Sei sulla pagina 1di 4

0263±8762/96/$07.00+ 0.

00
€ Institution of Chemical Engineers

HEAT EXCHANGER DESIGN THROUGH PARAMETER


PLOTTING
T. K. PODDAR* and G. T. POLLEY²
*Department of Chemical Engineering, New Jersey Institute of Technology, USA
²
Department of Chemical Engineering, UMIST, Manchester, UK

A new heat exchanger design procedure is presented. It can be used with any
existing state-of-the-art exchanger rating program. Rather than systematically
exploring the whole of the available range of exchanger sizes (diameters and tube
lengths), it determines the relationships between duty and tube length, pressure drop and
tube length, etc. for a range of diameters. This information is then used to clearly indicate
the full range of geometries that are suitable for a given duty and given constraints. As the
example presented in the paper shows, this can be achieved by undertaking very few rating
calculations (typically four to six). Conventional design procedures can involve hundreds
of such calculations. The graphical representation provides the designer with clear
guidance regarding the in¯ uence of allowable pressure drop on required geometry. It can
also be used to appraise rapidly the amount of thermal overdesign introduced through the
selection of a speci® c geometry.
Keywords: heat exchanger design; CAD; overdesign

INTRODUCTION This paper presents a graphical technique that over-


Heat exchanger design generally proceeds through the comes these drawbacks and lays bare the in¯ uence of the
examination of the performance of a range of potential `secondary constraints’ on design.
geometries. This is usually undertaken by computer. A The procedure was ® rst developed by one of the
typical program would systematically explore a range of authors whilst working in the Dharamsi Morarji
shell diameters, tube lengths, tube pass arrangements Chemical Co. in 1990. That version was based on the
and number of units in series and/or parallel. It is usually use of analytical equations derived from those presented
left to the designer to specify the range of geometries to by Kern1 for the prediction of exchanger performance.
be covered. Obviously, the larger the range speci® ed, the Unfortunately, the use of Kern’s equations has a number
longer the program takes to execute but the larger the of drawbacks. First, the shellside equation is restricted to
number of potential designs generated. exchangers that have a 25% ba‚ e cut and a ba‚ e
The objective in such procedures is to identify those spacing equal to the shell diameter. To apply the
geometries which satisfy the following constraints: equation to other geometries is to move outside the
bounds of its application. Secondly, Kern’s equations
1. the exchanger is capable of transferring the speci® ed have not proved to be very accurate and are no longer
load; used by designers. Far more accurate `stream analysis’
2. the tubeside pressure drop is less than or equal to a methods2 are now available to the designer in standard
speci® ed maximum value; exchanger design programs. What is needed is a
3. the shellside pressure drop is less than or equal to a methodology that yields results that are consistent with
speci® ed maximum value; these methods.
4. the tubeside velocity is less than a maximum speci® ed In this paper the authors present a procedure that can
value and greater than a minimum value; be applied with any available heat exchanger rating or
5. the tube length is less than a maximum speci® ed value; simulation program.
6. the shell diameter is less than a maximum speci® ed
value. INITIAL GEOMETRICAL DESIGN SPACE
The procedure su ers from at least three signi® cant The principal geometrical constraints are shell
drawbacks. First, the designs generated are often `over- diameter and tube length. The design space is initially
designs’ and, despite conventional wisdom, overdesign is de® ned in terms of the maximum values for these two
not always safe design. Second, the designer must parameters (shown in Figure 1). The `space’ starts as the
undertake a number of runs in order to establish the box ABCD.
in¯ uence of what may be termed the `secondary’
constraints (pressure drop, velocity and tube length) on
design and third, the designer may not necessarily cover VELOCITY CONSTRAINTS
the best range of potential geometries. The velocity is dependent on tube count which, in turn
849
850 PODDAR and POLLEY

Figure 3. P.D. constraints.


Figure 1. Geometrical space.
just a small percentage (usually 10%) of the total
is directly dependent on the shell diameter. The allowable value. The aim is to absorb the rest of the
maximum allowable tube velocity can easily be con- pressure drop in surface friction inside the exchanger in
verted into the minimum acceptable tube count. Then, by order to promote heat transfer. So, for each shell
referring to tube count tables (such as those presented by diameter studied, the straight tube pressure drop per
Kern1 and by Saunders3 , the shell diameter coinciding unit length is determined. By dividing the allowable
with the maximum velocity can be determined. The straight tube pressure drop by this value, the length of
approach can be repeated to yield the shell diameter that exchanger that coincides with the absorption of the
coincides with the minimum acceptable velocity (or, if allowable pressure drop is determined.
appropriate, minimum permitted tubeside Reynolds The relationship between shell diameter and the tube
Number). length for maximum pressure drop can now be
Multiple tube passes are easily catered for by incorporated into the plot. The result could be that
calculating both the pass and overall tubecounts and shown in Figure 3. Acceptable designs lie above and to
again referring to the relevant tube count table. Thus, the left of this line; the design space has now been
velocity constraints can now be incorporated into the reduced to EFIJH.
geometrical space (see Figure 2). The result could well be The data generated on shellside pressure drop per unit
the reduction of the search space to box EFGH. length of exchanger can be treated in exactly the same
way. The result is also incorporated in Figure 3. For this
INFORMATION GENERATED FROM problem it is the shellside pressure drop that is
EXCHANGER RATING controlling and the design space is now reduced to
EFKLH.
Most state-of-the-art programs for the rating of shell- Finally, the information on overall heat transfer
and-tube heat exchangers present a wealth of informa- coe cients and e ective mean temperature di erence
tion on exchanger performance. Virtually all of these can be used to determine the area needed for the given
programs will inform the user of the E ective Mean duty as a function of shell diameter. For each diameter
Temperature Di erence, the Overall Heat Transfer this can then be related to tube length necessary for the
Coe cient, the straight tube pressure drop and the heat transfer duty. This relationship can now be placed
shellside pressure drop for a single ba‚ e space. By on the plot. The result could be that shown in Figure 4.
running a rating program for a series of shell diameters Acceptable designs are those which are above and to the
with a selected ba‚ e con® guration all of the perfor- right of this line; the design space is ® nally reduced to
mance information can be related to shell diameter. MFKN.
The importance of the maximum allowable pressure Any exchanger having the chosen ba‚ e con® guration
drop can be determined as follows. First, allowance is and length and diameter values falling within this space
made for the exchanger nozzles. Normal design practice will transfer (at least) the required heat load whilst
is to design the exchanger nozzles such that they absorb obeying all of the speci® ed constraints.

Figure 2. Velocity constraints. Figure 4. Area requirement.

Trans IChemE, Vol 74, Part A, November 1996


HEAT EXCHANGER DESIGN THROUGH PARAMETER PLOTTING 851

Figure 5. Design margin. Figure 6. Relaxed shellside pressure drop constraint.

DESIGN MARGINS of the minimum and maximum shell diameters. The


The space MFKN may well contain designs that have maximum tube velocity has been set at 2m s- 1. This
considerably better performance than requested. The occurs at a tube count of 62 which is equivalent to a shell
quantity of over surface involved is very easily estab- diameter of 0.34 m. The minimum Reynolds Number is
lished. A 10% shift of the duty line to the right (see set at 10,000. A quick calculation shows that (with tubes
Figure 5) gives designs having 10% oversurface. The of 25.4 mm O.D. and 2 mm wall thickness) this occurs at
e ect of pressure drop on the design `envelope’ is also a velocity of 0.35 m s- 1. This, in turn, is found to occur at
easily determined. Since, pressure drop is also linearly a tube count of 355 which is equivalent to a shell
dependent on tube length, a 10% shift of the pressure diameter of 0.76 m. A state-of-the-art commercial
drop line to the right shows the e ect of increasing computer program is then used to rate exchangers of
allowable pressure drop by 10% (Figure 6). the following diameters: 0.34 m (minimum) 0.44m
0.54 m 0.64 m 0.76m (maximum).
In each case the ba‚ e cut was set at 25% and the ba‚ e
EXAMPLE PROBLEM spacing was set equal to the shell diameter. The results
A shell-and-tube heat exchanger is required for the obtained are presented in Table 1.
following duty. Naphtha, ¯ owing at a rate of This information was then transformed into the length
116,000kg h- 1 (32.22 kg s- 1 ), is to be heated from 90ÊC limits associated with the maximum use of available
to 110ÊC by means of a light oil, ¯ owing at 70,000 kg h- 1 pressure drops and into lengths required for the
(19.44 kg s- 1) which has an initial temperature of 180ÊC. necessary heat transfer. The results give the identi® ed
The allowable pressure drops for each of the streams is range of exchanger geometries that satisfy the required
5 kPa.
Assuming 10% of this pressure drop to be taken up by
Table 1. Results of rating calculations.
the exchanger nozzles, the straight tube and the bundle
allowable pressure drop are both 4.5 kPa. Shellside Tubeside Heat
The physical properties of the two streams are as Overall pressure pressure transfer
follows: heat drop per drop per area per
transfer unit tube unit tube unit tube
Diameter coe cient length length length
m Wm2 K- 1 kPa m- 1 kPa m- 1 m
Property Naphtha Light Oil
0.34 569 2.06 6.37 4.7
-3
Density (kgm ) 720 760 0.44 437 0.77 2.08 8.5
Heat Capacity (kJ kg- 1 K- 1) 2.345 2.43 0.54 336 0.36 0.85 13.5
Viscosity (cps) 0.54 1.5 0.64 285 0.20 0.56 20.7
Th. Conductivity (W m- 1 K- 1) 0.143 0.127 0.76 230 0.14 0.28 28.3

A fouling resistance of 0.007 m2 K W- 1 is to be used. A


heat balance on the Naphtha stream indicates that the
load on the exchanger is 1.511MW. A heat balance on
the light oil stream then indicates that the oil outlet
temperature will be 148ÊC. A single tube pass exchanger
is to be used. Thus, the E ective Mean Temperature
Di erence for the unit is 63.7ÊC.
Following convention, the higher viscosity liquid will
be placed on the shellside of the exchanger. This provides
better overall heat transfer when plain tubed units are
being used4 .
The ® rst step in ® nding solutions is the identi® cation Figure 7. Solution to example problem.

Trans IChemE, Vol 74, Part A, November 1996


852 PODDAR and POLLEY

Table 2. Identi® ed geometrical space. and duties and the principle design variables (in this case
Length to Length to Length
shell diameter and tube length) to identify clearly the
tubeside shellside required space within which successful designs are to be found.
Diameter pressure pressure for heat Since, the relationships between design margin and tube
m drop limit drop limit Area duty length is linear, the level of overdesign presented by any
geometry within the design space can be established
0.34 2.18 0.71 41.7 8.87
0.44 5.84 2.16 54.3 6.39 quickly and easily. Similarly, since the relationship
0.54 12.50 5.29 70.6 5.23 between pressure drop and tube length is linear, the
0.64 22.50 8.03 83.2 4.00 in¯ uence of allowable pressure drop on the design space
0.76 32.1 16.07 103.1 3.64 is also easily established. The results of the exercise are
best displayed graphically.
The procedure can be used manually with any existing
exchanger rating program. By interfacing it with state-
duty, given the speci® ed constraints. They are presented of-the-art exchanger rating programs, it will be possible
in Table 2 and plotted in Figure 7. to produce design programs that are more suitable for
use on desktop computers than are conventional design
CONCLUSIONS algorithms.
Conventional shell-and-tube heat exchanger design
algorithms proceed by systematically rating a large REFERENCES
number of possible exchanger geometries (termed the
1. Kern, D. Q., 1951, Process Heat Transfer (McGraw-Hill, New
design space) and subsequently identifying those which York).
successfully transfer (at least) the required heat load 2. Palen, J. W. and Taborek, J., 1969, Chem Eng Prog Symp Ser,
whilst providing tubeside and shellside pressure drops 65(92): 53±63.
that are less than maximum allowed values. This 3. Saunders, E. A. D., Heat Exchangers (Longmans Scienti® c,
approach can require a large amount of computation. London).
4. Polley, G. T. and Gibbard, I., 1995, Proc Eng, August.
Despite this, the approach does not necessarily provide
the best designs, nor does it give the designer a great deal
of insight into the design problem.
In this paper, a new approach to design is presented. ADDRESS
The philosophy is ® rst to use the constraints placed upon Correspondence concerning this paper should be addressed to Dr
the design to reduce the possible design space quickly G. T. Polley, Department of Chemical Engineering, UMIST, PO Box
88, Manchester M60 1QD, UK.
and systematically. Then, by undertaking just a few
rating calculations (typically six) within this space, to The manuscript was received 22 December 1995 and accepted for
derive relationships between the remaining constraints publicationafter revision 1 July 1996.

Trans IChemE, Vol 74, Part A, November 1996

Potrebbero piacerti anche