Вы находитесь на странице: 1из 6

Design of Multipass Heat

S. Ahmad Exchangers: an Alternative


B. Linnhoff Approach
Multipass heat exchangers are often designed according to the rule of thumb
B. Smith F T >0.75. This criterion can give rise to poor designs. It is the slope (dF T /9P) R that
Department of Chemical Engineering,
is important, not just the absolute value of F T . Furthermore, in multishell ar-
UMIST, rangements, the number of shells required is usually evaluated by trial and error.
Manchester, England This paper introduces an approach based on a new parameter that is an alternative
to F T factors. The method is fully compatible with established design practice, takes
account of the F T slope, and evaluates the number of shells explicitly in multishell
cases.

1 Introduction
One of the most common types of shell-and-tube exchanger R = CPH/CPC = (TCo - Tc,)/{Tm - TH0) (3)
is the 1-2 design (1 shell pass-2 tube passes). Because the flow and
arrangement involves part countercurrent and part cocurrent
flow, the effective temperature difference for heat exchange is P- (7/fi ~ THo)/(THi - Ta) (4)
reduced compared with a purely countercurrent device. This is Some designers use the alternative definitions
accounted for in design by the introduction of the FT factor
into the basic heat exchanger design equation, as discussed by R' = CPC/CPH = {Tm - THo)/(TCo - Tci) (3a)
Bowman et al. (1940) and
Q=UAATLMFT (1) P'=(TCo-Ta)/(THi-Ta) (4a)
The FT correction factor is usually correlated in terms of two instead of R and P. Whichever of the two conventions is
dimensionless ratios, the thermal effectiveness of the ex- adopted is immaterial in determining the value of FT for given
changer (P) and the ratio of the two heat capacity flow rates exchanger profiles. This results because
(R). R' = l/R, P'=RP
At the early stages of design, engineers need to be able to
screen alternative designs before going forward to more de- and
tailed calculations. These alternatives include not only dif- FT{R, P)=FT(l/R, RP) = FT(R', P') = FT(\/R', R'P')
ferent types of exchanger, but also in the case of 1-2 designs,
multiple shell arrangements forced by unacceptably low values Three basic situations can be encountered when using 1-2
of FT. If multiple shells are required then the most common exchangers (Fig. 1):
practice is to adopt a trial-and-error approach in which the (a) We have a temperature approach. This situation is
number of shells in series is progressively increased until a straightforward to design for using a single 1-2 shell.
satisfactory value of FT is obtained for each shell. (b) We have a small temperature cross. This situation is
This paper describes an alternative approach to the design usually straightforward to design for and again can probably
of 1-2 exchangers, particularly useful in the early stages of be accommodated in a single shell.
design, which does not rely directly on FT factors. The ap- (c) We have a large temperature cross. Local reversal of
proach is fully compatible with established design procedures heat flow may be encountered, which is wasteful in heat
and offers two distinct advantages: transfer area. The design might even become infeasible.
(0 It generates better designs than those based on tradi- One basic question therefore remains: How much
tional rules of thumb for specifying minimum permissible FT temperature cross can we tolerate before encountering reversal
be avoiding uncertain areas of the FT chart more effectively. of heat flow or infeasibility? Clearly, infeasible exchanger
(H) For multishell arrangements, the number of shells does designs returnF T <0, but havingF T >0 is not enough to make
not need to be evaluated by trial and error but can by a design practical. A commonly used rule of thumb requires
evaluated explicitly. FT-> 0.75 for the design to be considered practical as discussed
by Kern (1950).
2 F T Correction Factors However, the use of the criterion FT>0.15 for 1-2 ex-
Graphical and analytical correlations exist for FT based on a changers is only a rule of thumb and can lead to poor designs
number of assumptions concerning the heat transfer in a 1-2 if not used with caution.
exchanger (see Bowman et al., 1940). Important among these The fact that designs using values of FT less than 0.75 are
are that the streams assume constant heat capacity flow rates unacceptable is not solely due to the inefficiency of the chosen
and that the overall heat transfer coefficients remains fixed. configuration. Bell (1984) has pointed out that the whole
The correlations are then normally expressed in the form method is only approximate since it is based on certain simpli-
fying assumptions. Any violation of the simplifying assump-
FT=f{R, P) (2) tions will tend to have a particularly significant effect in areas
where of the FT chart where slopes are particularly steep. For exam-
ple, in an exchanger with a close temperature approach, bun-
dle and baffle bypassing not accounted for in the approach
Contributed by the Heat Transfer Division for publication in the JOURNAL OF
HEAT TRANSFER. Manuscript received by the Heat Transfer Division February can result in an exchanger that is not just inefficient but is
23, 1987. Keywords: Heat Exchangers, Modeling and Scaling. thermodynamically infeasible. Any uncertainties or inac-

304/Vol. 110, MAY 1988 Transactions of the ASME


Copyright 1988 by ASME
Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 09/14/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
1 Shell Pass- 2 Tube Passes
Temperature Temperature

FT

0 0-1 02 0-3 0-4 0-5 0-6 0-7 0-8 0-9 10

Fig. 2 A line of constant slope can be defined to separate regions of


preferred design from regions of unacceptable design

been acknowledged previously in qualitative form by Taborek


(1979, 1983b) and Liu et al. (1985), but they do not appear to
be widely practiced. This is probably because no easily inter-
preted rules of thumb are available.
These arguments can be made quantitative by introducing a
line of constant slope, that is
(dFr/dP)R = const
The expression for (dFr/dP)R for 1-2 design has been derived
byMitson(1984)as
(dFT/dP)R = [FT/P(l-P)]

- [(1 - P)F2T/^2P] [13/(2 - P/3) - a/(2 - Pa)] (5)


0 0-1 0-2 0-3 0-4 05 0-6 0-7 0-8 0-9 1-0
where a = 2 - V2 and 0 = 2 + V2.
A line of constant slope that follows the same slope as the
Fig. 1 Three basic temperature situations can be encountered when point FT = 0.75 at R=\ has (3FT/dP)R = - 2 . 8 , and is il-
using 1-2 exchangers lustrated in Fig. 2. Design can be carried out with far greater
confidence above the line of constant slope irrespective of
FT>Q.15. The line of constant slope shown in Fig. 2 was
curacies in design data also have a more significant effect chosen simply to be in line with current acceptable design
when slopes are steep. practice. More conservative design would require a line of less
Consequently, if we are,to be confident in a design, we steep slope, say for example (dFT/dP)R = - 1. The value of the
should avoid those parts of the FT chart where slopes are slope to be tolerated is at the discretion of the designer, similar
steep, irrespective of F r > 0 . 7 5 . These considerations have to the lowest value of the FT correction factor in the tradi-

Nomenclature

A = heat exchange area,


trr
CPr = heat capacity flow rate PN,2N - overall thermal effec- AT C = cold stream
of cold stream, tiveness for N 1-2 temperature change,
kW/C shells in series C
CP = heat capacity flow rate Q = exchanger heat duty, ATH - hot stream
of hot stream, kW/C kW temperature change,
FT = logarithmic mean R = heat capacity flow-rate C
temperature difference ratio, equation (3) ATLM = logarithmic mean
correction factor T 's = inlet and outlet temperature dif-
iV = number of 1-2 shells temperatures to an ex- ference, C
in series changer, C U = overall exchanger heat
P = heat exchanger thermal Ta = cold stream inlet transfer coefficient,
effectiveness, equation temperature, C kW//i22C
(4) TCo = cold stream outlet XP = Ratio of actual to
maximum (asymptotic) temperature, C maximum thermal ef-
value of P, equation THj = hot stream inlet fectiveness in a single
(7) temperature, C 1-2 exchanger, equa-
1,2 thermal effectiveness THo = hot stream outlet tion (7)
for a single 1-2 shell temperature, C {dFT/dP)R = slope of FT line
l,2max maximum (asymptotic) ATt = temperature difference against P for a given
value of Pi 2 , equation at exchanger hot end, value of R on the FT
(7) C chart

Journal of Heat Transfer MAY 1988, Vol. 110/305

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 09/14/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use


tional approach. The slope becomes a particularly important or the FT slope too large. If this happens, they may be forced
consideration when 1-2 exchangers are to be used in situations to consider either different types of shell or multipe shell ar-
in which the overall heat transfer coefficient varies ap- rangements (Fig. 5). The present paper will concentrate on
preciably over the length of the exchanger (see Bannerot and multiple shell arrangements of the 1-2 type. The profiles
Mahajan, 1978). shown in Fig. 5 could in principle be achieved either by two
1-2 shells in series or a single 2-4 shell. These two ar-
3 A Simple Design Criterion rangements will be considered equivalent in design terms,
displaying the same temperature profiles. Various ways of
Although the constant slope criterion outlined in the evaluating design situation that require multiple 1-2 shells in
previous section is an effective means of avoiding undesirable series will now be explored.
areas of the FT chart, it is rather complex to evaluate and use.
A simpler approach, which retains the essential advantages of (a) Traditional Design Practice. Traditionally, the
the constant slope approach, is possible and will now be designer would approach a problem requiring multiple shells
developed. by trail and error. The design begins by assuming a number of
This alternative approach is based upon the observation shells, usually one in the first instance, and the FT is
that for any value of R there is a maximum asymptotic value evaluated. If the FT is not acceptable then the number of shells
for P, say Pm3X, which is given as FT tends to -<. This in series is progressively increased until a satisfactory value for
represents the limit of thermal effectiveness in feasible 1-2 ex- FT is obtained for each shell. The task is simplified by having
changer design for given heat capacity flow rates. An expres- subcharts that evaluate the performance of the overall unit for
sion for P m a x has been stated by Taborek (1983a) and also arrangements of 2, 3, 4, etc., shells in series (see Bowman et
derived by Mitson (1984); it takes the form
P, = 2/(i? + l+Vtf 2 + l) (6)
A 1-2 exchanger designed for P=Pmax will not be feasible.
However, exchangers with P<Pmax will be feasible. We can
define a practical design to be limited to some fraction of Jmax
according to
P=XPPmm, where 0 < A > < 1 (7)
and XP is a constant defined by the designer.
A typical profile for equation (7) with XP = 0.9 is shown in
Fig. 3 and compared with the line of constant slope for
(dFT/dP)R= - 2 . 8 . It can be seen that the line XP = const
follows the same basic profile as the line of constant slope.
In fact, the relationship between the slope (dFT/dP)R and
the parameter XP can be expressed concisely (Fig. 4). To ob-
tain the simple mapping between the constant slope and con-
stant XP criteria, P = XPPmax = 2XP/(R + 1 + Vfl2 + 1) is sub-
stituted into the expression for (dFT/dP)R. In practice, the
slope of the line R=\ on FT charts can be taken as a
reasonable median from which to obtain XP values against
(dFT/dP)R. In summary, two competing criteria have been
developed for exchanger design that are equally good. The
constant XP approach is simpler to use than the constant slope
approach. However, we have only considered designs that re-
quire single 1-2 shells. Once designs that require multiple
shells are considered, the constant XP approach shows even
more significant advantages over the constant slope approach.

4 Design with Multiple 1-2 Shells


Designers often encounter situations where the FT is too low
0 0-1 0-2 0-3 0-4 0-5 0-6 0-7 0-8 09 1-0
XP
1 Shell Pass - 2 Tube Passes
Fig. A The relationship between lines of constant slope and parameter
Xp can be determined knowing the value of ft

Temperature Temperature
FT

+- temperature
' crosses
smaller
0 0-1 0-2 0-3 0-4 0-5 0-6 07 08 09 10 <c=>
i> *1>
Enthalpy Length
Fig. 3 A line of constant slope can be closely approximated by a line of
constant X p Fig. 5 Two 1-2 shells in series or a single 2-4 shell

306/Vol. 110, MAY 1988 Transactions of the ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 09/14/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use


al., 1940). We emphasize again that achieving minimum FT (2) P being constant for all shells as a result of each step
may not be sufficient to achieve a satisfactory slope. being geometrically similar.
The traditional approach can be demonstrated using a sim- Since both R and P are constant for each step then FT must
ple example. A hot stream is to be cooled from 410C to also be constant.
110C by exchange with a cold stream being heated from 0C Recognizing that this method gives constant FT in each shell
to 360C, using 1-2 shells. For the problem overall /? = 1.2 and is conservative in not allowing any temperature cross, a
and P=0.73. Try one shell initially and the problem is simple modification to the technique can allow us to step-off
infeasible. for any prespecified FT.
Adding another shell in series does not solve the problem
since FT is still infeasible. Three shells return a feasible (c) Stepping-Off With Prespecified FT. Rather than
F r = 0.65 against the appropriate chart for three shells. This stepping-off as shown in Fig. 6, we could step-off with a
FT is too low and yet another shell is required. For four shells temperature cross as shown in Fig. 7. Providing the same
F r = 0.8 against the corresponding chart. This looks to be a geometry is maintained for each step, then FT must remain
reasonable solution to the problem as far as F r >0.75 is con-
cerned, leaving for the moment FT slope considerations.
Although this basic approach is the one most commonly
Stepping-off for FT >0-75
used to determine the minimum number of 1-2 shells in series,
trial and error is in fact not needed. Various ways of tackling
this will now be explored.
(b) Simple Stepping-Off. The simplest method for
h Temperature
multishell design is "stepping-off" as discussed by Bell (1983)
and Liu et al. (1985). The method is illustrated in Fig. 6.
Starting at the final cold stream temperature we project across
horizontally to the hot stream, then vertically down to the cold
stream and so on. Each step then corresponds to one shell. 410
The basis of this method is to terminate each shell such that 360
the outlet temperature of the hot stream equals the outlet
temperature of the cold stream, giving zero temperature cross.
This is often conservative in design because, in practice, some
degree of temperature cross in each shell can usually be
tolerated. Therefore, fewer shells are often required than sug-
gested by this method. FT > 0-75
One important point that should be understood about sim- FT = 0-75
ple stepping-off is that each shell will have the same FT. This 110
results from:
(1) R being constant as a result of heat capacity flow rates
being constant.
Enthalpy
Stepping - off
4 Shells suggested
Fig. 7 Stepping-off can also be performed for temperature cross to
achieve a prespecified value of FT in each 1-2 shell
u Temperature

i Temperature

410
360

110c

110

Enthalpy

Enthalpy
5 Shells suggested
Fig. 8 The slope of the first step is given by AT.,, fi, and P 1 2 (which
Fig. 6 The simplest method for multishell design is "stepping-off" depends on X p ). Continuing to step-off with the same slope will deter-
(with no temperature cross allowed) mine the number of shells required to satisfy a given value of X p

Journal of Heat Transfer MAY 1988, Vol. 110/307

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 09/14/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use


constant in each shell. The designer can then step-off with a 5 An Analytical Procedure
specified FT as follows.
It is possible to transform the graphical stepping-off pro-
The unknown temperatures for the first step, THo and Tci,
cedure into a more convenient analytical equivalent by making
can be calculated from R, FT, THh and TCo using equations
use of the expression of Bowman (1936). This relates the
(2), (3), and (4). This then establishes the step slope and the re-
overall P for TV shells in series of 1-2 design, PNi2N, w r t n the
mainder of the problem can be stepped-off with the same
individual P for each shell, P l i 2 , and is of the form:
slope as shown in Fig. 7. (Note that the slope referred to here
in stepping-off is fundamentally different from the slope of N.2N =f{R, P, 2 , TV)(11)
(dFT/dP)R discussed earlier.) If we apply equation (7), the definition of XP, to this unit
It makes no difference whether the FT for stepping-off is overall then
chosen on the basis of traditional rules of thumb, such as
FT>0.15, or a chosen value for (dFT/dP)R, becauseF T is con- =f(R, XP P ,
,2max
stant for each step in either case. However, we have yet TV) (12)
another possibility for a stepping-off criterion, that is, a Recognizing from equation (6) that Pi, 2max is only a function
chosen value for XP. As we will see, this offers definite advan- of R,then
tages over (dFT/dP)R = const.
= / ( * , XP, TV) (13)
(d) Stepping-Off With Specified X P . The temperature If we invert this function we can evaluate TV directly
span covered by a single 1-2 shell will determine its value of P.
Next, we know that P m a x for each shell is determined fully by N=f-^R,PN,2N,XP) (14)
the ratio of heat capacity flow rates, R, in equation (6). Details of the derivation have been given by Ahmad (1985)
Introducing XP, say XP = Q.9, will fix P through equation and the final expression is
(7). The value of P so chosen in each shell, P 1 2 , is sufficient R*l:
information to place the steps. The size of the first step is eas- N=ln[(l-RPNi2N)/(l~PNi2N)]/\n W (15)
ily deduced by reference to Fig. 8.
P , 2 for a given step-size is defined by where
Pli2 = ATH/(ATl+ATc) (8) W= (R + 1 + -JR2 + 1 ~2RXP)/(R + 1 + Vi?2 + 1 -2XP) (16)
Since R = ATC/ATH, then equation (8) gives R = l:
Pia = ATH/ATl+RATH) (9) N= (PN,2N/1 - P, 2 N)(1 + (V2/2) - XP)/XP (17)
Equations (15)-(17) return a value of TV that satisfies precisely
which can be rearranged to a chosen value of XP throughout the series of 1-2 shells. Tak-
ing the number of shells to be the next largest integer above TV
ATH = ATl[Plt2/(\-RPl.2)] (10)
will result in XP becoming marginally improved (decreased)
Since P12 is determined fully from R and XP, equations (6) over the originally specified value. These expressions evaluate
and (7), the step slope given by ATH ultimately depends only explicitly the minimum number of 1-2 shells required and, at
on the designer's choice of XP. the same time, ensure that each shell in the design satisfies the
To complete the design we carry on stepping-off with the
same step slope. Keeping the same step slope ensures the same
Xp, the same (dFT/dP)R, and the same FT in each shell. Temperature
Figure 9 shows the example introduced earlier with the steps
completed for XP = 0.9. This corresponds to FT>0.75 in each
shell and the problem requires four shells, but now ensuring !
410
that uncertain areas of the FT chart according to XP = 0.9 have
been avoided. If a more conservative value of XP is chosen, 360

say XP = 0.8, then the step slope will be different, resulting in 310J/t5V^
a different number of shells to solve the problem.
! 210 <^y?k"
! 160>^
,, Temperature
110

i /t60

410 Enthalpy

360

overall

899
N
crf ^% = 990

= 1 660
110 N

Xp = 09(ie.F T = 0-75] N, ffi = 1 239

N = 4 Shells N,n = 0 520

Nm, = 0 470
0
Enthalpy
Fig. 10 No matter how we divide the problem into sections, there is
Fig. 9 Stepping-off to meet a given value of Xp is consistent with complete additivity among the parts in determining the overall number
achieving the corresponding value of FT of shells

308/Vol. 110, MAY 1988 Transactions of the AS ME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 09/14/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use


required FT slope criterion given by the specification for XP. for the computer models. This is being explored in current
Applying this analytical approach to the problem in Fig. 9 research.
with XP = 0.9 we obtain the answer that 3.9 shells are required On a different subject, additivity also allows the prediction
(four shells in practice). of the minimum number of shells required for the heat ex-
The expressions given above for the number of shells are changer network of an entire process ahead of design, allow-
simple and straightforward to use. FT slope requirements for ing predesign optimization to be carried out (Linnhoff and
each individual shell are observed through the definition of the Ahmad, 1986).
XP parameter. The reason the method is so straightforward is
that we no longer include FT explicitly in the analysis. For a 7 Conclusions
given problem R is fixed and, XP having been specified, FT
has also been specified implicitly but is not used directly. The following points can be made concerning the XP
The authors believe that the introduction of the XP parameter approach as introduced in this paper. For 1-2 shells
parameter both simplifies and improves procedures for the in general:
8
design of heat exchangers: significantly better exchanger designs than those based on
(i) Better designs are generated than those based on tradi- rules of thumb for minimum FT, such as F r > 0 . 7 5 ;
tional rules of thumb, such as F r > 0 . 7 5 , because uncertain similarly good designs as those based directly on constant
areas of the FT chart are avoided more effectively. FT slope.
(;7 ) For multishell arrangements the number of shells can A new criterion for 1-2 exchanger feasibility has been pro-
be evaluated explicitly, requiring neither trial and error nor posed that does not relate to FT. Instead it is based on the
graphic constructions. limiting 1-2 exchanger effectiveness, that is, Pmax.
(HI ) For evaluating the number of 1-2 shells in series, In the case of multishell arrangements we have shown that
recourse to FT charts is no longer necessary. different FT criteria can be interpreted as variations of
stepping-off. The XP parameter approach is also represented
as a modification of stepping-off. Next, it has been shown that
the XP parameter is ideally suited to obtaining a simple expres-
6 Additivityan Important Property sion determining explicitly the minimum number of 1-2 shells
in series for multishell designs. This alters the traditional
Now that we have the inversion design approach for multishell arrangements.
N=f~l(T's, desirable i > ) It should be borne in mind, however, that any results at this
level of exchanger design based on minimum FT, FT slope, or
=f-\T's,XP) (18) XP are still subject to detailed evaluation. Given that any of
these F T -based results are really initializations, we do not need
one further most important property can be demonstrated.
to be absolutely precise. More importantly, design practice re-
Consider again the example introduced earlier. If, instead of
quires speed and simplicity at this stage of design. The XP
solving the overall problem for the number of shells, we solve
parameter approach achieves fast evaluation at remarkable
the problem in parts and add up the parts, will we always get
reliability.
the same result?
This is demonstrated in Fig. 10. The problem overall re-
quires 3.889 shells. If we divide the problem arbitrarily into Acknowledgments
two parts, S, and T as shown, then part S requires 2.899 and The authors would like to thank their colleague, Dr. G. T.
part T requires 0.990 giving a total of precisely 3.889. It does Polley, for useful criticism and help in the preparation of this
not matter how many sections we divide the problem into and paper, and the Journal referees for their useful comments.
how big the sections are, we always get identically the same
result provided we work in fractional shells. When the References
problem is divided into four arbitrary parts A, B, C, and D,
adding up the individual shell requirements gives precisely Ahmad, S., 1985, "Heat Exchanger Networks: Cost Tradeoffs in Energy and
Capital," Ph.D. Thesis, University of Manchester Institute of Science and
3.889 again. Technology, United Kingdom.
Concerning consistency between FT charts and the Bowman Bannerot, R. B., and Mahajan, K. K., 1978, "The Caloric Temperature Fac-
equations we have now established the theoretical link be- tor for a 1-2 Heat Exchanger With an Overall Heat Transfer Coefficient Vary-
tween the number of shells N and FT via the parameter XP. ing Linearly With Tube Side Temperature," AIChE Symp. Series, Vol. 74, No.
171, p. 61.
We are told that for a given FT (as specified by R and XP) the Bell, K. J., 1983, Heat Exchanger Design Handbook, Vol. 3, Hemisphere
number of shells is additive over any arbitrary number of sec- Publishing Corporation, Washington, D. C.
tions over the exchanger profile so long as we maintain the Bowman, R. A., 1936, "Mean Temperature Difference Correction in
same FT in each section. Additivity in these charts has always Multipass Exchangers," lnd. Eng. Chem., Vol. 28, p. 541.
Bowman, R. A., Mueller, A. C , and Nagle, W. M., 1940, "Mean
existed but probably gone unnoticed because: (i) FT charts Temperature Difference in Design," Trans. ASME, Vol. 62, p. 283.
could not be constructed for fractional values of TV and (ii) Kern, D. Q., 1950, Process Heat Transfer, McGraw-Hill, New York.
The inversion N=f~x{T's, FT) has not been expressed Linnhoff, B., and Ahmad, S 1986, "SUPERTARGETING, or the Op-
previously. timisation of Heat Exchanger Networks Prior to Design," World Congress III
of Chemical Engineering, Tokyo, Japan.
The additive property takes on fundamental practical Liu, Y. A., Pehler, F. A., and Cahela, D. R., 1985, "Studies in Chemical
significance when complex problems are addressed where heat Process Design and Synthesis. Part VII: Systematic Synthesis of Multipass Heat
capacity flow rates and overall heat transfer coefficients are Exchanger Networks," AIChE Journal, Vol. 31, p. 487.
not constant. In this context, there exists no additivity in the Mitson, R. J., 1984, "Number of Shells Versus Number of Units in Heat Ex-
changer Network Design," M.Sc. Dissertation, University of Manchester In-
traditional approach that would allow the overall problem to stitute of Science and Technology, United Kingdom.
be decomposed into meaningful subproblems. Faced with a Taborek, J., 1979, "Evolution of Heat Exchanger Design Techniques," Heat
complex problem the designer has no recourse at present by to Transfer Eng., Vol. 1, p. 15.
resort to the use of detailed computer models. Development of Taborek, J., 1983a, Heat Exchanger Design Handbook, Vol. 1, Hemisphere
the current approach may lead either to a reduction in the need Publishing Corporation, Washington, D. C.
Taborek, J., 1983b, Heat Exchanger Design Handbook, Vol. 3, Hemisphere
for detailed computer models or at least to better initialization Publishing Corporation, Washington, D. C.

Journal of Heat Transfer MAY 1988, Vol. 110/309

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 09/14/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Вам также может понравиться