1
AbstractThe present study explores the use of a nontraditional
optimization technique; Genetic Algorithm (GA), for
optimization of steam surface condenser. The code for GA is
developed and successfully applied for the optimization of same
by varying the design variables such as shell internal diameter,
tube outer diameter, tube thickness and tube material. The
objective function for capital cost is derived and same is used for
optimization. The optimized results obtained using GA are found
to be superior to the earlier obtained results.
Index termsSurface condenser, Optimization, Genetic
algorithm
I. INTRODUCTION
team surface condenser is one type of shellandtube heat
exchangers (STHEs) used in power plant to condense the
steam coming from the turbine exhaust. Many researchers
have investigated how to optimize the single phase STHEs
using different optimization techniques such as simulated
annealing, global sensitivity analysis, harmony search
algorithm, particle swarm optimization method, differential
evolution strategies, and genetic algorithm considering
different objective functions. The optimization of two phase
steam surface condenser is not studied extensively and thus
there is enough scope for research.
Babu B. V. et al. [1] used differential evolution (DE)
optimization method and its various strategies for the optimal
design of STHEs. The authors had analyzed the problem
considering the main objective as the estimation of minimum
heat transfer area required for a given heat duty. Selbas et al.
[2] used GA for optimum design of STHEs, in which pressure
drop was applied as a constraint for achieving optimum
design parameters. The authors had considered minimization
of total cost as an objective function for STHEs.
Fesanghary et al. [3] used global sensitivity analysis to
identify the most influential geometrical parameters that affect
the total cost of STHE in order to reduce the size of
optimization problem and carried out the optimization of
influential parameter of STHE from economic point of view
by applying the harmony search algorithm.
Patel V. K. et al. [4] used particle swarm optimization
(PSO), for design optimization of STHEs from an economic
view point. Minimization of total annual cost was considered
as an objective function of heat exchanger. The authors had
selected three design variables such as shell internal diameter,
outer tube diameter and baffle spacing for optimization.
I. DESIGN OPTIMIZATION METHOD
Steam surface condenser is designed using a number of
design variables i.e. shell inner diameter, tube outer diameter,
tube thickness, tube pitch, tube layout angle, tube material
and so on. The present study explores the use of Genetic
Algorithm (GA) to optimize the cost of steam surface
condenser using four variables for optimization of design i.e.
shell inner diameter, tube outer diameter, tube thickness, and
tube material. In GA, these four variables are coded to
evaluate the design of condenser and hence find out the
optimum solution.
Genetic Algorithm:
The basic principle of genetic algorithm was first proposed
in the 1970s by John Holland.
The genetic algorithm is based on the natural selection, which
was found in biological evolution process. In the optimization
design application, before a genetic algorithm can be put to
work, a method is needed to encode potential solutions to the
optimization problem in a form that a computer can process.
One common approach is to encode solutions as binary
strings: sequences of 1s and 0s, where the digit at each
position represents the value of some aspect of the solution. A
metric called a fitness function that allows each potential
solution (individual) to be quantitatively evaluated. After a
random initial population in the ranges of design variables is
generated, the algorithm creates a sequence of new
generations iteratively until the stopping criterion is met. In
this process, the selection of parents is based on their fitness;
children (next generation or population) are produced by
making random changes to a single parent (mutation) or by
combining the vector entries of a pair of parents (crossover),
and then replace the current population with the children to
form the next generation. The algorithm selects individuals
with better fitness values as parents, and eliminates the
inferior. This guarantees the algorithm converges to a best
individual, which probably represents the best solution of the
given problem [5].
Steam Surface Condenser Design based on
Cost Optimization using Genetic Algorithm
A. Nirmalkumar P.Bhatt, B. A. M. Lakdawala, C.
V. J. Lakhera
3
A, B & C. Mechanical Engg. Department, Institute of Technology, Nirma University, Ahmedabad
S
INTERNATIONAL CONFERENCE ON CURRENT TRENDS IN TECHNOLOGY, NUiCONE 2011
2
II. MATHEMATICAL MODEL
1) Heat transfer:
Generally in steam surface condenser, steam expands from
turbine exhaust on shell side and it condenses by cooling
water in tube side. Therefore, Thermal analysis is done on
both tube side and shell side for heat transfer and its area.
According to flow regime with low Reynolds number, the
tube side heat transfer coefficient is computed from the
following correlation,
( )
( )
(
+
+ =
3 . 0
33 . 1
/ Pr Re 1 . 0 1
/ Pr Re 0667 . 0
657 . 3
L d
L d
d
k
h
i t t
i t t
i
t
t
. (1)
(If Re
t
<2300 [6])
Where, d
i
is the tube inner diameter and L is the length of
tube.
According to flow regime with transition flow, the tube
side heat transfer coefficient is computed from Gnielinski
correlation,
( )( )
( ) ( ) 1 Pr 2 / 7 . 12 1
Pr 1000 Re 2 / h
3 / 2 2 / 1
t
+
=
t
t t
t
f
f
k
d
. (2)
(If 2300<Re
t
<10,000 [7]
where, ( )
2
28 . 3 Re ln 58 . 1
=
t
f
According to flow regime with higher Reynolds number,
the tube side heat transfer coefficient is computed from
Petukhov and Kirillov correlation,
( )
( ) ( ) 1 Pr 2 / 7 . 12 07 . 1
Pr Re 2 / h
3 / 2 2 / 1
t
+
=
t
t t
t
f
f
k
d
. (3)
(For Re
t
>10,000 [7])
where, ( )
2
28 . 3 Re ln 58 . 1
=
t
f
Re
t
is the tube side Reynolds number and given by,
t
i t t
t
d v
= Re . (4)
Pr
t
is the tube side Prandlt number and given by,
t
l t
t
k
cp
= Pr . (5)
Flow velocity for tube side is found by,
( )


\

=
t t t
t
t
N
n
d
m
v
2
4 /
. (6)
where, N
t
is the number of tubes and n is the number of
tube passes which can be found approximately from the
following correlation [7],
( )
2 2
2
875 . 0
o
s
t
d PR
D
CL
CTP
N 
\

= . (7)
where, D
s
is shell diameter, which would contain right
number of tubes, N
t
of tube diameter d
o
. CTP is the tube
count calculation constant and CL is the tube layout constant.
For twotube pass and 30 tube layout angle, CTP and CL
values are suggested,
CTP=0.9
CL=0.87
Using Nusselt correlation [8], heat transfer coefficient for a
single horizontal tube with assume laminar film condensation
is computed from the following correlation,
( )
(
(
=
l w l
o fg g l l
o
l
s
k T
d gi
d
k
h
3
728 . 0 . (8)
The temperature difference T
w
is given by,
" q R T T T T
t w sat w
= =
where, T is the local temperature difference between the
streams; and R
t
is the sum of all other resistance (based on the
tube outer diameter); and q" is the local heat flux, which are
given by,
fo
m
o
w
w
fi
i
o
t i
o
t
R
D
d
k
t
R
d
d
h d
d
R + + + =
1
and
T U q
o
= "
where, D
m
is approximated as:
( )
( )
i o
i o
i o
m
d d
d d
d d
D + =
=
2
1
/ ln
and t
w
is the wall thickness; and U
o
is the overall heat transfer
coefficient based on the tube outer diameter.
Considering splashing effect on horizontal tube bundle, the
average heat transfer coefficient of tube bundle is computed
from the following Kern correlation [9],
6 / 1
,
= N h h
s N s
. (9)
The overall heat transfer coefficient based on the tube outer
diameter (U
o
) depends on both the shell side and tube side
heat transfer coefficients and fouling resistances is given [7],
INSTITUTE OF TECHNOLOGY, NIRMA UNIVERSITY, AHMEDABAD 382 481, 810 DECEMBER, 2011
3
N s
t
o
h
R
U
,
1
1
+
= . (10)
Considering for a condensation, the logarithmic mean
temperature difference LMTD is computed from the
following formula [10],
( ) ( ) [ ]
out c sat in c sat
in c out c
T T T T
T T
LMTD
, ,
, ,
/ ln
= . (11)
Considering overall heat transfer coefficient, the heat
exchanger surface area A is computed by,
LMTD F U
Q
A
o
= . (12)
where, the LMTD correction factor F is found as one for
condensation and the heat capacity ratio R is zero [7].
Correlation of the heat transfer rate is given either by
condensation or by sensible heat transfer which is computed
from following formula,
( )
in c out c pc c fg s
T T C m h m Q
, ,
= = . (13)
where, m
s
is the mass flow rate of steam; and h
fg
is the
enthalpy of vaporization which is the difference between
enthalpy of saturated steam h
v
and enthalpy of water m
c
is the
mass flow rate of cooling water. T
c,out
and T
c,in
are the
temperature of cooling water at inlet and outlet respectively.
Based on the heat exchanger surface area A, the necessary
tube length L is,
t o
N d
A
L
= . (14)
2) Pressure drop:
In the heat exchanger there is a close physical and
economical affinity between pressure drop and heat transfer.
This pressure drop is the static fluid pressure which may be
expanded to drive the fluid through heat exchanger.
Increasing the flow velocity will cause a rise of heat transfer
coefficient which results in compact heat exchanger design
and lower investment cost. However increase of flow velocity
will cause more pressure drop in exchanger which results in
additional running cost. For this reason when designing a heat
exchanger pressure drop must be considered with heat
transfer.
Tube side pressure drop P
t
is computed from the
following correlation [4],
n p f
d
L v
P
i
c c
c


\

+ =
2
2
. (15)
where,
c
is the density of cooling water, n is the tube pass, p
is a constant and its different values are considered by
different authors, kern assumed p=4, while Sinnot et al.
assumed p=2.5. [4]
For steam surface condenser, only tube side is a running
cost of condenser, while shell side steam expands in a
condenser from turbine exhaust directly. Thats why total
pumping power is computed from tube side pressure drop
only. This power is computed by following formula,


\

=
c
c
c
P
P
m
E
1
. (16)
where, is the efficiency of pump and m
c
is the mass flow
rate of cooling water.
i. Objective function:
Total cost C
tot
is taken as the objective function, which
includes capital cost C
c
and operating cost C
o
.
o i tot
C C C + = . (17)
The capital cost of the exchanger C
i
is computed as a function
of heat exchanger surface area A using following formula [3],
b
c
aA C = . (18)
where, a and b are the cost coefficient constants. For
Admiralty the value of a and b are shown in Table I.
Thus,
( )
( ) 1 1
1
+
+
=
TL
R
TL
R R
c i
i
i i
C C . (19)
The operating cost of exchanger is computed using following
formula,
P op
E ec TP C
= . (20)
Thus,
( )
( ) 1 1
1
+
+
=
TL
R
TL
R R
op o
i
i i
C C . (21)
TP is the period of the time of operation per year; i
R
is the
interest rate; ec is the unit cost of the energy; and TL is the
technical life of condenser.
INTERNATIONAL CONFERENCE ON CURRENT TRENDS IN TECHNOLOGY, NUiCONE 2011
4
The cost of exchanger using materials other than Admiralty
is computed using proper materialcost ratio of the materials
(cost of material/Cost of Admiralty). The current cost of the
materials can be finding from London Metal Exchange [11].
TABLE I
COST COEFFICIENT OF CONDENSER
Range a b
A < 9 m
2
3135 0.463
9 m
2
< A < 90 m
2
1957 0.679
A < 9 m
2
1042 0.810
The steam surface condenser design is optimized using GA
in the C program code. The results are compared with
Siemens design considering different design variables range
(design variables range used by Siemens) which are given as
for shell inner diameter (mm): 4900.0900.0; tube outer
diameter (mm): 24.019.0; Tube thickness (mm): 1.245
0.559; and Tube materials: Admiralty brass, Titanium, Cu
Nickel, SS304.
III. RESULTS AND DISCUSSION
The effectiveness of the present approach using GA is
assessed by analyzing three case studies.
Case 1: 18 MW Coal Based Captive Cogeneration Power
Plant, Doc No: 1CYJ185288, by Siemens Ltd, Baroda
Case 2: 9.9 MW Bua Sommai, Suwannaphum, Doc No:
1CYJ185182 by Siemens Ltd, Baroda
Case 3: 24 MW KPR Sugar Plant, Doc No: 1CYJ185162
by Siemens Ltd, Baroda
The specifications of above three cases are analyzed by
considering twotube pass with 30 tube layout using 1.85 to
2.2 pitch ratio. Cleanliness factor for condenser is considered
as 0.85. For pumping power, the pump efficiency is taken as
0.85. For economic calculation, the operating time of
condenser is taken 8000 hour/yr, electric cost 5 Rs/yr,
Technical life 10 yr, and interest rate of condenser and
pumping cost is considered as 20%. For GA operation, the
population size per iteration is taken 40, the one site
crossover probability is taken 0.9 and mutation probability is
taken 0.04.
i. Case 1: 18 MW Coal Based Captive Cogeneration Power
Plant, Doc No: 1CYJ185288, by Siemens Ltd, Baroda
The process specification of Case 1 for condenser is given
that the steam with 20.641 kg having dryness fraction 0.88
from the turbine is condensed in condenser at pressure 0.111
bar. The cooling water is available at temperature of 32C
TABLE II.
GA RESULTS COMPARED WITH DESIGN MADE BY SIEMENS LTD.
USING SIEMENS DESIGN VARIABLES RANGE FOR CASE 1
Parameter obtained
when minimization of
capital cost as an
objective function
By Siemens Ltd,
Doc No:
1CYJ185288
This work
(GA)
D
s
(mm) 2500 2280.25
d
o
(mm) 19.05 20.80
Tube thickness (mm) 0.7 0.72
Tube material SS304 SS304
Number of Tubes 4246 3178
v
t
(m/s) 2.25 2.24
L (mm) 5.05 6.03
P
t
(kPa)  19.90
E
P
(kW)  25.54
U (w/m
2
k)  2854.66
A (m
2
) 1263 1252.82
C
i
(Rs/yr) 1 91 507 1 90 256
this enters in condenser inlet and comes out at temperature of
41.56C.
The results of inhouse GA code are compared with
Siemens data. The design is optimized by considering capital
cost of the condenser as an objective function and results are
found that the cost of design found by GA is 0.80% less than
the Siemens work. And at the same time the velocity will be
less by 0.44% using GA than Siemens Design, so the
operating cost found by this work is also less than the
Siemens design. The almost same results are found for heat
transfer area and so the capital cost within 98 iteration of GA.
ii. Case 2: 9.9 MW Bua Sommai, Suwannaphum, Doc No:
1CYJ185182 by Siemens Ltd, Baroda
The process specification of Case 2 for condenser is given
that the steam with 9.12 kg/s having dryness fraction 0.88
from the turbine is condensed in condenser at pressure 0.103
bar. The cooling water is available at temperature of 32C
which enters in condenser inlet and comes out at temperature
of 41.01C.
For Case 2, the capital cost of design found by GA is 2.59%
less than the Siemens work. And at the same time the velocity
is also less by 21.53% using GA than Siemens Design, so the
operating cost found by this work will be less than the
Siemens design. The almost same results are found for heat
transfer area and so the capital cost within 24 iteration of GA.
iii. Case 3: 24 MW KPR Sugar Plant, Doc No:
1CYJ185162 by Siemens Ltd, Baroda
INSTITUTE OF TECHNOLOGY, NIRMA UNIVERSITY, AHMEDABAD 382 481, 810 DECEMBER, 2011
5
TABLE III.
GA RESULTS COMPARED WITH DESIGN MADE BY SIEMENS LTD.
USING SIEMENS DESIGN VARIABLES RANGE FOR CASE 2
Parameter obtained
when minimization of
capital cost as an
objective function
By Siemens Ltd,
Doc No:
1CYJ185182
This work
(GA)
D
s
(mm) 2100 1900.98
d
o
(mm) 19 20.47
Tube thickness (mm) 0.7 0.60
Tube material SS304 SS304
Number of Tubes 2458 2282
v
t
(m/s) 1.95 1.53
L (mm) 5.5 5.28
P
t
(kPa)  9.18
E
P
(kW)  5.50
U (w/m
2
k)  2709.12
A (m
2
) 801 775.43
C
i
(Rs/yr) 1 32 431 1 28 996
The process specification of Case 3 for condenser is given
that the steam with 20.641 kg having dryness fraction 0.88
from the turbine is condensed in condenser at pressure 0.111
bar. The cooling water is available at temperature of 32C
which enters in condenser inlet and comes out at temperature
of 41.56C.
For Case 3, the capital cost of design found by GA is
5.27% less than the Siemens work. And at the same time the
velocity is also less by 6.31% using GA than Siemens Design,
so the operating cost found by this work will be less than the
Siemens design. The almost same results are found for heat
transfer area and so the capital cost within 118 iteration of
GA.
The cost found by GA is compared with Siemens result
which shown in Fig. 1. From analyzing above three case
studies, the results found better for this work than the Siemens
design. Siemens got the design using experimental results of
Heat Exchanger Institute (HEI) standard for overall heat
transfer coefficient. In present work overall heat transfer
coefficient is calculated using empirical correlation (Nusselt
correlation and Kern relation for multiple tubes) which gives
less overall heat transfer coefficient than the experimental
result. Even though using this correlation, the results of GA is
better than the Siemens design. Therefore, GA can be
successfully employed for optimization of steam surface
condenser considering different design variables.
TABLE IV.
GA RESULTS COMPARED WITH DESIGN MADE BY SIEMENS LTD.
USING SIEMENS DESIGN VARIABLES RANGE FOR CASE 3
Parameter obtained
when minimization of
capital cost as an
objective function
By Siemens Ltd,
Doc No:
1CYJ185162
This work
(GA)
D
s
(mm) 2300 2389.74
d
o
(mm) 23 20.47
Tube thickness (mm) 0.7 0.64
Tube material SS304 SS304
Number of Tubes 3136 3606
v
t
(m/s) 2.22 2.08
L (mm) 7.0 6.34
P
t
(kPa)  17.72
E
P
(kW)  27.06
U (w/m
2
k)  2709.12
A (m
2
) 1572 2842.08
C
i
(Rs/yr) 2 28 652 2 16 610
Fig.1 Capital cost comparison
IV. CONCLUSIONS
An algorithm for the optimum design of steam surface
condensers based on genetic algorithm has been presented.
The model uses the Nusselts correlation with Kern Method
for calculation of heat transfer coefficients in the shell side.
The objective function can include any kind of function which
is non linear and influence the cost of condenser can be
INTERNATIONAL CONFERENCE ON CURRENT TRENDS IN TECHNOLOGY, NUiCONE 2011
6
handled in mathematical programming techniques using
genetic algorithm. The Genetic algorithm provides better
solution to detect global optimum results while reducing the
number of evolution for the objective function. The solution
to case studies taken from the Siemens data sheet show how
previously reported designs can be improved through the use
of the new approach presented in the present work.
V REFERENCES
[1] B.V. Babu, S.A. Munawar, Differential evolution strategies for
optimal design of shellandtube heat exchangers, Chemical
Engineering Science 62 (2007) 3720 3739
[2] Resat Selbas, Onder Kzlkan, Marcus Reppich, A new design
approach for shellandtube heat exchangers using genetic algorithms
from economic point of view, Chemical Engineering and Processing,
45 (2006) 268275
[3] M. Fesanghary, E. Damangir, I. Soleimani, Design optimization of
shell and tube heat exchangers using global sensitivity analysis and
harmony search algorithm, Applied Thermal Engineering 29 (2009)
10261031
[4] V.K. Patel, R.V. Rao, Design optimization of shelland tube heat
exchanger using particle swarm optimization technique, Applied
Thermal Engineering 30 (2010) 14171425
[5] Jiangfeng Guo, Lin Cheng, Mingtian Xu, Optimization design of
shellandtube heat exchanger by entropy generation minimization
and genetic algorithm, Applied Thermal Engineering 29, 2009,
p.p.29542960
[6] Antonio C. Caputo, Pacifico M. Pelagagge, Paolo Salini, Heat
exchanger design based on economic optimization, Applied Thermal
Engineering, 28 (2008) 11511159
[7] Sadik Kakac, Hongtan Liu, Heat Exchangers Selection, Rating, and
Thermal Design. Second Edition, 1998
[8] Robert W. Serth, Process Heat Transfer. Second Edition, 2007
[9] G. F.Hewitt, G.L. Shires, T.R. Bott, Process Heat Transfer. Second
Edition, 1994
[10] Heat Exchanger Institute, Inc. Standards for Steam Surface
Condensers, Heat Exchanger Institute, 10th Edition, 2006
[11] London Metal Exchange, URL: http://www.lme.com/
Much more than documents.
Discover everything Scribd has to offer, including books and audiobooks from major publishers.
Cancel anytime.