Received October 17, 2020, accepted November 12, 2020, date of publication November 17, 2020,

date of current version December 1, 2020.

Digital Object Identifier 10.1109/ACCESS.2020.3038740

Economic Power and Heat Dispatch in

Cogeneration Energy Systems Using
Manta Ray Foraging Optimizer
ABDULLAH M. SHAHEEN 1 , AHMED RABIE GINIDI1 ,
RAGAB A. EL-SEHIEMY 2 , (Senior Member, IEEE), AND
SHERIF S. M. GHONEIM3,4 , (Senior Member, IEEE)
1 Department of Electrical Engineering, Faculty of Engineering, Suez University, Suez 43518, Egypt
2 Department of Electrical Engineering, Faculty of Engineering, Kafrelsheikh University, Kafrelsheikh 33516, Egypt
3 Department of Electrical Engineering, College of Engineering, Taif University, Taif 21944, Saudi Arabia
4 Faculty of Technology and Education, Suez University, Suez 43527, Egypt

Corresponding author: Ragab A. El-Sehiemy (elsehiemy@ieee.org)

This work was supported by the Taif University Researchers Supporting Project, Taif University, Taif, Saudi Arabia, under
Grant TURSP-2020/34.

ABSTRACT Economic Power and Heat Dispatch (EPHD) in Cogeneration Energy Systems (CES) is
considered as one of non linear hard optimization problems. It is optimally scheduling the of heat and power
generation units. It aims at minimizing the total fuel cost (TFC) of cogeneration units considering their
operational limits. In this paper, a Manta Ray Foraging Optimization Algorithm (MRFOA), which is a recent
meta-heuristic optimization technique, is developed to solve the EPHD problem in CES with additional
non-convex valve point effects. The simplicity and effectiveness motivate the attempt of employing the
MRFOA to minimize the TFC for power units only, cogeneration units and heat units only. The equality
constraints by supplying the total loading of power and heat is are maintained. In addition, the inequality
operational bounds of power only and heat only units are satisfied while the dynamic operational bounds of
cogeneration units are not jeopardized. Three test systems are analyzed to estimate the MRFOA performance
for solving the EPHD problem in CES, which involve 5 units, 7 units, and 48 units. It is worth noticing that
the optimal solutions demonstrate MRFOA capability, feasibility and efficiency of better solutions obtained
in terms of TFC compared with other optimization methods and the ability of implementation of MRFOA
on EPHD issue in CES..

INDEX TERMS Cogeneration energy systems, optimal dispatch, manta ray foraging algorithm, non-convex
optimization, valve point.

I. INTRODUCTION The EPHD optimization problem in CES has more com-

Energy systems are pivotal basis for the survival and the plicated cost functions, which depends on both heat-power
development of national economy. The demand of energy in capacity and heat–power demand [3], [4].
the economic society is becoming huge. Cogeneration Energy A plethora of conventional and mathematical approaches
Systems (CES) are capable of simultaneous generating ther- were developed for solving the EPHD optimization prob-
mal and electrical energy with high efficiency [1] and great lem like as assequential quadratic programming (SQP) [5],
reduction of the environment contaminant gas emissions [2]. Lagrangian relaxation (LR) [6], [7], and Benders decomposi-
Therefore, the basic target of Economic Power and Heat Dis- tion (BD) [8]. These techniques with surrogate sub-gradient
patch (EPHD) in CES is getting the minimum operating costs updating multiplier has been utilized. However, the con-
of CES with optimum scheduling of power and heat units straints relaxation may yield to local optimal. Additional non-
with maintaining the corresponding operational constraints. linearity characteristics reflects higher complexity such as the
valve-point impact, prohibited operation zones of power-only
The associate editor coordinating the review of this manuscript and units, and transmission power losses. Therefore, it is required
approving it for publication was Weixing Li. to adopt the heuristic optimization techniques to overcome

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
VOLUME 8, 2020 208281
 A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

non-convex and non-smooth operational cost function and to immune system algorithm [32], line-up competition algo-
handle varied constraints. rithm [33], oppositional teaching learning based optimiza-
Heuristic optimizers are employed to the considered prob- tion algorithm (OTLBO) [34], multi-player harmony search
lem because they have capability of solving the complexity of (MPHS) [35], quantum optimization (QO) [36], modi-
this issue by treating it as a black box. The researchers have fied GSO (MGSO) [37] approach, and gravitational search
employed many optimization algorithms such as harmony algorithm (GSA) [38]. These techniques utilized definite
search (HS) [9], evolutionary programming [10], genetic procedures on the basis of randomness to explore the
algorithm (GA) [11], and the modified particle swarm opti- space of feasible solution instead of the differential rules
mization (MPSO) [12] to obtain the optimal scheduling of in the mathematical models. Then, these methods are quite
power and heat generation units with minimum total cost. suitable for dealing with the EPHD problems with non-
Additionally, in [13], a self-adaptive real-coded genetic algo- differentiability, discontinuity, and non-convexity. Despite
rithm (SARGA) was applied to the EPHD problem in CES by these extensive applications, there is still continuous progress
adopting a strategy for penalty constraint-handling. A hybrid in this area searching for reliable and effective techniques.
PSO and weighted vertices based (WVO), has been applied One of these new effective tools is the Manta ray foraging
in [14] to obtain the optimal solution of EPHD problem optimizer (MRFOA) that was recently proposed, in 2020,
in CES. PSO with time-varying acceleration coefficients by W. Zhao and others [39]–[41]. It presented three strategies
(TVAC-PSO) [15], [16] has been implemented with adding for foraging of the manta ray to develop an optimization algo-
a sinusoidal term to the polynomial cost function to dis- rithm for solving numerous engineering issues. Therefore,
play the effect of the valve point. Whale optimization algo- in the current paper, the MRFOA was employed to solve the
rithm (WOA) [17] and real-coded genetic algorithm with EPHD problem in CES considering the valve-point loading
improved Mu’ hlenbein mutation (RCGAIMM) [18] have effects and the operational constraints such as equality and
been employed on this issue considering the effect of both inequality limits. The MRFOA is developed for handling
transmission losses and valve-point. In [19], scheduling opti- the equality and inequality operational limits via involving
mization model of the combined energy system has been an adaptive penalty parameter to penalize the infeasible
developed to meet the requirements of heating, cooling, and solutions with different degrees based on its remoteness from
electrical loads. In [20], a quasidynamic pipeline model for the nearest feasible point. MRFOA is efficiently applied
EPHD with start-stop schedule of heat exchange stations with high superiority and constraints fulfillment on three
is developed. Also, an invasive weed optimizer (IWO) [21] test instances including 5-unit, 7-unit, and 48-unit systems
based on the process of weed colonization/distribution has with diverse output demand and thermal demand. Various
been applied optimal operation of the CES for small scale comparisons are compared with recent techniques that are
EPHDproblems. In addition, an integrated technique, which reported in the literature. The simulation results illustrate
combined the civilized swarm optimization (CSO) and the the MRFOA capability, efficiency and feasibility of better
Powell’s pattern search (PPS) method [22], has been carried solutions attained in terms of fuel cost with respect to the
out to solve the EPHDproblem in CES. Furthermore, utilizion other optimization methods and the ability of implementation
and assessment for the available reserve capacity of CHP of MRFOA on different test systems.
units is develoloped in [23]. A nonlinear dynamic model of The rest of this article is demonstrated as follows: The
a EPHD is proposed to solve the unknown parameters with formulation of EPHD problem in CES is given in Section II.
absorption heat pump [24]. Moreover, crisscross optimization Section III introduces the MRFOA for obtaining the optimal
(CSO) algorithm [25] has been applied to solve the EPHD- solution of this optimization issue. The MORF is employed
problem in CES. In [26], grey wolf optimization (GWO) has on three test instances of the EPHD optimization issue in CES
been subjected to solve it with additional constraints of spin- in Section IV. The MRFOA is compared with the state of
ning reserve constraint and ramp-rate limits. Coordination of the arts that displayed in the previous research studies and
both operation of electric power systemand district heating assessment to the other optimization techniques is developed.
system is introduced in [27]. In [28], improved version of Eventually, the paper conclusion is displayed in Section V.
group search optimization (IGSO) has been compared to the
conventional GSO in solving the EPHDproblem in CES while II. PROBLEM FORMULATION
a stochastic fractal search (SFS) has been applied, in 2020 to In the EPHD problem with CES, it is required to reduce
ensure achieving equality and inequality constraints to differ- the total fuel cost (TFC) of cogeneration, power only, and
ent instance systems in [29] but very small systems of 4 units heat only units to fulfill the demands of power and heat.
and 5 units have been utilized. Compining non-dominated The generation costs of three categories of power and heat
sorting GA with multi-objective PSO [30] have been effi- production plants can be written as [17]:
ciently employed for the same purpose but they are valid for
just small scale applications of 5- unit and 7 unit systems. Np Nh NC
p
X X X
Cj Hjh + k , HK )($/h)
Ck (PC C
There are some other swarm intelligent algorithms Min Ci Pi +
that has been integrated to the EPHD problems in i=1 j=1 k=1
CES such as bee colony optimization [31], artificial (1)

208282 VOLUME 8, 2020

A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

p p p
Ci (Pi ) = ai (Pi )2 + bi Pi + ci ($/h) (2) respectively; Pckmin (Hkc ) and Pckmax (Hkc ) express the minimum
Cj (Hjh ) =
p
aj (Hjh )2 + bj Pj + cj ($/h) (3) and maximum outputs of k th cogeneration unit which are
p functions of generated heat Hkc ; Hkcmin (Pck ) and Hkcmax (Pck )
Ck (Pck , Hkc ) = ak (Pck )2 + bk Pk + ck + dk (Hkc )2 give the lower and upper bounds of the k th heat generation
+ ek Hkc + fk Hkc Pck ($/h) (4) which are functions of produced power Pck . The fifth and the
most important constraint is the consideration of valve point
where the overall cost can be manifested by C; Np , Nh , and Nc
impact. It is highly required to study and analyze the impact of
represent the number of power-only plants, heat-only plants,
valve point that because of the ample steam-admitting valves
and cogeneration units, respectively. The symbols of the out-
in the power-only plants. It can be said that the valve-point
puts of power and heat generation units are demonstrated
impacts could be demonstrated by sinusoidal term, which is
by P and H . The indices i, j, and k reflect the number of
added to the objective function because the EPHD problem
the power only plants, heat only plants, and cogeneration
p in CES is considered as non-differentiable and non-convex
units, respectively. The term Ci (Pi ) reflects the conventional
p problem with considering the impacts of the valve point [41].
thermal fuel feature of unit i for generating Pi MW at
Thus, the objective function of the EPHD problem in CES
one-hour time horizon. The coefficients of the cost function
can be described as in (11).
of ith power-only plant are expressed by ai , bi , and ci . Equa-
p p p p p
tion (2) manifests the quadratic function approximation of the Ci (Pi ) = ai (Pi )2 + bi Pi + ci + λi sin(ρi (Pi min − Pi ))
cost function of power-only plant. The cost of jth heat-only (11)
plant is demonstrated by Cj (Hjh ) and produces heat (H h )
MWth. The coefficients of cost function of jth heat-only plant where, λi and ρi refer to the valve-point impact cost coeffi-
are illustrated by aj ; bj , and cj as shown in (3). The term cients.
Ck (Pck , Hkc ) displays the operational cost of k th cogeneration
III. MANTA RAY FORAGING ALGORITHM FOR
unit and the cost coefficients of the k th unit are reflected by
OPTIMIZED EPHD IN CES
ak , bk , k ck , dk , ek and fk . Considering (4), power output Pc
In this paper, the new recent optimization algorithm called
and heat output H c are the two parameters of the cost function
Manta Ray Foraging Optimization Algorithm (MRFOA) is
of the cogeneration units.
developed for solving EPHD problem in the CES. MRFOA
In the next points, five constraints are described and
elaborates three intelligent foraging strategies of manta rays.
explained to get the optimal solution of EPHD problem in
These strategies are chain foraging, cyclone foraging and
CES. The first constraint is the balance between the power
somersault foarging. For the first startgey, the manta rays
generation and demand which is expressed in Eq. (5), while
forage in a minor cooperative way are ordered in an orderly
the second constraint is the balance between the heat genera-
line to funnel the highest quantity of plankton into their gills.
tion and demand is expressed as in Eq. (6):
The second strategy, numerous manta rays are accompanied
Np Nc in a spiral shape to produce a centralized spiraling peak.
p
X X
Pi + Pck = Pd , (5) Consequently, this forces the water up towards the surface and
i=1 j=1 pull out the plankton inside their mouths. The last strategy
Nc
X Nh
X is the somersault foraging, where the manta rays search for
Hjc + Hkh = Hd , (6) the planktons’ position and swim towards them. The manta
j=1 k=1 rays show the individuals in the search space that seek for
where, Pd denotes the electric power demand of the system the minimum fitness which is the planktons’ position. The
and Hd is the system heat demand. MRFOA is initialized randomly within the operating lim-
The third constraint is the limitations of power-only plants its of the control variables for population number of the
capacity that are illustrated in Eq. (7), whilst the fourth con- manta rays (PM) and maximum number of iterations (Itm)
straint is the generation limits of heat-only units that are elab- for D-dimensional individuals (M) Thus in the first strategy,
orated in Eq. (8). Moreover, capacity limits of cogeneration the chain foraging, everyone is updated as follows:
(
energy units are depicted in Eqs. (9-10). ∗ Mm + (MB − Mm )(r + σ ) if m = 1
Mm =
p p
Pi min ≤ Pi ≤ Pi max
p
i = 1, . . . , Np , (7) Mm + r(Mm−1 − Mm ) + σ (MB − Mm ) else
Hjhmin ≤ h
Hj ≤ Hj hmax
j = 1, . . . , Nh , (8) (12)
σ = 2.r. |log(r)|
p
(13)
Pckmin (Hkc ) ≤ c cmax
Pk ≤ Pk (Hk ) k = 1, . . . , Nc ,
c
(9)
Hkcmin (Pck ) ≤ Hkc ≤ Hkcmax (Pck ) k = 1, . . . , Nc , (10) where the symbol (Mm) represents the individual of each
manta ray (m); r refers to random uniformly distributed num-
In the above formula, the lower and upper limits of out- ber within range [0,1]. M∗ and MB illustrate the new position
p p
put power of each power-only unit i are Pi min and Pi max in and the best position of the manta ray in the population,
MW, respectively; Hjhmin and Hjhmax reflect the lower and respectively. σ describes the weight coefficient which varies
upper bounds of the outputs of heat-only unit jth in MWth, each iteration.

VOLUME 8, 2020 208283

 A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

the balance constraints of the power and heat loads are guar-
anteed as illustrated in (5) and (6). For the related constraints
to the operational limits of the units, they are divided into two
types. The first is self-dependent, which are the outputs of
power-only and heat-only units as manifested in (7) and (8).
Both units start within their limits and if any of them is
violated during the iterations, they are set to the nearest limit.
The second type of mutual-dependent cogeneration units is
shown in Fig. 1. They are handled based on quadratic penalty
function to be augmented in the considered fitness function
(F) that can be expressed mathematically as follows:
NC
X
F = TFC + ψv BI .{PC C CLimit
k (Hk ) − Pk (HkC )} (19)
k=1
FIGURE 1. Dependency between power and heat for cogeneration unit. where; (ψv ) is a violation penalty coefficient of cogeneration
operating point, whilst (BI) is a binary term equals 1 if there
is violation or zero if not. The term (PCLimit
k (HkC )) is the
In the 2nd strategy, the iterations are similarly divided into relevant power limit to the set heating output of the cogen-
two parts. The first one improves the MRFOA exploration. eration k. Based on this formulation, the farthest violated
Then each individual is updated as described in (14-16). The operating point has greater penalty term and vice versa, how-
second part of the iterations in the cyclone foraging improves ever, the fulfilled operating point has no penalty as reflected
the MRFOA exploitation. Therefore, everyone is updated as in Figure 1. Figure 2 illustrates the main steps of the MRFOA
depicted in (17). for handling the formulated EPHD problem in CES.
(
∗ MR + (MR − Mm )(r + β) if m = 1
Mm = IV. SIMULATION RESULTS
MR + r(Mm−1 − Mm ) + β(MR − Mm ) else In this part, MRFOA is employed on three test instances. The
(14) first two test systems are investigated and analyzed includ-
min max min ing 5 units and 7 units for evaluating the performance of
MR = M + r.(M −M ) (15)
Itm −It+1 MRFOA. Additionally, 48-unit system is handled efficiently.
β = 2 er1 ( Itm ) .sin(2π r1 ) (16)
( It has an advantage of less parameters because it includes
MB + (MB − Mm )(r + β) if m = 1 just two main internal parameters which are the population
M∗m =
MB + r(Mm−1 − Mm ) + β(MB − Mm ) else size (PM ) and the number of iterations (Itm ). The optimal
(17) solution obtained when number of iterations and individuals
become 300 iterations and 50 individuals, respectively for the
where, MR expresses a created individual in a random way 5-unit and 7-unit systems, whereas number of iterations and
within the limited bounds. Moreover, β represents an adaptive individuals for 48-units system become 3000 iterations and
weight coefficient the symbol (It) reflects the current iteration 100 individuals, respectively. In this paper, the test systems
and the random uniformly distributed number, r1 , is in the are considered with lossless networks. This consideration is
range [0,1]. acceptable since the power losses in the system is very small
In the third strategy, each individual is updated around the compared to the demand.
best extracted position as follows:
A. SIMULATION RESULTS OF THE 5-UNIT TEST SYSTEM
M∗m = Mm + S(r2 MB − r3 Mm ) (18)
The first test system is small-sized that contains 1 conven-
where, S refers to the somersault coefficient (S = 2) that tional thermal power unit, 3 cogeneration energy units, and
controls the somersault domain of the manta rays. r2 and r3 1 heat-only unit. Two cases are discussed depending on the
are two random numbers defined within range [0, 1]. variation of both the load and heat demand. The data for this
For handling the EPHD problem in CES, two types of system is adopted from [12]. The two cases are given below
constraints, equality and inequality constraints, are generally as follows.
taken into account. These constraints are the as described
in the section. In this study, the loading of power and heat 1) CASE 1
are kept by considering the large unit as a slack unit. Thus, For this case, MRFOA is employed on 300 MW power
the outputs of power and heat of all units are created and demand and 150 MWth heat demand, where the opti-
updated outputs based on the MRFOA mechanism, while mal solution is compared with respect to those illus-
the slack units’ outputs are accommodated to satisfy the trated by various optimization techniques as in Table 1.
gap between the load and their aggregation. Consequently, It is apparently seen that the MRFOA finds a minimum

208284 VOLUME 8, 2020

A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

TABLE 1. Simulation results for 5-unit system-Case 1 with power demand of 300 MW and Heat demand of 150 MWth.

FIGURE 3. Violated operating point of CHP 1 of EDHS [9] and CPSO [15].

in Figures 3 and 4. From Figure 3, the operating points

that provided by EDHS [9] for cogeneration energy unit or
Combined Heat and Power (CHP) 1 is out of the limits, where
they account for (18.16 MW and 84.06 MWth). Additionally,
the operating points that provided by CPSO [12] for CHP 3,
as seen in Figure 4, are out of the limits, where they account
for (133.8 MW and zero thermal power).
Furthermore, Table 1 shows that the equality balance con-
straints based on the proposed MRFOA are effectively satis-
fied with 100% accuracy. The convergence characteristic of
MRFOA implementation on EPHD problem in CES is shown
in Figure 5 which demonstrates the consecutive improvement
of the optimal solution through the iterations.
FIGURE 2. Main steps of the MRFOA.

2) CASE 2
For this case, MRFOA is implemented with 250 MW
operational costs of 13673.13 $ that is lower than HS [8], power demand and 175 MWth heat demand and the opti-
firefly algorithm (FA) [42], TVAC-PSO [15], WVO [14], mal solution is compared with respect to those illus-
WVO-PSO [14], GSO [37], and IWO [21] where they trated by various optimization techniques as manifested
acquired TFC of 13724.24, 13683.22, 13692.52, 13723.2, in Table 2. It is apparently clear that the minimum opera-
13677.07, 13688.4836, and 13683.65 $, respectively. On the tional cost is obtained by the MRFOA which is 12117.72 $,
other side, it can be observed that EDHS [9] and CPSO [12] whereas HS [11], GA [11], CPSO [15], TVAC-PSO [15],
attained less TFC values compared to MRFOA, but their WVO [11],WVO-PSO [14], GSO [37], and IWO [21]
operating points were not feasible solutions as described acquired TFC of 12284.62, 12503.69, 12133.02, 12118.04,

VOLUME 8, 2020 208285

 A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

TABLE 2. Simulation results for 5-unit system-Case 2 with 250 MW power demand and 175 MWth heat demand.

FIGURE 4. Violated operating point of CHP 3 of EDHS [9] and CPSO [15].
FIGURE 6. Violated operating point of CHP 1 related to EDHS [11].

FIGURE 5. MRFOA convergence for 5-units system-case 1.

FIGURE 7. Violated operating point of CHP 2 related to EDHS [9].

12139.2, 12117.29, 12129.02, and 12134.33 $, respectively.

On the other hand, the obtained solutions based on EDHS [9], 85.85 MWth), (0 MW and 56.32 MWth), and (114.9 MW
WVO [14], and WVO-PSO [14] were not feasible solutions and 0 MWth) as shown in Figure 6, Figure 7, and Figure 8,
as depicted in Figures 6-9. For EDHS [9], the operating respectively.
points that provided are out of the limits for CHP 1, CHP 2, Furthermore, the operating point obtained by WVO and
and CHP 3, where they account for (0.1112 MW and WVO-PSO [14] are slightly less than the limit for CHP 2,

208286 VOLUME 8, 2020

A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

TABLE 3. Comparative results for 7-Unit System with power demand of 600 Mw and heat demand of 150 Mwth.

FIGURE 9. Violated operating point of CHP 2 related to WVO [14].

FIGURE 8. Violated operating point of CHP 3 related to EDHS [9].

Table 3 manifests the optimal solution based on MRFOA to

where they account for (10.05 MW and 40.12 MWth) as solve the EPHD problem in CES, while Figure 10 illustrates
displayed in Figure 9. Thus, the obtained results based on the its related convergence characteristics. It can be observed
MRFOA has the capability of providing an optimal and fea- from Table 3 and Figure 10 that the minimum TFC is
sible solution in terms of TFC compared with other methods. achieved to 10092.3 $ where all the operational limits
are kept. Additionally, Table 3 illustrates detailed com-
B. THE 7-UNIT TEST SYSTEM parison between the optimal solution based on MRFOA
This test instance includes 4 conventional thermal units, with those manifested by various optimization techniques.
1 heat-only units, and 2 CHP units, where the load demand This table illustrates the great effectiveness and superiority
and heat demand become 650 MW and 150 MWth, respec- of the proposed MRFOA in finding the minimum opera-
tively. The data for this test system is obtained from [14]. tional cost compared to EP [10], PSO [10], RCGA [31],

VOLUME 8, 2020 208287

 A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

FIGURE 12. Violated operating point of CHP 12 related to GSO [37].

FIGURE 10. MRFOA convergence for 7-units system.

FIGURE 13. Violated operating point of CHP 10 related to MGSO [37].

FIGURE 11. Violated operating point of CHP 2 related to LCA [33].

RCGA-IMM [18], CSO [25], AIS [32], CSO&PPS [22],

LCA [33], IGA-NCM [43], DE [10], BCO [31], CPSO [34],
TVAC-PSO [15], TLBO [34], TVAC-PSO [16], WVO [14],
and WVO-PSO [14]. It is lucid that the proposed MRFOA
achieves 100% accuracy for equality constraints that the sum-
mation of power and heat demands are precisely 600 MW and
150 MWth, respectively.
On the other side, the obtained solutions based on
LCA [33] is not feasible solutions as depicted in Figure 11,
where the operating points that provided are out of the limits
for CHP 2 and account for (40 MW and 74.5 MWth). Also,
the obtained result from the latter technique is not better than
the MRFOA technique.
FIGURE 14. Violated operating point of CHP 11 related to MGSO [37].
C. THE 48-UNIT TEST SYSTEM
For the 48-unit test system, the power and heat demands are
4700 MW and 2500 MWth, respectively. This test system for this test system is cusomised from [15]. The proposed
contains 26 conventional thermal power units, 10 heat-only MRFOA and other optimization techniques are employed on
units, and 12 cogeneration units, respectively. The data this case and the related the tabulated solutions in Table 4.

208288 VOLUME 8, 2020

A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

FIGURE 17. Violated operating point of CHP 11 related to OTLPO [34].

FIGURE 15. Violated operating point of CHP 12 related to MGSO [37].

FIGURE 16. Violated operating point of CHP 11 related to TLPO [34].

FIGURE 18. Violated operating point of CHP 2 related to GSA [38].

This table includes the results of a plethora of optimiza-

tion techniques, which are TVAC-PSO [15], CPSO [15], OTLBO [34], CPSO [12], TVAC_PSO [15], GSA [38], and
TVAC_PSO [15], CSO&PPS [22], TLBO [34], GSO [37], MPSO [12] overwhelmed the MRFOA for achieving min-
MGSO [37], OTLBO [34], GSA [38], and MPSO [12], imum TFC. However, by verifying the operating points of
as depicted in Table 4 (a), (b), and (c) respectively. such techniques, great violations of the operating points of
Table 5 tabulates the TFC objectives that are related to cogeneration energy units are obviously clear, where they are
the recorded operating points in Table 4. For comparative outside their permissible limits. The details of these violations
statistical purposes, several recent algorithms are adopted are stated as follows:
for the same algorithms as well like CSO [44], DE [45], • The obtained solution based on GSO [37] is not fea-
GWO [46], MVO [47] and SSA [48]. They are employed with sible solution because the operating point that pro-
the same number of individuals and iterations for 10 separate vided for CHP 12 is out of the limits as depicted
times where the Best, Mean, Worst and Standard Deviation in Figure 12.
are estimated and recorded in Table 5. As shown in this table, • Another violation is noticed using MGSO [37] as the
the proposed MRFOA minimizes the total operational costs operating points provided for CHP 10, CHP 11, and CHP
of the system to 117336.9$. It clearly reflects the effective- 12 are out of the limits as illustrated in Figs.13-15.
ness and superiority of the proposed MRFOA compared to the • The operating points that provided for CHP 12 of both
other optimizers not only in achieving the minimum best TFC TLBO [34] and OTLBO [34] are out of the limits as
value but also for all the indices. In addition to this, it pro- depicted in Figures 16 and 17, respectively.
vides great robustness with the smallest standard deviation • The operating points that provided for CHP 2 and CHP
of 450.7414. It is manifested from Table 5 that the results 4 of GSA [38] are out of the limits as manifested in
of GSO [37], MGSO [37], CSO&PPS [22], TLBO [34], Figures 18 and 19, respectively.

VOLUME 8, 2020 208289

 A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

TABLE 4. Comparative results for 48-unit system with power demand of 4700 MW and heat demand of 2500 Mwth.

208290 VOLUME 8, 2020

A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

TABLE 5. Summarized Comparative Results for all power and heat demand based on the various optimization techniques for 48-unit system.

FIGURE 19. Violated operating point of CHP 4 related to GSA [38]. FIGURE 21. Violated operating point of CHP 8 related to MPSO [12].

FIGURE 20. Violated operating point of CHP 1 related to MPSO [12].

FIGURE 22. Violated operating point of CHP 10 related to MPSO [12].

• The operating points that provided for CHP 1, CHP 8 Despite CPSO [15] and TVAC_PSO [15] give feasi-
and CHP 10 of MPSO [12] are outside the permissible ble operating points of the units, there is considerable
limits as demonstrated in Figures 20-22, respectively. deviation between the reported TFC value of 119708.8818 $

VOLUME 8, 2020 208291

 A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

FIGURE 23. Compartive convergence rates for 48-units system.

FIGURE 24. Operating points of power only units for 48-unit system.

and 117824.8956 $ and the calculated TFC value In addition, Figure 23 displays the convergence characteristic
of 120918.918 $ and 118962.5402 $, respectively. Further- of MRFOA compared to as CSO, DE, GWO, MVO and SSA.
more, for CSO&PPS [22], there is a mismatch of 1562.3 kW This curve declares the great convergence rate of the MRFOA
in the balance constraint of the power load Sum(pg) where the compared to the others for solving the EPHD problem in CES.
value is 4698.4377 MW but the required load is 4700 MW Furthermore, the effectiveness of the proposed technique
which makes the TFC value of CSO&PPS [22] less than in finding optimal and feasible operating points for bigger
the developed MRFOA. Therefore, the comparison of the systems is validated. In Fig. 23, all the optimization methods
proposed MRFOA with other techniques, which are not are coded and applied with the same stopping condition when
violated due to the heat-power dependency characteristics, the maximum number of iterations is reached. This figure
confirms the superiority of the proposed algorithm to achieve shows that some methods may get better results if the iteration
the desired objective without any violation of any constraint. number is increased. However, this argument elucidates the

208292 VOLUME 8, 2020

A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

FIGURE 25. Operating points of heat only units for 48-unit system.

FIGURE 26. Operating points of cogeneration units for 48-unit system.

quicker convergence of the proposed MRFOA. The superi- Similarly, all the cogeneration units operate within their
ority of the proposed MRFOA is also shown, as it has the acceptable ranges as indicated in Figure 26.Finaly, in the
smallest standard deviation of 450,74 relative to the others. view point of economical point of view, Table 6 shows the
Figures 24 and 25 illustrate that all the power only units annual cost savings of the proposed MRFOA compared to
and heat only units operate within their acceptable ranges. the maximum reported TFC for the three studied systems.
VOLUME 8, 2020 208293
 A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

TABLE 6. Annual savings for all tested systems by MRFOA. REFERENCES

[1] E. E. Elattar, ‘‘Environmental economic dispatch with heat optimization in
the presence of renewable energy based on modified shuffle frog leaping
algorithm,’’ Energy, vol. 171, pp. 256–269, Mar. 2019.
[2] J. Yu, L. Guo, M. Ma, S. Kamel, W. Li, and X. Song, ‘‘Risk assessment of
integrated electrical, natural gas and district heating systems considering
solar thermal CHP plants and electric boilers,’’ Int. J. Electr. Power Energy
Syst., vol. 103, pp. 277–287, Dec. 2018.
[3] T. Jayabarathi, A. Yazdani, and V. Ramesh, ‘‘Application of the invasive
weed optimization algorithm to economic dispatch problems,’’ Frontiers
Energy, vol. 6, no. 3, pp. 255–259, Aug. 2012.
[4] X. Huang, Z. Xu, Y. Sun, Y. Xue, Z. Wang, Z. Liu, Z. Li, and W. Ni, ‘‘Heat
and power load dispatching considering energy storage of district heating
system and electric boilers,’’ J. Mod. Power Syst. Clean Energy, vol. 6,
no. 5, pp. 992–1003, Jan. 2018.
[5] M. A. Gonzalez Chapa and J. R. V. Galaz, ‘‘An economic dispatch algo-
Also, this tables reports the average computational time for rithm for cogeneration systems,’’ in Proc. IEEE Power Eng. Soc. Gen.
the 5-, 7- and 48 test systems as 42.42, 44.306 and 428.72 sec, Meeting, 2004, pp. 989–994.
respectively. [6] A. Sashirekha, J. Pasupuleti, N. H. Moin, and C. S. Tan, ‘‘Combined heat
and power (CHP) economic dispatch solved using lagrangian relaxation
with surrogate subgradient multiplier updates,’’ Int. J. Electr. Power Energy
V. CONCLUSION Syst., vol. 44, no. 1, pp. 421–430, Jan. 2013.
This paper has been developed for handling the EPHD [7] T. Guo, M. I. Henwood, and M. V. Ooijen, ‘‘An algorithm for combined
optimization problem in CES. The EPHD problem in CES heat and power economic dispatch,’’ IEEE Trans. Power Syst., vol. 11,
no. 4, pp. 84–1778, 1996.
is solved considering various constrains such as capacity [8] C. Lin, W. Wu, B. Zhang, and Y. Sun, ‘‘Decentralized solution for com-
limits of the generation units, the impact of valve-point bined heat and power dispatch through benders decomposition,’’ IEEE
effects of conventional thermal power plants, and con- Trans. Sustain. Energy, vol. 8, no. 4, pp. 1361–1372, Oct. 2017.
[9] E. Khorram and M. Jaberipour, ‘‘Harmony search algorithm for solving
straints on power-heat dependence of the CES. In this paper, combined heat and power economic dispatch problems,’’ Energy Convers.
the MRFOA is handled effectively the equality and inequality Manage., vol. 52, no. 2, pp. 1550–1554, Feb. 2011.
operational limits via adaptive penalty parameter to penalize [10] M. Basu, ‘‘Combined heat and power economic dispatch by using differen-
tial evolution,’’ Power Compon. Syst., vol. 38, no. 8, pp. 996–1004, 2017.
the infeasible solutions with different degrees based on its
[11] A. Vasebi, M. Fesanghary, and S. M. T. Bathaee, ‘‘Combined heat and
remoteness from the nearest feasible point. This penalty con- power economic dispatch by harmony search algorithm,’’ Int. J. Electr.
straint incorporation guarantees achieving feasible operating Power Energy Syst., vol. 29, no. 10, pp. 713–719, Dec. 2007.
point and making use of the high searching characteristics of [12] M. Basu, ‘‘Modified particle swarm optimization for non-smooth non-
convex combined heat and power economic dispatch,’’ Electr. Power Com-
the MRFOA. It is applied effectively on three test instances pon. Syst., vol. 43, no. 19, pp. 2146–2155, Nov. 2015.
including 5-unit, 7-unit, and 48-unit systems with different [13] P. Subbaraj, R. Rengaraj, and S. Salivahanan, ‘‘Enhancement of combined
power and thermal demand. The delivered solutions for the 5-, heat and power economic dispatch using self adaptive real-coded genetic
algorithm,’’ Appl. Energy, vol. 86, no. 6, pp. 915–921, Jun. 2009.
7-, and 48-unit test systems are compared with those reported [14] S. Dolatabadi, R. A. El-Sehiemy, and S. GhassemZadeh, ‘‘Scheduling of
in the recent studies.That is because the results were not the combined heat and generation outputs in power systems using a new hybrid
same as those that mentioned the literature, where some oper- multi-objective optimization algorithm,’’ Neural Comput. Appl., vol. 32,
no. 14, pp. 10741–10757, Nov. 2019.
ating point obtained of some cogeneration units are outside [15] B. Mohammadi-Ivatloo, M. Moradi-Dalvand, and A. Rabiee, ‘‘Combined
the feasible region. Besides, the superiority of the MRFOA in heat and power economic dispatch problem solution using particle swarm
dealing with constrained and non-convex EPHDoptimization optimization with time varying acceleration coefficients,’’ Electr. Power
Syst. Res., vol. 95, pp. 9–18, Feb. 2013.
issue in CES is demonstrated. It achieves the minimum TFC [16] Y. A. Shaabani, A. R. Seifi, and M. J. Kouhanjani, ‘‘Stochastic multi-
for the 5-unit and 7-unit systems among the other techniques. objective optimization of combined heat and power economic/emission
As well, it acquires the minimum TFC for the 48-unit sys- dispatch,’’ Energy, vol. 141, pp. 1892–1904, Dec. 2017.
[17] M. Nazari-Heris, M. Mehdinejad, B. Mohammadi-Ivatloo, and
tem with outperformed robustness as it provides the smallest G. Babamalek-Gharehpetian, ‘‘Combined heat and power economic
standard deviation compared to the others. Also, the obtained dispatch problem solution by implementation of whale optimization
operating point based on the MRFOA is verified within the method,’’ Neural Comput. Appl., vol. 31, no. 2, pp. 421–436, Jun. 2017.
[18] A. Haghrah, M. Nazari-Heris, and B. Mohammadi-Ivatloo, ‘‘Solving com-
feasible region for all power only, heat only units, and cogen- bined heat and power economic dispatch problem using real coded genetic
eration units. Therefore, the obtained optimal solutions for algorithm with improved Mühlenbein mutation,’’ Appl. Thermal Eng.,
the three test instances of EPHD problem in CES are demon- vol. 99, pp. 465–475, Apr. 2016.
[19] X. Yu and D. Zheng, ‘‘Cross-regional integrated energy system scheduling
strated and the convergence characteristics are manifested, optimization model considering conditional value at risk,’’ Int. J. Energy
which reflect the superiority and effectiveness of the MRFOA Res., vol. 44, no. 7, pp. 5564–5581, Jun. 2020.
compared with other optimization techniques when handling [20] C. Lin, W. Wu, B. Wang, and Y. Sun, ‘‘Combined heat and power dispatch
with start-stop schedule of heat exchange stations,’’ in Proc. IEEE Power
practical systems. Energy Soc. Gen. Meeting (PESGM), Aug. 2018, pp. 1–5.
[21] T. Jayabarathi, A. Yazdani, V. Ramesh, and T. Raghunathan, ‘‘Combined
ACKNOWLEDGMENT heat and power economic dispatch problem using the invasive weed opti-
The authors would like to acknowledge the financial sup- mization algorithm,’’ Frontiers Energy, vol. 8, no. 1, pp. 25–30, Nov. 2013.
port received from Taif University Researchers Supporting [22] N. Narang, E. Sharma, and J. S. Dhillon, ‘‘Combined heat and power
economic dispatch using integrated civilized swarm optimization and Pow-
Project Number (TURSP-2020/34), Taif University, Taif, ell’s pattern search method,’’ Appl. Soft Comput., vol. 52, pp. 190–202,
Saudi Arabia. Mar. 2017.

208294 VOLUME 8, 2020

A. M. Shaheen et al.: EPHD in CES Using Manta Ray Foraging Optimizer

[23] Y. Zhou, W. Hu, Y. Min, and Y. Dai, ‘‘Integrated power and heat dispatch [46] A. M. Shaheen and R. A. El-Sehiemy, ‘‘Optimal co-ordinated allo-
considering available reserve of combined heat and power units,’’ IEEE cation of distributed generation units/capacitor banks/voltage regula-
Trans. Sustain. Energy, vol. 10, no. 3, pp. 1300–1310, Jul. 2019. tors by EGWA,’’ IEEE Syst. J., early access, May 6, 2020, doi:
[24] Y. Gao, D. Zeng, L. Zhang, Y. Hu, and Z. Xie, ‘‘Research on modeling and 10.1109/JSYST.2020.2986647.
deep peak regulation control of a combined heat and power unit,’’ IEEE [47] A. M. Shaheen and R. A. El-Sehiemy, ‘‘Application of multi-verse opti-
Access, vol. 8, pp. 91546–91557, 2020. mizer for transmission network expansion planning in power systems,’’
[25] A. Meng, P. Mei, H. Yin, X. Peng, and Z. Guo, ‘‘Crisscross optimization in Proc. Int. Conf. Innov. Trends Comput. Eng. (ITCE), Feb. 2019,
algorithm for solving combined heat and power economic dispatch prob- pp. 371–376.
lem,’’ Energy Convers. Manage., vol. 105, pp. 1303–1317, Nov. 2015. [48] A. M. Shaheen and R. A. El-Sehiemy, ‘‘A multiobjective salp optimiza-
[26] N. Jayakumar, S. Subramanian, S. Ganesan, and E. B. Elanchezhian, ‘‘Grey tion algorithm for techno-economic-based performance enhancement of
wolf optimization for combined heat and power dispatch with cogener- distribution networks,’’ IEEE Syst. J., early access, Jan. 27, 2020, doi:
ation systems,’’ Int. J. Electr. Power Energy Syst., vol. 74, pp. 252–264, 10.1109/JSYST.2020.2964743.
Jan. 2016.
[27] Z. Li, W. Wu, M. Shahidehpour, J. Wang, and B. Zhang, ‘‘Combined heat ABDULLAH M. SHAHEEN was born in Tanta,
and power dispatch considering pipeline energy storage of district heating Egypt, in 1985. He received the B.Sc. degree
network,’’ in Proc. IEEE Power Energy Soc. Gen. Meeting, Jul. 2017, p. 1. from Suez Canal University, Port-Said, Egypt,
[28] M. T. Hagh, S. Teimourzadeh, M. Alipour, and P. Aliasghary, ‘‘Improved in 2007, and the M.Sc. and Ph.D. degrees from
group search optimization method for solving CHPED in large scale power Menoufia University, Shebin El-Kom, Egypt,
systems,’’ Energy Convers. Manage., vol. 80, pp. 446–456, Apr. 2014. in 2012 and 2016, respectively. He is cur-
[29] M. I. Alomoush, ‘‘Optimal combined heat and power economic dispatch rently with the Department of Electrical Engi-
using stochastic fractal search algorithm,’’ J. Mod. Power Syst. Clean neering, Faculty of Engineering, Suez University,
Energy, vol. 8, no. 2, pp. 276–286, 2020. El-Suweis, Egypt. His research interests include
[30] A. Sundaram, ‘‘Combined heat and power economic emission dispatch power system operation, control, and planning,
using hybrid NSGA II-MOPSO algorithm incorporating an effective con- the applications of optimization algorithms in electric power systems, renew-
straint handling mechanism,’’ IEEE Access, vol. 8, pp. 13748–13768, able integration, and smart grids.
2020.
[31] M. Basu, ‘‘Bee colony optimization for combined heat and power eco- AHMED RABIE GINIDI received the B.Sc.
nomic dispatch,’’ Expert Syst. Appl., vol. 56, pp. 13527–13531, Mar. 2011. degree in electrical engineering from Fayoum Uni-
[32] M. Basu, ‘‘Artificial immune system for combined heat and power eco- versity, Fayoum, Egypt, in 2007, and the M.Sc.
nomic dispatch,’’ Int. J. Electr. Power Energy Syst., vol. 43, no. 1, pp. 1–5, and Ph.D. degrees in electrical engineering from
Dec. 2012. Cairo University, Cairo, Egypt, in 2010 and 2015,
[33] B. Shi, L.-X. Yan, and W. Wu, ‘‘Multi-objective optimization for combined respectively. He is currently an Assistant Pro-
heat and power economic dispatch with power transmission loss and fessor of Electrical Engineering with Suez Uni-
emission reduction,’’ Energy, vol. 56, pp. 135–143, Jul. 2013. versity, Suez, Egypt. His main interests include
[34] P. K. Roy, C. Paul, and S. Sultana, ‘‘Oppositional teaching learning based system engineering, automatic control, intelligent
optimization approach for combined heat and power dispatch,’’ Int. J. systems, smart grids, power systems operation,
Electr. Power Energy Syst., vol. 57, pp. 392–403, May 2014. fuzzy systems, energy policy, optimization and integration of renewable
[35] M. Nazari-Heris, B. Mohammadi-Ivatloo, S. Asadi, and Z. W. Geem, energy sources, and operations research.
‘‘Large-scale combined heat and power economic dispatch using a novel
multi-player harmony search method,’’ Appl. Thermal Eng., vol. 154, RAGAB A. EL-SEHIEMY (Senior Member, IEEE)
pp. 493–504, May 2019. was born in Menoufia, Egypt, in 1973. He received
[36] B. Deng, Y. Teng, Q. Hui, T. Zhang, and X. Qian, ‘‘Real-coded quantum the B.Sc. and M.Sc. degrees in electrical engi-
optimization-based bi-level dispatching strategy of integrated power and neering and Ph.D. degree in electrical power sys-
heat systems,’’ IEEE Access, vol. 8, pp. 47888–47899, 2020. tems from Menoufia University, Egypt, in 1996,
[37] E. Davoodi, K. Zare, and E. Babaei, ‘‘A GSO-based algorithm for com- 2005, and 2008, respectively. He is currently a
bined heat and power dispatch problem with modified scrounger and ranger Professor of Electrical Power Systems with the
operators,’’ Appl. Thermal Eng., vol. 120, pp. 36–48, Jun. 2017. Electrical Engineering Department, Faculty of
[38] S. D. Beigvand, H. Abdi, and M. La Scala, ‘‘Combined heat and power Engineering-Kafrelsheikh University, Egypt. His
economic dispatch problem using gravitational search algorithm,’’ Electr. research interests include power systems opera-
Power Syst. Res., vol. 133, pp. 160–172, Apr. 2016.
tion, planning and control, smart grids, renewable energy, AI, and its appli-
[39] W. Zhao, Z. Zhang, and L. Wang, ‘‘Manta ray foraging optimization: An
cation to power systems. He was a receipt of the Prof. Mahmoud Khalifia
effective bio-inspired optimizer for engineering applications,’’ Eng. Appl.
Award in Power System Engineering from the Academy of Research and
Artif. Intell., vol. 87, Jan. 2020, Art. no. 103300.
Technology, in 2016. He has published more than 110 articles in International
[40] E. E. Elattar, A. M. Shaheen, A. M. Elsayed, and R. A. El-Sehiemy, ‘‘Opti-
mal power flow with emerged technologies of voltage source converter sta-
journals and conferences. He is an Editor of the International Journal of
tions in meshed power systems,’’ IEEE Access, vol. 8, pp. 166963–166979, Engineering Research in Africa.
2020. SHERIF S. M. GHONEIM (Senior Member,
[41] B. Mohammadi-Ivatloo, A. Rabiee, and A. Soroudi, ‘‘Nonconvex dynamic IEEE) received the B.Sc. and M.Sc. degrees from
economic power dispatch problems solution using hybrid immune-genetic the Faculty of Engineering, Shoubra, Zagazig Uni-
algorithm,’’ IEEE Syst. J., vol. 7, no. 4, pp. 777–785, Dec. 2013.
versity, Egypt, in 1994 and 2000, respectively, and
[42] A. Yazdani, T. Jayabarathi, V. Ramesh, and T. Raghunathan, ‘‘Combined
the Ph.D. degree in electrical power and machines
heat and power economic dispatch problem using firefly algorithm,’’ Fron-
from the Faculty of Engineering, Cairo University,
tiers Energy, vol. 7, no. 2, pp. 133–139, May 2013.
in 2008. Since 1996, he has been teaching with
[43] D. Zou, S. Li, X. Kong, H. Ouyang, and Z. Li, ‘‘Solving the combined
heat and power economic dispatch problems by an improved genetic the Faculty of Industrial Education, Suez Canal
algorithm and a new constraint handling strategy,’’ Appl. Energy, vol. 237, University, Egypt. From the end of 2005 to the
pp. 646–670, Mar. 2019. end of 2007, he was a Guest Researcher with the
[44] A. M. Shaheen and R. A. El-Sehiemy, ‘‘Optimal allocation of capacitor Institute of Energy Transport and Storage (ETS), University of Duisburg–
devices on MV distribution networks using crow search algorithm,’’ Open Essen, Germany. He joined as an Associate Professor with the Electri-
Access Proc. J., vol. 2017, no. 1, pp. 2453–2457, Oct. 2017. cal Engineering Department, Faculty of Engineering, Taif University. His
[45] A. M. Shaheen, R. A. El-Sehiemy, and S. M. Farrag, ‘‘A reactive power research areas include grounding systems, dissolved gas analysis, breakdown
planning procedure considering iterative identification of VAR candidate in SF6 gas, and AI technique applications.
buses,’’ Neural Comput. Appl., vol. 31, no. 3, pp. 653–674, Mar. 2019.

VOLUME 8, 2020 208295