼1 j¼1 k¼1

where Np, Nc and Nh are the number of conventional Equality constraints


thermal units, cogeneration units, and heat-only units,
respectively. H and P are the heat and power output of & Heat and power balance constraints
unit, respectively.
X
Np X
Nc
Ci(Ppi ), Cj(Pcj , Hcj ) and Ck(Hhk ) represent the fuel cost PPi þ Pcj ¼ Pdemand ð6Þ
of conventional thermal unit i, cost function of the i¼1 j¼1
cogeneration unit j, and the cost of heat-only unit k,
respectively.
Quadratic cost function of thermal units can be writ- X
Nc X
Nh
ten as follows: H cj þ H hk ¼ H demand ð7Þ
j¼1 k¼1
 . 
2
C i ðPpi Þ ¼ αi ðPpi Þ þ β i Ppi þ γ i $ h ð2Þ where Pdemand and Hdemand are the system power and
thermal demands, respectively.
where αi, βi, and γi are the cost coefficients of the i th
conventional thermal unit. In a practical generation unit,
steam valve admission effects lead to the ripple in the Inequality constraints
production cost. In order to model this effect more
accurately, a sinusoidal term is added to the quadratic & Capacity limit constraints
cost function (Behnam Mohammadi-Ivatloo et al.
2013b). Considering valve-point effects make the prob- PPi min ≤ PPi ≤ PPi max i ¼ 1; ⋯; N p ð8Þ
lem non-convex and non-differentiable. The unit cost
function considering valve-point effects is formulated as    
follows: Pcjmin H cj ≤ Pcj ≤ Pcjmax H cj j ¼ 1; ⋯; N c ð9Þ
      . 
C i ðPpi Þ ¼ αi ðPpi Þ þ β i Ppi þ γ i þ λi sin ρi PPi min −Ppi 
2
$ h
   
ð3Þ H cjmin Pcj ≤ H cj ≤ H cjmax Pcj j ¼ 1; ⋯; N c ð10Þ
where λi and ρi are the cost coefficients for modeling
valve-point effects.
The production cost of cogeneration and heat-only H hk min ≤ H hk ≤ H hk max k ¼ 1; ⋯; N h ð11Þ
units are defined as follows: p p
where Pi min and Pi max are the minimum and the
   2  2  . 
maximum power outputs of the i th conventional
C j Pcj ; H cj ¼ a j Pcj þ b j Pcj þ c j þ d j H cj þ e j H cj þ f j H cj Pcj $ h  
ð4Þ power-only unit in megawatt. Pcjmin H cj ; Pcjmax
     
H cj ; H cjmin Pcj and H cjmax Pcj are the func-
tions that define the limits of CHP units considering
heat and power dual dependency for j th CHP unit.
   2
C k H hk ¼ ak H hk þ bk H hk þ ck ð$=hÞ ð5Þ H hk min and H hk max are also the minimum and the
maximum thermal outputs of the k th heat-only unit.
where, aj, bj, cj, dj, ej and fj are the cost coefficients of It is obvious that the heat production limits of CHP
the j th cogeneration unit; ak, bk and ck represent the units are dependent to the unit power production,
cost coefficients of k th heat-only unit. and the power production limits of CHP units are
Energy Efficiency

dependent to the heat productions. The heat-power Cuckoos search for the many proper area to lay eggs
feasible operation region (FOR) for a combined heat in order to maximize their eggs survival rate. After
and power unit is depicted in Fig. 1. The upper and remained eggs become mature cuckoos, they make
lower bounds of heat and power units are restricted some societies so that each society has its habitat region
by their own generation limits. to live in. The best habitat of all societies will be the
destination for the cuckoos in other societies, and they
immigrate toward this best habitat (Rajabioun 2011).
Egg laying radius is dedicated to each cuckoo with
Cuckoo optimization algorithm considering the number of eggs and the cuckoo’s dis-
tance to the best habitat. After defining egg laying radius
Cuckoo optimization algorithm (COA) was presented in (ELR), cuckoo starts to lay eggs in some random nests
2011 by R. Rajabioun (Rajabioun 2011), about two inside her ELR. This process continues until the best
years after the publication of the original cuckoo search position with maximum profit value is obtained and
algorithm by Yang and Deb in 2009 (Yang and Deb most of the cuckoo population is gathered around the
2009). The COA algorithm was inspired by cuckoos same position (Rajabioun 2011).
behavior. Special characteristics of cuckoos in breeding
and egg laying had been the basic motivation for devel-
opment of cuckoo optimization algorithm. Cuckoos Implementation of cuckoo optimization algorithm
used in this modeling are divided into two forms, i.e., for CHPED problem
mature cuckoos and eggs. Mature cuckoos lay eggs in
some other birds’ nest, and if these eggs are not recog- The following parameters of cuckoo optimization algo-
nized and not killed by host birds, they grow and be- rithm are considered in this paper. These parameters are
come a mature cuckoo (Rajabioun 2011). The flowchart selected as appropriate parameters by doing simulations
of the proposed COA method for solution of CHPED on case II with different values for parameters.
problem is shown in Fig. 2. The proposed algorithm, Number of initial population (cuckoos) = 4; mini-
like other evolutionary algorithms, starts with an initial mum number of eggs for each cuckoo = 2; maximum
population of cuckoos. These cuckoos have some eggs number of eggs for each cuckoo = 4; maximum iteration
to lay in some host birds’ nests. Some of these eggs = 1500; motion coefficient = 4; radius coefficient (con-
which are more similar to the host bird’s eggs have the trol parameter of egg laying) = 2.
opportunity to grow up and become a mature cuckoo The steps of solution of non-convex CHPED prob-
(Rajabioun 2011). If a host bird realize that the eggs are lem using cuckoo optimization algorithm are presented
not its own, it will either throw these eggs away or in the following.
abandon its nest and build a new nest elsewhere.
Step 1 Initialize the population of cuckoo (n = 4);
The problem independent variables are ini-
P(MW)
tialized considering their feasible numerical
range. The independent variables such as
125.8
(Nh − 1) heat-only units and real power output
110.2 of (Np − 1) power generating units, real power
and heat output of CHP units of CHPED prob-
lem are initialized randomly within their spec-
ified operating limits as follows:
44  
40 p p p
Ppi; j ¼ P j min þ rand ð0; 1Þ  P j max −P j min j ¼ 1; ⋯; N p −1

ð12Þ

15.9 135.6 H(MWth)  


32.4 75 Pci; j ¼ Pcjmin ð:Þ þ rand ð0; 1Þ  Pcjmax ð:Þ−Pcjmin ð:Þ j ¼ 1; ⋯; N c
Fig. 1 The heat-power feasible regions for a combined heat and
power unit (CHP unit 3 in case I) ð13Þ
Energy Efficiency

Start

Initialize the population of


cuckoo using (12)-(15) based
on step 1

Determine egg laying radius


for each cuckoo
Lay eggs in different nests

Some of eggs are detected and


killed based on step 5
Move all cuckoos toward best
environmant

Kill cuckoos are worst NO Is population less than


area based on step 7 max value ? (based on
step 7) Determine cuckoo societres

Yes

Check survival of eggs in


Find nest with best survival
nests using (18) (based on
rate
step 8)

Is stop condition
sastisfied ? (based on Let eggs grow
step 11) NO

Yes

END

Fig. 2 Flowchart for cuckoo optimization algorithm

      
H ci; j ¼ H cjmin Pci; j þ rand ð0; 1Þ  H cjmax Pci; j −H cjmin Pci; j where rand(0, 1) is a uniform distributed ran-
j ¼ 1 ; ⋯ ; Nc dom generated number between 0 and 1. In
ð14Þ order to meet the equality constraints of the
  heat demand and power demand, power gener-
H hk; j ¼ H hj min þ rand ð0; 1Þ  H hj max −H hj min j ¼ 1; ⋯; N h −1
ation of the Npth power generating unit (with-
ð15Þ out considering the power loss) and heat output
Energy Efficiency

Table 1 Cost function parameters of the CHP units of case I

CHP units a($/MW2) b ($/MW) c ($) d ($/MWth2) e ($/MWth) f ($/MW MWth) Feasible region coordinates [Pc,Hc]

2 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]
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]

of the Nhth heat generating unit are calculated Step 4 Let cuckoos to lay eggs inside their correspond-
using the following equations: ing egg laying radius;
Step 5 Kill those eggs that are recognized by host birds;
Step 6 Let eggs hatch and chicks grow;
N p −1þN c
X Step 7 Limit cuckoos Bmaximum number in the envi-
PN p ¼ Pdemand − Pj ð16Þ ronment and kill those who live in worth
j¼1
habitats^;
Step 8 Evaluate the habitat of each newly grown cuckoo;
N h −1þN c
The habitat of each newly grown cuckoo is
X evaluated using the fitness functions. The fitness
H N h ¼ H demand − Hj ð17Þ
j¼1 function should be minimized while satisfying
the equality and inequality constraints. The pop-
Step 2 Assign some eggs to each cuckoo (maximum ular way of the constraint handling is adding
number of eggs for each cuckoo = 4 and min- penalties for violated constraints. The fitness
imum number of eggs for each cuckoo = 2); function is defined as the heat and power pro-
Step 3 Define egg laying radius for each cuckoo (ra- duction cost (1) plus penalties as stated in the
dius coefficient = 2); following.

Table 2 Comparison of optimal results for case I

Load Method P1 P2 P3 P4 H2 H3 H4 H5 Total


generation
cost ($)

LP1 GA (Subbaraj et al. 2009) 135.0000 70.8100 10.8400 83.2800 80.5400 39.8100 0.0000 29.6400 13779.5000
CPSO (Behnam Mohammadi-Ivatloo et al. 2013a) 135.0000 40.7309 19.2728 105.0000 64.4003 26.4119 0.0000 59.1955 13692.5212
EDHSa (Khorram and Jaberipour 2011) 135.0000 18.1563 13.0749 133.7688 84.0626 37.7657 0.0000 28.1118 13613.0000
TVAC-PSO (Behnam Mohammadi-Ivatloo et al. 2013a) 135.0000 41.4019 18.5981 105.0000 73.3562 37.4295 0.0000 39.2143 13672.8892
COA 135 40.7687 19.2313 105 73.5956 36.7760 0 39.6284 13672.8341
LP2 GA (Subbaraj et al. 2009) 119.2200 45.1200 15.8200 69.8900 78.9400 22.6300 18.4000 54.9900 12327.3700
CPSO (Behnam Mohammadi-Ivatloo et al. 2013a) 135.0000 40.3446 10.0506 64.6060 70.9318 39.9918 4.0773 60.0000 12132.8579
EDHSa (Khorram and Jaberipour 2011) 135.0000 0.1112 0.0000 114.8888 85.8178 56.3198 0.0000 32.8135 11836.0000
TVAC-PSO (Behnam Mohammadi-Ivatloo et al. 2013a) 135.0000 40.0118 10.0391 64.9491 74.8263 39.8443 16.1867 44.1428 12117.3895
COA 135 40 10 64.991 75 40 14.4001 45.6 12116.6
LP3 GA (Subbaraj et al. 2009) 37.9800 76.3900 10.4100 35.0300 106.0000 38.3700 15.8400 59.9700 11837.4000
CPSO (Behnam Mohammadi-Ivatloo et al. 2013a) 35.5972 57.3554 10.0070 57.0587 89.9767 40.0025 30.0232 60.0000 11781.3690
EDHSa (Khorram and Jaberipour 2011) 135.0000 0.0000 0.0000 25.0000 87.2560 58.1586 40.1823 34.3703 93181.0000
TVAC-PSO (Behnam Mohammadi-Ivatloo et al. 2013a) 42.1433 64.6271 10.0001 43.2295 96.2593 40.0001 23.7407 60.0000 11758.0625
COA 42.1497 64.6342 10 43.2161 96.2654 40 23.7346 60 11758.06
a
Out of the feasible region
Energy Efficiency

X
Np X
Nc   XNh
 
min C i ðPpi Þ þ C j Pcj ; H cj þ C k H hk
memory. Also, the number of initial population, mini-
i¼1
"
j¼1 k¼1
#2
mum, and maximum number of eggs for each cuckoo
X
Np X
Nc
are considered four, two, and four, respectively. Maxi-
þ ω1 Ppi þ Pcj −Pdemand −PPloss
i¼1 j¼1 mum number of the iterations is considered 1500 gen-
" #2
XNc X
Nh
erations for all test cases.
þ ω2 H cj þ H hk −H demand
j¼1 k¼1
Np D
X E Np D
X E
þ ω3 Ppij −Ppmin
j þ ω4 Ppmax
j −Ppij Five-unit test system
j¼1 j¼1
XNh D E X Nh D E
þ ω5 H hij −H hmin
j þ ω6 H cmax
j −H hij This test system consists of a conventional power unit,
j¼1 j¼1
Nc D
X  E X Nc D   E three cogeneration units, and a heat-only unit. This case
þ ω7 Pcij −Pcmin
j H cij þ ω8 Pcmax
j H cij −Pcij has been discussed in Vasebi et al. (2007). The test
j¼1 j¼1
Nc D
X  E Nc D
X   E system is considered for three power and heat demand
þω9 H cij −H cmin Pcij þ ω10 H cmax Pcij −H cij
j¼1
j j
j¼1
levels. The power and heat demands for three different
load profiles are 300 MW and 150 MWth (LP1),
ð18Þ
250 MW and 175 MWth (LP2), and 160 MW and
where ω1, …, ω10 are penalty parameters. The 220 MWth (LP3), respectively. The cost functions and
bracket operator 〈〉 denotes the absolute value of capacity limits of the conventional and heat-only units
the operand, if the operand is negative; if the are expressed in (19)–(23), respectively. Parameters of
operand is non-negative, it returns zero. cost function of CHP units are provided in Table 1, and
Step 9 Cluster cuckoos and find the best group and the cost functions of these units are obtained by (4).
select the goal habitat;
X
1
X
3   X1
 
Step 10 Let the new cuckoo population migrate to the min C i ðPpi Þ þ C j Pcj ; H cj þ C k H hk ð$=hÞ
goal habitat; i¼1 j¼1 k¼1

Step 11 If stopping condition (number of function ð19Þ


evaluations = 26,000) is satisfied, then
stop or else go to step 2 and end. C 1 ðP1 Þ ¼ 254:8863 þ 7:6997P1 þ 0:00172P21 þ 0:000115 P31
ð20Þ

Simulation results and discussions C 5 ðH 5 Þ ¼ 950 þ 2:0109H 5 þ 0:038H 25 ð21Þ

Subject to:
The COA described above has been implemented on
three test systems to demonstrate the performance of the 35≤ P1 ≤ 135 M W ð22Þ
COA method. The first system includes five units, a
conventional power unit, three co-generation units, and 0 ≤ H 5 ≤ 60 MWth ð23Þ
a heat-only unit with three different demand levels. The The problem is solved using the proposed cuckoo
second system includes 13 power-only units, 6 CHP optimization algorithm. The simulation results of these
units, and 5 heat-only units considering valve-point techniques are shown in Table 2. To validate the pro-
effects, and the third system is a 48-unit test system posed approach, the result of same test system is com-
considering valve-point effects. The third test system is pared with GA (Subbaraj et al. 2009), harmony search
large enough to address the scalability of the proposed algorithm (EDHS) (Khorram and Jaberipour 2011),
method. classic PSO (CPSO) (Behnam Mohammadi-Ivatloo
et al. 2013a), and time varying acceleration coefficients
Simulation parameters PSO (TVAC-PSO) (Behnam Mohammadi-Ivatloo et al.
2013a) algorithms. It should be mentioned that the
The proposed algorithm is implemented using the results of the compared algorithms have been directly
MATLAB 7.0 software and run on a PC with Intel(R) quoted from the corresponding references. It is found
Core(TM) i3-2330 M CPU 2.20 GHz 2GB RAM that the proposed COA approach provides lower
Energy Efficiency

minimum cost than the available methods in the litera- Meng et al. 2010) test system which has a lot of
ture for different load profile of this case. local minima and is one of the challenging eco-
nomic dispatch test cases (Behnam Mohammadi-
24-unit test system Ivatloo et al. 2013a). Power and heat demands are
200 MW and 115 MWth, respectively. The system
In this case, a medium-scale system is used to data are listed in Table 3. The detailed optimal
show the validity and effectiveness of the pro- results of COA for this case are given in
posed algorithm. This case study is originally pro- Table 4. From these results, it can be observed
posed in Behnam Mohammadi-Ivatloo et al. that the proposed COA finds a better solution
(2013a). The system consists of 6 CHP units, 13 ($58076.7422) in comparison with other tech-
power-only units, and 5 heat-only units. The data niques. The same CHP dispatch problem has been
of power only units are based on the 13-unit solved previously by the CPSO in Behnam
standard ED (B Mohammadi-Ivatloo et al. 2012; Mohammadi-Ivatloo et al. (2013a), TVAC-PSO in

Table 3 Cost function parameters of medium test system II

Unit α β γ λ ρ
Ppmin ðMWÞ Ppmax ðMWÞ

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
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($/MW2) b ($/MW) c ($) d ($/MWth2) e ($/MWth) f ($/MW MWth) Feasible region coordinates [Pc,Hc]
CHP units
14 0.0345 14.5 2650 0.030 4.200 0.031 [98.8, 0], [81, 104.8], [215, 180], [247, 0]
15 0.0435 36.0 1250 0.027 0.600 0.011 [44, 0], [44, 15.9], [40, 75], [110.2, 135.6],
[125.8, 32.4], [125.8, 0]
16 0.0345 14.5 2650 0.030 4.200 0.031 [98.8, 0], [81, 104.8], [215, 180], [247, 0]
17 0.0435 36.0 1250 0.027 0.600 0.011 [44, 0], [44, 15.9], [40, 75], [110.2, 135.6],
[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 20.0 1565 0.020 2.340 0.040 [35, 0], [35, 20], [90, 45], [90, 25], [105, 0]
a ($/MWth2) b ($/MWth) c ($) H hmin
ðMWthÞ H hmax
ðMWthÞ
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
Energy Efficiency

Behnam Mohammadi-Ivatloo et al. (2013a), teach- Table 4 Optimal dispatch results for case II using proposed
algorithm
ing learning-based optimization (TLBO) in Roy
et al. (2014), and oppositional TLBO (OTLBO) Output Proposed
in Roy et al. (2014). Table 5 compares the results
of COA and the mentioned methodologies for this P1 539.0637
test system. It should be mentioned that the results P2 299.3440
of the compared algorithms have been directly P3 297.8142
quoted from the corresponding references. The re- P4 108.3624
sults show that the proposed approach can still P5 109.7182
find a better solution than other methods and is P6 109.6883
an efficient approach for solving the medium-size P7 109.9509
CHP dispatch problems. By comparing the pre- P8 109.6829
sented results in Table 5 for COA and TVAC- P9 108.9633
PSO, it can be found that the obtained cost of P10 76.3384
COA is $136.406 less than TVAC-PSO for 1 hour. P11 77.4041
The daily saving of the proposed algorithm in P12 55.6846
comparison with TVAC-PSO will be $3273.744. P13 55.0000
Convergence curve of the proposed method is P14 81.3721
depicted in Fig. 3. It may be observed from P15 41.4938
Fig. 3 that the cost function value converges to P16 81.2203
the optimum value after about 26,000 number of P17 40.0477
function evaluations (about 1000 iterations). P18 10.1198
P19 39.4421
48-unit test system H14 104.8911
H15 76.2440
A large-scale system is used in this case in order H16 104.9041
to evaluate the performance of the proposed algo-
H17 75.0328
rithm. This test case consists of 26 power-only
H18 39.9883
units, 12 CHP units, and 10 heat-only units, and
H19 22.0161
the total power demand of this case is 4700 MW
H20 467.3292
and thermal demand is 2500 MWth. The required
H21 59.9883
data of this case is presented in Table 6.
H22 59.9955
The detailed results of the CHPED solution
H23 119.7519
using COA are shown in Table 7 for this case.
H24 119.8573
The results demonstrated that the proposed ap-
Total generation cost ($) 57938.32741
proach gives better solution than other approaches.
The best solutions attained by CPSO (Behnam
Mohammadi-Ivatloo et al. 2013a), TVAC-PSO
(Behnam Mohammadi-Ivatloo et al. 2013a), TLBO
Table 5 Results obtained by different methods for case II
(Roy et al. 2014), OTLBO (Roy et al. 2014), and
proposed COA methods are compared in Table 8. Solution technique Best cost ($)
It should be mentioned that the results of the
compared algorithms have been directly quoted CPSO (Behnam Mohammadi-Ivatloo et al. 2013a) 59736.2635
from the corresponding references. From these re- TVAC-PSO 58122.7460
(Behnam Mohammadi-Ivatloo et al. 2013a)
sults, it can be observed that the proposed COA TLBOa (Roy et al. 2014) 58006.9992
provides lower total operation cost among all the a
OTLBO (Roy et al. 2014) 57856.2676
above mentioned methods. For example, the ob-
COA 57938.32741
tained cost of COA is $1034.98024 less than
a
TVAC-PSO, which means daily saving of Out of the feasible region
Energy Efficiency

Fig. 3 Convergence characteristics of cuckoo method for case II

$24839.526 in total cost. Therefore, COA shows that the results of the compared algorithms have
its effectiveness in solving the problems with high been directly quoted from the corresponding
dimensions and provides a more robust perfor- references.
mance for solving different case studies. Conver- We have also provided statistical results of the
gence characteristic of the proposed method is proposed method for 100 independent runs.
depicted in Fig. 4. It can be evidently seen from Table 10 compares the statistical information of
this figure that the value of the objective function the 100 independent runs for test case II. It can
settles at the minimum after about 26,000 function be observed that the maximum cost of the COA is
evaluations (1000 iterations). less than the minimum cost of CPSO method,
which confirms the performance of the proposed
method over the CPSO. Table 11 presents the
minimum, mean, and maximum cost of the pro-
Computation time
posed algorithm for 100 independent runs for case
III. It can be observed from this table that the
The proposed methodology for solving CHPED
maximum cost of the proposed algorithm is lower
problem is implemented in Matlab 7 and executed
than the minimum cost of TVAC-PSO algorithm,
with Intel(R) Core(TM) i3-2330 M CPU 2.20 GHz
which confirms the applicability of the proposed
2GB RAM personal computer. In order to compare
algorithm for large-scale test systems.
the computation time of the proposed algorithm,
absolute CPU time and relative simulation times
for all case study systems are provided in Table 9.
Computation time has a direct relation with CPU Conclusion
speed. Relative simulation time is calculated by
multiplying relative CPU speed by reported simu- In this work, cuckoo optimization algorithm is
lation time (Behnam Mohammadi-Ivatloo et al. used to solve the non-convex and non-linear com-
2013a). It should be mentioned that both CPU plex CHPED problem. Three test cases with dif-
speed and simulation times for some methods were ferent computation burden are solved using the
not available in literature. It should be mentioned proposed method. Valve-point effects, heat-power
Energy Efficiency

Tab 6 Cost function parameters of large test system III

Unit α β γ λ ρ
Ppmin ðMWÞ Ppmax ðMWÞ

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
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
14 0.00028 8.1 550 300 0.035 0 680
15 0.00056 8.1 309 200 0.042 0 360
16 0.00056 8.1 309 200 0.042 0 360
17 0.00324 7.74 240 150 0.063 60 180
18 0.00324 7.74 240 150 0.063 60 180
19 0.00324 7.74 240 150 0.063 60 180
20 0.00324 7.74 240 150 0.063 60 180
21 0.00324 7.74 240 150 0.063 60 180
22 0.00324 7.74 240 150 0.063 60 180
23 0.00284 8.6 126 100 0.084 40 120
24 0.00284 8.6 126 100 0.084 40 120
25 0.00284 8.6 126 100 0.084 55 120
26 0.00284 8.6 126 100 0.084 55 120
a ($/MW2) b($/MW) c($) d($/MWth2) e($/MWth) f($/MW MWth) Feasible region coordinates [Pc,Hc]
CHP units
27 0.0345 14.5 2650 0.030 4.200 0.031 [98.8, 0], [81, 104.8], [215, 180], [247, 0]
28 0.0435 36.0 1250 0.027 0.600 0.011 [44, 0], [44, 15.9], [40, 75], [110.2, 135.6],
[125.8, 32.4], [125.8, 0]
29 0.0345 14.5 2650 0.030 4.200 0.031 [98.8, 0], [81, 104.8], [215, 180], [247, 0]
30 0.0435 36.0 1250 0.027 0.600 0.011 [44, 0], [44, 15.9], [40, 75], [110.2, 135.6],
[125.8, 32.4], [125.8, 0]
31 0.1035 34.5 2650 0.025 2.203 0.051 [20, 0], [10, 40], [45, 55], [60, 0]
32 0.072 20.0 1565 0.020 2.340 0.040 [35, 0], [35, 20], [90, 45], [90, 25], [105, 0]
33 0.0345 14.5 2650 0.030 4.200 0.031 [98.8, 0], [81, 104.8], [215, 180], [247, 0]
34 0.0435 36.0 1250 0.027 0.600 0.011 [44, 0], [44, 15.9], [40, 75], [110.2, 135.6],
[125.8, 32.4], [125.8, 0]
35 0.0345 14.5 2650 0.030 4.200 0.031 [98.8, 0], [81, 104.8], [215, 180], [247, 0]
36 0.0435 36.0 1250 0.027 0.600 0.011 [44, 0], [44, 15.9], [40, 75], [110.2, 135.6],
[125.8, 32.4], [125.8, 0]
37 0.1035 34.5 2650 0.025 2.203 0.051 [20, 0], [10, 40], [45, 55], [60, 0]
38 0.072 20.0 1565 0.020 2.340 0.040 [35, 0], [35, 20], [90, 45], [90, 25], [105, 0]
a ($/MWth2) b ($/MWth) c ($) H hmin
ðMWthÞ H hmax
ðMWthÞ
Energy Efficiency

Tab 6 (continued)

Unit α β γ λ ρ
Ppmin ðMWÞ Ppmax ðMWÞ

Heat only units


39 0.038 2.0109 950 0 2695.20
40 0.038 2.0109 950 0 60
41 0.038 2.0109 950 0 60
42 0.052 3.0651 480 0 120
43 0.052 3.0651 480 0 120
44 0.038 2.0109 950 0 2695.20
45 0.038 2.0109 950 0 60
46 0.038 2.0109 950 0 60
47 0.052 3.0651 480 0 120
48 0.052 3.0651 480 0 120

Table 7 Optimal dispatch results for case III using proposed


algorithm

Output Proposed Output Proposed


dual dependency, transmission loss, and capacity
limits are considered in studied systems. The ob-
P1 628.0052 P31 10.2119 tained results validate the applicability of the COA
P2 223.3670 P32 38.5893 and show the effectiveness and high potential of
P3 224.2239 P33 82.9338
the proposed framework for solving the non-con-
P4 160.4662 P34 40.4870
P5 109.4100 P35 81.7879
vex, non-smooth, and non-linear real world prob-
P6 158.5215 P36 44.8094 lems. The proposed method reduces the operation
P7 108.0529 P37 11.8972 costs of the CHP-based systems. Simulations on a
P8 109.6997 P38 54.0504 test system verified that the presented method
P9 159.5992 H27 106.2509 could results in a daily saving of $24839.526 in
P10 40.0587 H28 124.6536
comparison with available methods. The advantage
P11 40.0024 H29 104.8896
of COA for proper convergence provides feasible
P12 55.0956 H30 80.7382
P13 92.0292 H31 40.0726
results in lower iterations. As future work, CHPED
P14 447.8708 H32 21.4735 problem can be extended by considering more
P15 224.4969 H33 105.8367 practical constraints like as heat losses and multi
P16 73.9883 H34 75.4084 period modeling.
P17 159.5506 H35 105.2044
P18 109.6294 H36 79.1119
P19 159.3689 H37 40.4439
P20 159.3796 H38 28.4086
P21 158.4264 H39 444.7157 Table 8 Results obtained by different methods for case III
P22 159.3175 H40 59.5783
P23 41.4182 H41 60.0000 Solution technique Best cost ($)
P24 40.1996 H42 119.8104
P25 90.8237 H43 119.6650 CPSO 119708.8818
P26 92.4161 H44 427.0172 (Behnam Mohammadi-Ivatloo et al. 2013a)
P27 84.0293 H45 59.1377 TVAC-PSO 117824.8956
P28 97.5254 H46 59.2648 (Behnam Mohammadi-Ivatloo et al. 2013a)
TLBOa (Roy et al. 2014) 116739.3640
P29 81.3338 H47 119.4493
P30 46.9334 H48 118.8751 OTLBOa (Roy et al. 2014) 116579.2390
Total cost ($) COA 116789.91535
116789.91535 a
Out of the feasible region
Energy Efficiency

Fig. 4 Convergence characteristics of cuckoo method for case III

Table 9 Comparison of the absolute and relative CPU time for all cases

Case Method CPU speed (GHz) Absolute time (s) Relative CPU time (s)

Case I CPSO (Behnam Mohammadi-Ivatloo et al. 2013a) 2 1.5 1.5


TVAC-PSO (Behnam Mohammadi-Ivatloo et al. 2013a) 2 1.52 1.52
COA 2.2 1.672 1.672
Case II CPSO (Behnam Mohammadi-Ivatloo et al. 2013a) 2 53.36 53.36
TVAC-PSO (Behnam Mohammadi-Ivatloo et al. 2013a) 2 52.25 52.25
COA 2.2 62.7 62.7
Case III CPSO (Behnam Mohammadi-Ivatloo et al. 2013a) 2 93.32 93.32
TVAC-PSO(Behnam Mohammadi-Ivatloo et al. 2013a) 2 89.63 89.63
COA 2.2 98.59 98.59

Table 10 Comparison of minimum, mean and maximum cost for 100 independent runs for case II

Algorithms COA TVAC-PSO [13] CPSO [13]

Minimum cost ($) 57938.32741 58122.7460 59853.478


Maximum cost ($) 58175.1334 58359.552 60076.6903
Mean cost ($) 58010.285 58198.3106 59736.2635
STD 64.8930 NA NA

NA not applicable
Energy Efficiency

Table 11 Minimum, mean, and maximum cost for 100 independent runs for case III

Algorithms COA TVAC-PSO [13] CPSO [13]

Minimum cost ($) 116789.91535 117824.8956 119708.8818


Maximum cost ($) 117068.2693 NA NA
Mean cost ($) 116835.5491 NA NA
STD 81.3027 NA NA

NA not applicable

References dependability context. Journal of Scientific and Industrial


Research, 71(11), 715–721.
Mellal, M. A., Adjerida, S., & Williamsb, E. J. (2013). Optimal
Alipour, M., Mohammadi-Ivatloo, B., & Zare, K. (2014). selection of obsolete tools in manufacturing systems using
Stochastic risk-constrained short-term scheduling of industri- cuckoo optimization algorithm. Chemical Engineering, 33.
al cogeneration systems in the presence of demand response Meng, K., Wang, H. G., Dong, Z., & Wong, K. P. (2010).
programs. Applied Energy, 136, 393–404. Quantum-inspired particle swarm optimization for valve-
Ameryan, M., Akbarzadeh Totonchi, M. R., & Seyyed Mahdavi, point economic load dispatch. Power Systems, IEEE
S. J. (2014). Clustering based on Cuckoo optimization algo- Transactions on, 25(1), 215–222.
rithm. In Intelligent Systems (ICIS), 2014 Iranian Conference Mohammadi-Ivatloo, B., Rabiee, A., Soroudi, A., & Ehsan, M.
on 2014 (pp. 1–6): IEEE. (2012). Iteration PSO with time varying acceleration coeffi-
Ferreira, V. R., Augusto, C. M., Ribeiro, J. B., Gaspar, A. R., & cients for solving non-convex economic dispatch problems.
Costa, J. J. (2015). Increasing the efficiency of high temper- International Journal of Electrical Power & Energy Systems,
ature furnaces through a topping cycle cogeneration—a case 42(1), 508–516.
study. Energy Efficiency, 8(1), 85–95. Mohammadi-Ivatloo, B., Moradi-Dalvand, M., & Rabiee, A.
Fister, I., & Yang, X.-S. (2015). A short discussion about (2013a). Combined heat and power economic dispatch prob-
BEconomic optimization design of shell-and-tube heat ex- lem solution using particle swarm optimization with time
changers by a cuckoo-search-algorithm^. Applied Thermal varying acceleration coefficients. Electric Power Systems
Engineering, 76, 535–537. Research, 95, 9–18.
Gandomi, A. H., Yang, X.-S., & Alavi, A. H. (2013). Cuckoo Mohammadi-Ivatloo, B., Rabiee, A., & Soroudi, A. (2013b). Non-
search algorithm: a metaheuristic approach to solve structural convex dynamic economic power dispatch problems solution
optimization problems. Engineering With Computers, 29(1), using hybrid immune-genetic algorithm. IEEE Systems
17–35. Journal, 7(4), 777–785.
Huang, S.-H., & Lin, P.-C. (2013). A harmony-genetic based Nguyen, T. T., & Truong, A. V. (2015). Distribution network
heuristic approach toward economic dispatching combined reconfiguration for power loss minimization and voltage
heat and power. International Journal of Electrical Power & profile improvement using cuckoo search algorithm.
Energy Systems, 53, 482–487. International Journal of Electrical Power & Energy
Johansson, M. T., & Soderstrom, M. (2014). Electricity generation Systems, 68, 233–242.
from low-temperature industrial excess heat-an opportunity Piperagkas, G., Anastasiadis, A., & Hatziargyriou, N.
for the steel industry. Energy Efficiency, 7(2), 203–215. (2011). Stochastic PSO-based heat and power dispatch
Khorram, E., & Jaberipour, M. (2011). Harmony search algorithm under environmental constraints incorporating CHP and
for solving combined heat and power economic dispatch wind power units. Electric Power Systems Research,
problems. Energy Conversion and Management, 52(2), 81(1), 209–218.
1550–1554. Rajabioun, R. (2011). Cuckoo optimization algorithm. Applied
Le Anh, T. N., Vo, D. N., Ongsakul, W., Vasant, P., & Ganesan, T. Soft Computing, 11(8), 5508–5518.
(2015). Cuckoo optimization algorithm for optimal power Ramesh, V., Jayabarathi, T., Shrivastava, N., & Baska, A. (2009).
flow. In Proceedings of the 18th Asia Pacific Symposium on A novel selective particle swarm optimization approach for
Intelligent and Evolutionary Systems, Volume 1, (pp. 479– combined heat and power economic dispatch. Electric Power
493): Springer. Components and Systems, 37(11), 1231–1240.
Mellal, M. A., & Williams, E. J. (2014). Parameter optimization of Roy, P. K., Paul, C., & Sultana, S. (2014). Oppositional teaching
advanced machining processes using cuckoo optimization learning based optimization approach for combined heat and
algorithm and hoopoe heuristic. Journal of Intelligent power dispatch. International Journal of Electrical Power &
Manufacturing, 1-16. Energy Systems, 57, 392–403.
Mellal, M. A., Adjerid, S., Williams, E. J., & Benazzouz, D. Salgado, F., & Pedrero, P. (2008). Short-term operation planning
(2012). Optimal replacement policy for obsolete components on cogeneration systems: a survey. Electric Power Systems
using cuckoo optimization algorithm based-approach: Research, 78(5), 835–848.
Energy Efficiency

Song, Y., Chou, C., & Stonham, T. (1999). Combined heat and International Journal of Electrical Power & Energy Systems,
power economic dispatch by improved ant colony search 29(10), 713–719.
algorithm. Electric Power Systems Research, 52(2), 115–121. Wang, L., & Singh, C. (2008). Stochastic combined heat and
Su, C.-T., & Chiang, C.-L. (2004). An incorporated algorithm for power dispatch based on multi-objective particle swarm op-
combined heat and power economic dispatch. Electric Power timization. International Journal of Electrical Power &
Systems Research, 69(2), 187–195. Energy Systems, 30(3), 226–234.
Subbaraj, P., Rengaraj, R., & Salivahanan, S. (2009). Wong, K. P., & Algie, C. (2002). Evolutionary programming
Enhancement of combined heat and power economic dis- approach for combined heat and power dispatch. Electric
patch using self adaptive real-coded genetic algorithm. Power Systems Research, 61(3), 227–232.
Applied Energy, 86(6), 915–921. Yang, X.-S., & Deb, S. (2009). Cuckoo search via Lévy flights. In
Vasebi, A., Fesanghary, M., & Bathaee, S. (2007). Combined heat Nature & Biologically Inspired Computing, 2009. NaBIC
and power economic dispatch by harmony search algorithm. 2009. World Congress on, 2009 (pp. 210-214): IEEE.

You might also like