Research Journal of Applied Sciences, Engineering and Technology 4(9): 1062-1066, 2012
ISSN: 2040-7467
© Maxwell Scientific Organization, 2012
Submitted: October 31, 2011
Accepted: November 18, 2011
Published: May 01, 2012
Optimal Parameter Selection of Gas-Fired Units for Distributed Cogeneration
in an Industrial Town Based on an Industrial Service Company Perspective
Meisam Ansari, Mohammad Sadegh Payam, Mostafa Abdollahi and Asadollah Salami Dehkordi
Department of Electrical Engineering, Boroujen Branch, Islamic Azad University, Boroujen, Iran
Abstract: Industrial service companies incentivize the investors to install distributed generation, to improve
network environmental performance and efficiency. This paper proposes an approximated dynamic approach
for optimal parameter selection of gas-fired units to supply electrical and thermal loads in an industrial town
considering industrial service company perspective. A cost-benefit analysis with considering gas units, heat
recovery system and distribution network constrains is performed for Optimal sizing, placement and timing for
CHP installation. The main objective of the model is to minimize the company’s operating costs in five years.
Two aspects are considered in problem formulation: priority of the heat generation or electricity generation.
Also, a multi objective genetic algorithm is used to solve the problem. Moreover, the PIKO feeder, one of the
MANAVI substation feeders in TEHRAN, is employed as the study system. The simulation results show a
significant decrease in the industrial service company operating costs, especially loss costs in five years.
Key words: Distributed cogeneration, distributed network reinforcement, genetic algorithm
INTRODUCTION
Distributed systems are used commonly in the
networks to decrease transmission and distribution losses.
Using cogeneration systems, with total efficiency more
than %70, is one of the effective strategies for optimizing
energy consumption. However, if require parameters
don’t select properly, it may cause to decrease the planes
performance. Developing in technology has created a
wide choice especially in Combine Heat and Power
(CHP) units that can be competing in costs decrease. In
this case, Distributed Generation (DG) installation
problem is formulated with various objective functions
and is solved using several methods. In (Hashemi, 2009),
an offline model is developed by authors for optimal
operation in a three generation system (combine heat and
power and cooling system). Where, the objective function
is maximizing the annual benefit of the CCHP systems. In
(Sanaye and Reaessi, 2009) the authors are introduced a
method to estimate optimal capacity and number of gas
micro-turbines to use in a system with specific electrical
and thermal load curve. The objective function is
maximizing total benefit from selling power and heat. In
(Giaccone and Canova, 2009) a method is proposed to
specify a cogeneration system for industrial users while,
the objective function is maximizing net present value. In
(Sun, 2008) a method to determine the CHP
systemparameters from a manufactures perspective is
presented. In this paper, the objective function is
maximizing Internal Rate of Return (IRR). Also, in
(Favuzza et al., 2007) an approach is suggested to
reinforce distributed network by DG installation and a
dynamic ant colony optimization algorithm is used to
solve the problem. In Keane et al. (2007) and Harrison
et al. (2007) the authors are expressed the impact of DG
interconnection policy on distributed network and power
purchasing quality from DGs. In these studies, the
interaction between distributed generation developer and
distributed network operator is evaluated.
In this study, the expansion of CHP units in an
industrial town is developed. The problem is formulated
as an optimization problem based on industrial service
company perspective. The problem is solved using a
dynamic method to minimize the operation cost in
several years.
PROBLEM FORMULATION
In general view, it’s required to evaluate CHP
installation problem in several aspects such as technical
constraints, units characteristics, environmental affects,
economical evaluations and etc. The most important
parameters required to determine a cogeneration plane in
a distributed network, are placement, timing and capacity
of gas-fired units. The problem is definable based on
investor perspective, industrial company perspective or
local consumer perspective. In this study, the problem is
defined based on distributed company perspective.
Control variables, constraints and objective function are
three main parts that should be specified to formulate an
optimization problem.
Corresponding Author: Mohammad Sadegh Payam, Department of Electrical Engineering, boroujen branch, Islamic Azad
University, Boroujen, Iran, P.O. Box 88715/141, Tel.: +983824223812; Fax: +983824223812
1062
Res. J. Appl. Sci. Eng. Technol., 4(9): 1062-1066, 2012
Control variables include place, time and capacity of
gas-fired units. In this study, an approximated dynamic
approach is used to determine these variables. The
problem is solved for one year in this method. Then the
achieved plan is considered and the problem solve for
next year. This process continues until completion of
study. It is necessary to estimate the Load Duration Curve
(LDC) into several steps for mathematical formulation.
Objective function: The industrial companies add units
to their networks with operation cost decreasing purpose.
Thus the objective function defines by decreasing costs
purpose for N years. These costs include following items.
Cost of purchasing power from network: The power
price curve estimate into N steps similar to LDC.
Equation (1) shows the calculation of this cost:
f
NET 
pur ( n ,t )  (1  rN ) n . c N (t ).[ E load (n, t )  E CHP (n, t )]
E load (n, t )  (1  re ) n . PL(t ). h(t )
(1)
f ENS (n, t ) 
h( t )
 U .[(1  rN ) n . PLi (t )
8760 i i
max
 min{(1  rN ) n . PLi (t ), PCHP
i }].
Cs (t ).(1  rs )
(5)
n
Therefore, we can calculate the ENS in distributed
network without islanding possibility, before and after DG
installation.
Heat recovery revenue: This revenue belongs to DG
owners but it cause to give priority to the buses with
thermal load for install DG by subtracting this item from
total cost. There are two views in considering this item:
In the first view, the thermal load is on priority. In the
other word, the main goal of DG installation is supply
local thermal load and generated power can be used by
local electrical load or can be sold to the network. Total
thermal required by local load must supply by units. In
this view, heat recovery is equal to local thermal load and
it can be calculated by following equation:
E CHP (n, t )   i 1 PCHPi (n, t ). h(t ). KCHP
nCHP
R rec ( n , t ) 
U
n
bf
. C f .(1  rh ) . QDi ( t ). h ( t ). K CHP (6)
i
Cost of power purchasing from installed units:
Considering legal rules in IRAN, the DGs can sell their
power with bilateral contract by guaranty rate or
participate in the power market and compete with other
gencos. In this paper, bilateral contract by guaranty rate is
used to define the problem. Equation (2) shows the
formulation of this cost:
f CHP pur (n, t )  (1  rc )n. Cc (t ). E CHP (n, t ) (2)
In the second view, DG units will be installed on the
network to generate the power and selling to the network
or local electrical load. In this case heat recovery can be
used from local thermal load. Total thermal recovered
maybe cannot supply total requirement and then we have
to use some boilers to provide total requirement. Heat
recovery is equal or less than local consumption anyway.
Eq. (7) shows calculation of this parameter:
Rrec (n, t ) 
Loss cost: Equation (3) shows the calculation of loss cost:
f loss (n, t )  CN (t ).(1  rN ) n . Eloss (n, t )
nbus
nbus
1
Eloss (n, t )  h(t ). j 1
Rij . I ij2 (n, t )
j 1
2
 
h( t )
 U .(1  re ) n . PLi (t ). Cs (t ).(1  rs )n
8760 i i
ij . r j
Ui 

n
h
i
CHPi ( n, t )}.
(7)
i
U bf .C f .h(t ). KCHP
(3)
Finally total annual cost can be calculated using following
equations:
Cost of energy not supply: Equation (4) use to estimate
the Energy Not Supply (ENS) when there are no DG units
in network and it’s not possible to make island in
abnormal condition:
f ENS (n, t ) 
 min{(1  r ) .QD (t ),Q
N 1
tot  cos t 
  cos t (n, t ).(1  I )
n 0
N n
t
cos t (n, t )  f NET pur (n, t )  f
CHP pur
(n, t ) (8)
 f loss (n, t )  f ENS (n, t )  Rrec (n, t )
(4)
j
when DG units exist in the network and making island is
not possible, each unit can just supply the local load
connected to it. In this case Eq. (5) can be used to
estimate the ENS:
Constraints: The constraints include 3 parts: network
constraints, DG constraints and heat recovery system
constraints. Equation (9) shows network constraints.
These formulas define optimal power flow for the
network:
1063
Res. J. Appl. Sci. Eng. Technol., 4(9): 1062-1066, 2012
The second part is about the heat recovered limitation for
DG units, which shown as:
QCHPi (n, t )  kheat . PCHPi (n, t )
Units will be installed by two perspectives, if local
thermal load exist. This subject is expressed in section
1-3. Equation (12) and (13) show these constraints for
thermal priority and electrical priority respectively:
Fig. 1: Diagram for PIKO feeder
Table 1: Required parameter to solve the problem
rn
rh
rc
0.05
0.05
0.02
Cf
Ubf
ra
0.05
49
0.25
PFkmin
N
khrat
1.1
0.85
5
T
PL
H
Cn
1
3339
730
0.075
2
3172
730
0.072
3
3005
730
0.069
4
2838
730
0.066
5
2671
730
0.063
6
2504
730
0.060
7
2337
730
0.057
8
2170
730
0.053
9
2003
730
0.050
10
1836
730
0.047
11
1670
730
0.044
12
1503
730
0.041
Cc
0.048
0.048
0.048
0.048
0.048
0.048
0.048
0.048
0.048
0.048
0.048
0.048
rc
0.03
KCMP
0.9
I
0.12
Cs
0.075
0.072
0.069
0.066
0.063
0.060
0.057
0.053
0.050
0.047
0.044
0.041
 VV (G sin   B cos )
P  P   VV (G cos  B sin  )
1
 PG   PL  2   P . I
i
j
ij
ij
i
j
ij
ij
ij
Li
i 1
n bus
i 1
ij
i
i
i 1
i 1
j 1
ij
2
2
ij
0
(9)
Units constraints include two basic parts. First part is
electrical constraints. Equation (10) shows constraints for
this part:
Case study: The PIKO feeder, one of the MANAVI
substation feeders in TEHRAN, is considered as a case
study. The diagram of this feeder is shown in Fig. 1.
Network information, lines characterize and breakers
locations are illustrated in Fig. 1. Other parameters that
need to simulate the problem shown in Table 1.
SIMULATION RESULTS
min
max
PCHPi
 PCHPi  PCHPi
PCHPi
PCHPi  q CHPii
2
 PF min
(15)
constraints (c(x)#0) and it’s defined by (a = *d(x)* for
equality constraints (d(x) = 0). The algorithm may loss its
basic aim if penalty factors are very effective. We use
“it” to prevent this. This parameter decrease linear by the
algorithm iteration and its value is in [0 1] interval.
V min  Vi  V max
2
(14)
The “a” is defined by a  c( x )  c( x ) for inequality
/ Fij /  Fijmax
min
max
qCHPi
 qCHPi  qCHPi
cos ti
P rimary  cos t
penalty  fact  e it .a
ij
n bus
(13)
A penalty factor method is used to consider constraints in
the algorithm. These factors are multiplied by the fitness
function. Equation (15) shows that how the constraints are
considered.
ij
n bus
n bus
QCHPi (n, t )  Qboiler i (n, t )  QDi (t ).(1  rh ) n
fiti  1 
n bus
Gi
(12)
Performance by GA: A standard genetic algorithm is
used to solve this problem. Equation (14) shows the
fitness function for each chromosome.
n bus
i 1
QCHPi (n, t )  QDi (t ).(1  rh ) n
In (13), if Qboiler i is positive, this means that we need
auxiliary boiler to supply local thermal load and if its
negative, this means that heat recovered from units can
supply local thermal load.
Table 2: optimal plan specification in 5 years
Capacity (Kw)
----------------------------------------------------------------------No. Bus Year 1
Year 2
Year 3
Year 4
Year 5
6
848
0
0
0
0
12
0
0
0
729
0
13
1060
0
0
0
0
14
0
1131
0
0
0
Total
1908
3039
3039
3768
3768
qGi  q Li 
(11)
(10)
Table 2 shows optimal plan for gas-fired unit’s
installation in 5 next years. As it can be seen, the units are
installed at first, second and third years. Total capacity are
installed in the first year, is 848 kW in bus 6 and 1060 kW
in bus 13. Then at the second year, 1314 kW capacity
1064
Res. J. Appl. Sci. Eng. Technol., 4(9): 1062-1066, 2012
457.45
Year 1
Year 4
851.02
457.40
851.00
850.98
457.30
Mwh
Mwh
457.35
457.25
457.20
850.94
850.92
457.15
850.90
457.10
1
2
3
4
5
6
7
8
9
10
11 12
850.88
1
851.10
2
3
4
5
6
7
8
9
10
11 12
9
10
11 12
Year 2
851.10
851.05
Year 5
851.05
851.00
Mwh
Mwh
850.96
850.95
851.00
851.95
850.90
851.90
850.85
1
2
3
4
5
6
7
8
9
10
11 12
851.85
2000
1
Year 3
1500
2
3
4
5
6
7
8
Annual purchased energy from units
Mwh
Years
1000
1
2
3
4
5
500
0
1
2
3
4
5
6
7
8
9
10
11 12
Energy [Mwh]
5487.4
10211.6
10211.6
10211.7
18946.2
Fig. 2: Energy purchasing from units in each step
Table 3: Comparing parameters before and after units installation for
5 years (all parameters in Mwh)
without DG
with DG
Loss
240.1
104.4
purchasing from net
112820.5
57616.3
purchasing from DG
0.0
55068.5
Energy not supply
18.2
11.8
Heat recovered
0.0
49808.1
Total
0
0
Table 4: Comparing costs before and after units installation for 5 years
(all costs in 1000 Euros)
without DG with DG
Improvement (%)
Loss
18.9
9.0
52.3
purchasing from net 6827.9
3486.9
48.9
purchasing from DG 0.0
2615.8
0
Energy not supply
1.4
0.9
32.7
Heat recovered
0.0
716.0
0
Total
6846.8
6111.7
10.7
is installed in bus 14. Finally the plan is finished by
installing 729 kW in bus 12 at third year .Thus developing
units plan complete for 5 next years. Total electrical
capacity is installed on the feeder is 3768 kW.
Considering Fig. 1 and Table 2, the buses are selected for
installing DG are ended buses and have thermal
consumption. This means ended buses are suitable for DG
development.
Table 3 shows comparing parameters before and after
units’ installation for 5 years (all parameters in Mwh).
Also, Table 4 shows operation costs before and after
developing units in the network. As can be seen, the total
loss improvement is 82% in five years. Its equivalent
9.9e3 Euros considering power price in IRAN. ENS is
decreased about 54% in five years. It is equivalent 500
Euros. ENS is total energy that industrial company buys
from market but cannot sells to costumers because of
faults occur in its network. This value maybe increases, if
industrial companies have to pay fine to costumers
because of interruption service. After implementation the
plan, thermal recovered is more than 49 GWH equivalent
716e3 Euros. This reduction in fuel consumption is very
impressive. Comparing numerical results shows that the
most revenue of using CHP in network is heat recovery.
Finally total operation costs is reduced more than 14%
equivalent 735e3 Euros in five years.
Figure 2 shows energy purchased from gas-fired units
in each years. As can be seen, in the first steps -related to
full load hours- energy purchased is more than latest steps
related to light load hours. Because in the full load hours
market price is higher than billateral contracts rate and in
the light load hours, it is lower than bilateral contracts
rate.
1065
Res. J. Appl. Sci. Eng. Technol., 4(9): 1062-1066, 2012
Mwh
60
without DGs
with DGs
40
20
0
1
2
4
5
Fig. 3: Comparing between annual loss before and after units
installation
Mwh
6
without DGs
with DGs
4
constrains for gas units, heat recovery system and
distribution network. The obtained results from the case
study, show impressive revenue from heat recovery and
significant improvement in active power loss. Also results
show energy not supply decrease by using DG in the
network without any automatic equipment to manage
outages. Final results show that using CHP units allow the
industrial company to use of maximum possible network
capacity and response to load growth by using a full or an
approximated dynamic view to solve the problem is
possible. The method developed in this paper has a good
friendship by restructured environment in order to supply
energy for industrial towns.
REFERENCES
2
0
1
2
4
5
Fig. 4: Comparing between annual ENS before and after units
installation
Figure 3 and 4 show annual loss and annual energy
not supply respectively. Considering these figures, both
ENS and loss decrease after installing the units. Although
these figures show affect of developing units in the
network decrease in the last years. This is because of
occurring a kind of saturation in the network by
increasing the total capacity installed. Therefore, it is
necessary to consider total distributed capacity installed
in each network, shall not be more than a specific value.
Final results show that improvement in ENS is less than
loss because studied feeder not equipped by automatic
device for automation in abnormal condition. If automatic
equipments exist in feeder and we can make the islands,
results can be better.
CONCLUSION
Using cogeneration, gas-fired units in the distributed
network cause to improve energy efficiency. In this paper
optimal sizing, placement and timing for CHP installation
is obtained through a cost-benefit analysis subject to
Favuzza, S., G. Graditi, M.G. Ippolito and
E.R. Sanseverino, 2007. Optimal electrical
distribution systems reinforcement planning using
gas micro turbines by dynamic ant colony search
algorithm. IEEE T. Power Syst., 22(2): 580-587.
Giaccone, L. and A. Canova, 2009. Economical
comparison of CHP systems for industrial user with
large steam demand. Appl. Energ., 86(6): 904-914.
Harrison, G.P., A. Piccolo, P. Siano and A.R. Wallace,
2007. Exploring the tradeoffs between incentives for
distributed generation developers and DNOS. IEEE
T. Power Syst., 22(2): 821-828.
Hashemi, R., 2009. A developed offline model for
optimal operation of combined heating and cooling
and power systems. IEEE T. Energ. Conver., 24(1):
222-229.
Keane, A., E. Denny and M.O. Malley, 2007. Quantifying
the impact of connection policy on distributed
generation. IEEE T. Energ. Conver., 22(1): 189-196.
Sanaye, S. and M. Reaessi, 2009. Estimating the power
and number of micro turbines in small-scale
combined heat and power systems. Appl. Energ., 86:
895-903.
Sun, Z.G., 2008. Energy efficiency and economic
feasibility analysis of cogeneration system driven by
gas engine. Energ. Build., 40: 126-130.
1066