Академический Документы
Профессиональный Документы
Культура Документы
To cite this article: Geoff F. Hewitt & Simon J. Pugh (2007) Approximate Design and Costing Methods for Heat Exchangers,
Heat Transfer Engineering, 28:2, 76-86, DOI: 10.1080/01457630601023229
To link to this article: http://dx.doi.org/10.1080/01457630601023229
GEOFF F. HEWITT
Department of Chemical Engineering & Chemical Technology, Imperial College of Science, Technology & Medicine,
London, UK
SIMON J. PUGH
ESDU International plc, London, UK
Methodologies for the rapid sizing and costing of heat exchangers have been developed under the aegis of ESDU International
plc, London. This paper is a summary of a group of design guides (referred to as data items) that cover a wide range of heat
exchanger configurations. These data items are aimed at providing rapid selection, sizing, and costing at the process design
stage. For two-stream exchangers, the C value method has been adopted in which the costs are expressed per unit (Q/Tm ),
is the heat load and Tm the mean temperature difference. The development and applications of this method are
where Q
reviewed, with an emphasis on comparisons between various types of exchanger. The nature of variations from the standard
cases considered are also discussed. Though the C value method can be applied to two-stream plate-fin exchangers, such
exchangers often operate with multiple streams. Approximate calculations for the design of such multistream exchangers
can be made using the concept of the volumetric heat transfer coefficient. This methodology can be combined with standard
curves of the cost per unit volume as a function of volume to obtain an approximate costing of such exchangers.
general background that ESDU International plc, in collaboration with heat exchanger manufacturers and under the guidance
of independent committees, embarked upon a series of their data
items on the topic of selection and costing of heat exchangers in
1992.
The first data item in the series [1] presented data allowing
the selection between feasible types and also gave approximate
cost data for the majority of the types considered. Starting in
1994, further data items were issued developing the methodology further for specific heat exchanger types. These further data
items include ones dealing with shell-and-tube exchangers [2],
air-cooled heat exchangers [3], plate-and-frame heat exchangers [4], and plate-fin heat exchangers [5]. Work on the series
is ongoing [6, 7], and ESDU is incorporating the methods into
computer codes for rapid access by process designers.
The objective of the present article is to briefly survey some of
the work that has been done. The data items are quite extensive
and cover more than 200 pages of detailed information. Clearly,
it is impossible to present any more than a brief summary in the
present article, but it is hoped that this will give a feel for the
kind of work being done.
In what follows, a description is given of the initial selection
procedure, and the bases of the methods used for costing to allow
BACKGROUND
The conservation of thermal energy by the use of heat exchangers is of vital importance in any scenario for sustainable
development. The most important step in the design process for
heat exchangers within a process plant occurs at the initial process design stage. At this stage, the fundamental decisions are
made about the incorporation of heat recovery networks within
the process leading to the placement and specification of heat
exchangers. There are many types of heat exchanger available
and, depending on the process, a range of heat exchanger types is
feasible to meet the physical conditions (pressure, temperature,
corrosion resistance, size, etc.) imposed by the process. The designer then has to choose between feasible types; ideally, this
would be done on the basis of cost, but there is often a great deal
of conservatism in industry in choosing heat exchanger types
outside the normal range of experience. However, if approximate cost data are available for all the feasible types, then this
makes the selection process much more focused. It was with this
76
77
Q
U Tm
(1)
Table 1 Sample from ESDU [1] tables summarizing characteristics of heat exchanger types (construction material: carbon steel)
Heat exchanger
type
Maximum pressure
(bar, absolute)
Temperature
range ( C)
Fluid
limitation
Up to 980
Low-pressure gases
Scraped-surface
10
Up to 300
Shell-and-tube
300 (shell)
Spiral
18
Welded-plate
60 (higher in shells)
In excess of 650
10 to 1000 m2 (per
shellmultiple shells
can be used).
Up to 200 m2
>1000 m2
Special features
Inter-stream leakage must
be tolerated
Suitable for viscous and
crystallization systems.
Very adaptable and can be
used for nearly all
applications.
High heat transfer
efficiency. Cylindrical
geometry useful as
integral part of
distillation tower.
Differential pressure
should be less than
30 bar. Differential
expansion should be
borne in mind.
78
However, for various forms of heat exchangers, the definition of area is often complex and, of course, the definition of U
is linked to this. In the Cvalue method (first suggested by He
witt et al. [8]), C is defined as the cost per unit ( Q/T
m ). This
avoids difficulties in defining area and overall coefficient and
allows a direct comparison between heat exchangers in terms
and the available temperature driving force
of the duty ( Q)
(Tm ), which are related to the process specification. In both approaches, care has to be exercised in specifying Tm because this
may depart significantly from its value for pure counter-current
flow.
Figure 1 Relationship between E and NTU with R as a parameter. Shell-andtube heat exchangers with E-type shells and an even number of tube-side passes
[11].
/ Tm
Determination of Q
is readily determined from the enthalpy
The heat load Q
change of either of the streams. For pure counter-flow exchangers with single phase streams of constant specific heat capacity,
the mean temperature difference Tm is equal to the logarithmic
mean temperature difference TL M given by:
Tm = TL M =
(2)
where Th,in and Th,out are the inlet and outlet temperatures of the
hot stream and Tc,in and Tc,out are the inlet and outlet temperatures of the cold stream.
Where there is significant departure from pure countercurrent flow operation, then the mean temperature difference
is given by:
Tm = FTL M
(3)
(4)
UA
p )smaller
( Mc
(5)
p )smaller
( Mc
,
p )larger
( Mc
(6)
Q
p )smaller NTU
= UA = ( Mc
Tm
(7)
79
Table 2 Film coefficients, fouling factors and overall film coefficient for tubular exchangers [8]
Cold side
Hot side
Film coefficient
W/m2 K (clean)
Fouling factor
Km2 /W
Overall film
coefficient W/m2 K
Film coefficient
W/m2 K (clean)
Fouling factor
Km2 /W
Overall film
coefficient W/m2 K
112
682
5000
1667
210
5676
1667
0.0002
0.0002
0.0002
0.0004
0.0008
0.0003
0.0004
110
600
2500
1000
180
2100
1000
112
682
6000
1667
170
8182
1410
435
0.0002
0.0002
0.0005
0.0004
0.0008
0.0001
0.0002
0.0002
110
600
1500
1000
150
4500
1100
400
Figure 2 Cost per unit area as a function of area for BEM-type shell-and-tube
heat exchangers [2].
Figure 3 Cost per unit area as a function of area for plate-and-frame heat
exchangers [4].
The tables of C values (exemplified by Table 3) can be interpolated logarithmically. Thus, the value of C is given by:
ln[( Q/T
m )1 /( Q/Tm )2 ]
where C1 and C2 are the C values of the particular hot-side/cold
( Q/T
m ), as is exemplified by the following values taken for
the case of treated cooling water on the cold-side and a low
viscosity organic fluid on the hot-side.
For small duties, the double-pipe heat exchanger is more economical than the shell-and-tube heat exchanger, reflecting the
ability to mass-produce its components. The reverse is true for
large duties. Over the range shown, the plate-and-frame heat exchanger is notably more economical than the other types, though,
of course, there are potential problems with sealing. A fully
welded plate exchanger, on the other hand, is more expensive
than the tubular types for small duties but less expensive for large
duties. The printed circuit heat exchanger is highly compact and
80
Q/T
(W/K) Fluid
1000
5000
Low-pressure
gas (<1 bar)
Medium-pressure
gas (20 bar)
High-pressure
gas (150 bar)
Treated
cooling water
Low-viscosity
hydrocarbon
High-viscosity
hydrocarbon
Boiling
water
Boiling
organic liquid
Low-pressure
gas (<1 bar)
Medium-pressure
gas (20 bar)
High-pressure
gas (150 bar)
Treated cooling
water
Low-viscosity
hydrocarbon
High-viscosity
hydrocarbon
Boiling
water
Boiling organic
liquid
Parameter
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
U (W/m2 K)
C (/(W/K))
Lowpressure
gas
(<1 bar)
Mediumpressure
gas
(20 bar)
55
4.8
95
3.4
120
2.9
105
2.8
100
2.7
70
4.7
105
3.9
100
4
55
2.16
95
1.5
120
2.05
105
1.4
100
1.45
70
2.4
105
1.4
100
1.45
95
3.8
300
2.5
350
2.9
484
2.5
375
2.5
140
2.5
470
2.5
375
2.5
95
1.26
300
0.86
350
1.1
484
0.75
375
0.8
140
1.66
470
0.9
375
0.8
HighLow
pressure
viscosity
gas
Process hydrocarbon
(150 bar) water
liquid
125
2.9
350
2.9
400
2.9
500
2.9
425
2.9
175
2.9
550
2.9
430
2.9
125
1.1
350
1
400
1.1
500
1
425
1.05
175
1.2
550
1
430
1.05
105
3.8
430
2.5
500
2.9
940
2.5
600
2.5
160
2.5
875
2.5
600
2.5
105
1.23
430
0.76
500
1
940
0.5
600
0.8
160
0.95
875
0.5
600
0.8
100
3.8
375
5
400
2.9
715
2.5
500
2.5
155
2.5
670
2.5
500
2.5
100
1.22
375
0.8
400
1.1
715
0.72
500
0.9
155
1
670
0.6
500
0.9
High
viscosity
Condensing
hydrocarbon Condensing Condensing hydrocarbon
liquid
steam
hydrocarbon with inert gas
65
4.7
120
2.6
150
2.9
145
2.5
130
2.5
85
3.9
140
2.5
130
2.5
65
1.84
120
1.16
150
1.3
145
1
130
1.1
85
2.7
140
1.1
130
1.28
110
2.8
530
2.5
600
2.9
1610
2.5
820
2.5
175
2.5
1435
2.5
820
2.5
110
1.23
530
0.6
600
1
1610
0.4
820
0.5
175
0.82
1435
0.4
820
0.7
100
3.8
390
2.5
420
2.9
765
3.9
525
2.5
155
2.5
725
2.5
525
2.5
100
1.22
390
0.8
420
1.1
765
0.72
525
0.9
155
1.6
725
0.73
525
1
85
3.9
240
2.5
350
2.9
345
2.5
290
2.5
125
2.5
340
2.5
285
2.5
85
3.5
240
0.94
350
1.2
345
0.9
290
1.2
125
1.95
340
0.9
285
1.1
Q/T
m
= 5000 W/K
Q/T
m
= 100,000 W/K
Shell-and-tube
Double-pipe
Plate-and-frame
Printed circuit
Welded plate
0.91
0.72
0.140
2.4
1.0
0.134
0.140
0.045
0.400
0.108
Figure 4 Comparison of costs of stainless steel plate-and-frame heat exchangers and carbon steel shell-and-tube exchangers [2].
81
COST VARIATIONS
The numbers given in C value tables exemplified by Table 3
are for standard designs. For example, in the case of the table
for shell-and-tube exchangers, the standard is a fixed tube sheet,
two-pass carbon steel exchanger with a E-type shell (designated BEM in the TEMA [12] standards). Real designs depart,
of course, from these standard configurations, and allowance
must be made for these variations in assessing costs. A simplistic way of doing this is to use fixed multipliers, and this has
been adopted in the ESDU studies where appropriate. Table 5
shows the effect of various changes from the standard design for
plate-and-frame exchangers.
The cost-factor approach leads to serious errors where manufacturing costs are a significant proportion of the total cost. In
this case, the effect of the variation increases with an increasing
surface area. Examples of such variations are shown in Figures
6 and 7, both for the case of shell-and-tube heat exchangers.
In Figure 6, the effect of tube material is shown. For low
surface areas, manufacturing costs represent a bigger proportion
of the total costs, and the cost ratio is lower than the relative cost
Change
Factor on
C value
1.1
1.4
1.7
0.9
1.5
1.6
Figure 7 Rates of costs of floating head (AES) and fixed tube sheet (BEM)
shell-and-tube exchangers as a function of surface area [2].
82
(8)
(9)
au f 1 + f 1 a f 1
2
au f 2 + f 2 a f 2
2
(T1 TW 1 )1
(10)
(TW 2 T2 )2
(11)
= (TW 1 TW 2 )yz/(t p + t f )
dQ
W
(12)
s(t p + t f )
W
(13)
(15)
1
1
1
+
=
B
1
2
(16)
and
Although a wide range of fin geometries are possible with platefin heat exchangers, the most common type is the serrated fin, and
a suitable reference geometry would be one with a fin frequency
( f n ) of 708.7 fins/m (18 fins per inch), a fin thickness of 2.032
104 m (0.008 in), a plate gap/fin height (b) of 0.00635 m
(0.25 inches), and a parting sheet thickness (t p ) of 1.5 mm. The
procedure used in ESDU [5] was to estimate typical values of
i for various streams for this reference design (analogous to
the values given for area-based film coefficients in Table 2 for
shell-and-tube heat exchangers). In order to calculate guideline
coefficients, pressure drops of 0.1 bars were assumed for liquid and two-phase streams and 1% of the gas pressure for gas
streams. The fin efficiency was calculated from the equation:
=
tanh m
m
(17)
2t f W
(18)
Q
Va =
BTm
(19)
heat transfer engineering
83
1100
1400
80
400
1000
1200
60
300
of Q/T
m , and tables of C values obtained in this way are
presented in ESDU [5].
Method For Multistream Exchangers
The volumetric heat transfer coefficient methodology can be
extended to cover multistream plate-fin exchangers using the
following steps:
1. The stream data are represented in terms of hot and cold composite curves using the pinch analysis method (see Linhoff
and Smith [17] for a detailed description of the methodology).
A typical pair of composite curves is shown in Figure 10 for
the six stream example, which will be presented in more detail below. As will be seen, there is a pinch at which the
minimum temperature difference Tmin is 6 K.
2. The composite curve is divided into zones corresponding to
regions where the hot and cold composite curves are linear, as
exemplified by Figure 10. Each of these zones is then treated
separately.
3. A mean volumetric coefficient for a zone containing n
streams can be estimated from the expression:
n
z
i
Q
Q
=
Bz
i
i =1
(20)
84
Hot
or cold
p (kW/K)
Mc
Inlet
temperature (K)
Outlet
temperature (K)
H1
H2
H3
C1
C2
C3
Hot
Hot
Hot
Cold
Cold
Cold
10
5
8
15
5
20
300
250
200
90
120
170
150
100
150
130
210
250
z /Tm,z
Q
z
(21)
Figure 11 Network of two-stream exchangers for heat recovery between the hot and cold streams for the six-stream example shown in Table 7. Note: Numbers
in the diagram that are not assigned units are temperatures in C.
Here, there are six hydrocarbon gas streams (three hot and
three cold). The composite curves for this example are as shown
in Figure 10. Applying the procedure described above, a total
exchanger volume (V ) of 4.97 m3 is obtained, and using the
manufacturers curves for cost per unit volume as function of
volume, the cost of the multistream exchanger for this duty is
calculated as 95 k. It is interesting to compare the cost of a multistream exchanger with the costs of solving the same problem
using a network of two-stream exchangers. The network can be
developed using the methods of pinch analysis (see Linhoff and
Smith [17]) and is shown using the conventional grid diagram
in Figure 11.
As can be seen, ten two-stream heat exchangers are required.
The estimated cost of the exchangers is around 195 k, more
than double the cost of meeting the required duties in a single
multi-stream exchanger.
CONCLUSIONS
The following main conclusions can be drawn from the studies forming the basis of this article: the provision of information
for selection and budget costing is important for the process designer at the initial stage of design of the process. This is the
crucial stage in achieving the most economic solutions.
The C value method has proved an invaluable tool in the
selection and preliminary costing of heat exchangers, and is
already built into several proprietary computer codes.
The approximate design of multistream exchangers is aided
by using volumetric heat transfer coefficients, and a new methodology for estimating these is presented.
Considerable cost savings can be achieved by using multistream exchangers rather than a multiplicity of two-stream exchangers.
ACKNOWLEDGMENTS
The authors would like to gratefully acknowledge the assistance of Heat Transfer Steering Group of ESDU in advising
on the work described in this present article. They are also extremely grateful for the assistance of various working parties
(drawn mainly from industry) who advised on the work on the
specific heat exchanger types.
NOMENCLATURE
A
a
B
b
C
cp
E
F
fn
H
m
M
n
NTU
Q
R
s
T
t f , t p
T
U
V
Va
V
y
z
85
Greek Symbols
stream heat transfer coefficient, W/m2 K
film volumetric heat transfer coefficient, W/m3 K
thermal conductivity, W/mK
fin efficiency
Subscripts
c, h
f, u f
in, out
LM
m
max
W
z
1, 2
REFERENCES
[1] ESDU, Selection and Costing of Heat Exchangers, ESDU data
item No. 92013, ESDU International plc, London, UK, 1992.
[2] ESDU, Selection and Costing of Heat Exchangers, Shell-and-Tube
Type, ESDU data item No. 94042, ESDU International plc, London, UK, 1994.
[3] ESDU, Selection and Costing of Heat Exchangers, Air-Cooled
Type, ESDU data item No. 94043, ESDU International plc,
London, UK, 1994.
[4] ESDU, Selection and Costing of Heat Exchangers, Plate-andFrame Type, ESDU data item No. 95007, ESDU International
plc, London, UK, 1995.
86
[16] Taylor, M. A., Plate-Fin Heat Exchangers: Guide to Their Specification and Use, 1st ed., HTFS, Harwell Laboratory, England,
1987.
[17] Linnhoff, B., and Smith, R., Pinch Analysis for Network Design, Heat Exchanger Design Handbook, ed. G. F. Hewitt, Begell
House, New York, 1994.
Geoff Hewitt is Emeritus Professor of Chemical Engineering at Imperial College, London. He was formerly the head of the Thermal Hydraulics Division
and founder of the Heat Transfer and Fluid Flow
Service (HTFS) at the Harwell Laboratory of the
UKAEA. He moved to Imperial College (part-time
in 1985 and full-time in 1990) and has continued
to work there on multiphase flow and heat transfer.
He has authored and edited many books (including
Process Heat Transfer in 1994 and Encyclopedia of Heat and Mass Transfer in 1997) and published more than 400 papers and reports, mainly on gasliquid flow and evaporative heat transfer. He is the editor of Multiphase Science and Technology, executive editor of the Heat Exchanger Design Handbook, and one of the founding editors of Heat Transfer Engineering. He is
the recipient of the AIChE Donald Q. Kern Award, ASME Max Jakob Award,
the Nusselt Reynolds Prize, the Luikov Medal, and the IChemE Council and
Armstrong Medals. He has received honorary doctorates at Louvain, UMIST,
and Heriot Watt. He is a fellow of the Royal Academy of Engineering, fellow
of the Royal Society, and Foreign Associate of the US National Academy of
Engineering.
Simon Pugh is the head of thermofluids at ESDU
International plc of London, UK. His current role includes the management of all ESDUs heat transfer
work, which is undertaken under the guidance of international independent committees of experts from
industry and the universities. He is currently leading
the group of engineers working on the development
of a range of design guides to oil industry fouling
problems and computer programs for better selection,
design, and operation of heat exchangers. He holds a mechanical engineering
degree from Brunel University in UK.