Sie sind auf Seite 1von 6

Heat Transfer

COMPACT HEAT EXCHANGERS PART 3:

Retrofit Heat
Recovery Networks
With Plate Exchangers
Graham Polley,
www.pinchtechnology.com
Pinch analysis retrofit procedures can be
Christopher Haslego, extended to handle compact exchangers.
Alfa Laval

T he first two articles in this series demonstrated


the benefits of including alternative heat ex-
changer technologies at the very outset of the
design of new plants. This article looks at how
such a consideration can be made in revamp projects.
The recommended approach for the revamping of heat-
recovery networks (1) consists of three stages:
Such a plot usually shows that the energy consumption
of the plant is high for the amount of heat recovery surface
area being used. For instance, in the example shown in Fig-
ure 1, the minimum amount of hot utility required by the
plant is that associated with point A rather than point X.
To obtain energy savings, it is necessary to invest. The
investment comes in the form of additional heat-recovery
1. The application of economic analysis to the overall surface strategically positioned in order to improve the
heat-recovery problem, to individual component parts of manner in which the existing surface is used. Consequent-
the plant, and to combinations of those parts in order to ly, the retrofit follows a relationship similar to that shown
identify the potential benefits of linking the component in Figure 1.
parts and the linkage (if any) having the best economics. It is not possible to determine this relationship without
2. The determination of the best means for making the undertaking a very large amount of detailed design investi-
identified linkage (i.e., the actual change to be made to the gation, but it is possible to put a bound on its location. Tjoe
plant structure). and Linnhoff suggested that the way in which the existing
3. The combination of equipment design and economic plant used heat-recovery area could be characterized by an
analysis to identify the detailed design that achieves the de- area efficiency, defined as the ratio of the ideal heat-re-
sired economics. covery area needed for the current hot utility consumption
This article will review how the procedure was applied divided by the actual amount of heat-recovery surface
to an example using shell-and-tube heat exchangers (1) and used. They then suggested that the bound on the location of
then repeat the analysis using plate-and-frame exchangers. the relationship would be a curve involving the installation
of new area such that the retrofit projects had the same area
Traditional retrofit economics efficiency as the existing plant (Figure 2).
The traditional approach to retrofit economic analysis Given this bound, it becomes possible to relate ex-
was developed by Tjoe and Linnhoff (2). It starts with pected hot utility saving with additional heat-recovery
comparing the performance of the existing plant with what area (Figure 3). This, in turn, can be converted to a plot
is theoretically possible. This is done by superimposing of monetary saving against financial investment (Figure
plant performance (in terms of hot utility consumed and 4). By comparing the economic criteria (simple or differ-
heat-recovery area used) on a plot of the relationship be- ential payback, and investment ceiling) with this plot, the
tween hot utility consumption and the theoretical minimum scope (i.e., utility saving) of the retrofit project could be
heat-recovery area (Figure 1). identified.

46 www.cepmagazine.org November 2002 CEP


Energy Saved Saving

Area Added
Path of Retrofit Investment
A A
X
A X

Ideal Area B

E E

Figure 1. Comparison of actual and possible performance and Figure 3. Relating savings and investment based on the retrofit path.
typical retrofit path.

Project consistent with required


differential payback

Constant Area Slope = 1/differential


Efficiency payback Project
consistent
with
Saving

Economic maximum
Projects X investment
A
Project consistent with
Actual A required payback

Ideal A Slope = 1/payback Investment


Ceiling

E Investment

Figure 2. Division of area/energy space.


Figure 4. Relating project scope with the economic criteria.

The extension of this approach to encompass the use of


a new exchanger type (e.g., plate-and-frame exchangers) in tacting pattern is equivalent to making heat recovery
a plant using shell-and-tube exchangers has two problems. matches that are vertically aligned between the compos-
First, since the performance of the new exchangers will ite curves. The composite curves can then be divided into
differ substantially from that of the existing units, how is enthalpy intervals and the heat-recovery area (A) for
the area vs. energy plot developed? Second, what basis is each interval (Aint) estimated by:
there for using the concept of area efficiency?
1 Q
Why heat-recovery area is used inefficiently Aint =
LMTD
hi (1)
i
The estimation of the heat-recovery area used in the
area vs. energy plot (Figure 1) assumes that the best use of where LMTD is the log-mean temperature difference, Qi is
temperature driving force is achieved if hot and cold the heat load on stream i in the interval j, and hi is the heat-
streams are in pure countercurrent flow. This thermal con- transfer coefficient for stream i.

CEP November 2002 www.cepmagazine.org 47


Heat Transfer

If the hot streams and the cold streams exhibit uniform


heat-transfer coefficients (i.e., all the hot streams have the Hot Stream
same coefficient and all the cold streams have the same Cooler
coefficient [which may differ from that of the hot
streams]), the estimate of heat-recovery area for the inter-
val is rigorous. Cold Stream
The heat-recovery area for the network (Anetwork) can be Heat Recovery Heater
determined by summing Eq. 1 over the enthalpy intervals: Unit

Figure 5. Simple network components ideally aligned.


Anetwork =
1 Q
LMTD j h i
(2 )
Where the heat-transfer coefficients for the individual
streams differ from each other (which occurs in most real 174
cases), Eq. 2 provides an approximation of the required
area. Consequently, a small deviation between actual E5
area and predicted minimum area can be due to differ- Zone D
E1
ences in individual stream coefficients. However, major
deviations are due to poor use of the temperature driving 160
4
force i.e., deviation from the pure countercurrent
E3
flow. This situation occurs if the heat-recovery matches New Match
are not ideally placed i.e., they are not vertically 100
aligned. Tjoe (3) recognized that one objective of a 5
retrofit should be to improve the use of the temperature Zone A
driving force by improving the alignment of matches be-
tween the composite curves.
It should be recognized that there are situations in
which non-vertical alignment is very beneficial and Figure 6. Position of new match automatically aligns.
should be retained. One example is where expensive ma-
terials are required for the contacting of a small number the plot itself, for the new area is being added ideally. The
of hot and cold streams but not for the bulk of the pro- retrofit curve would follow the plot.
cess. It is often cheaper to match these particular Now consider how two components of a plant are
streams than to try to integrate them into the rest of the linked through a new heat-recovery match. Figure 6 shows
process. Similarly, where pressure drop is of particular how two components (A and D, which are part of the ex-
importance (e.g., refrigeration systems, compressed ample discussed below and depicted in Figure 7) would be
gases, etc.), power costs can outweigh exchanger costs linked. The position of the new match is dictated by the
and the use of abnormally large temperature driving temperature field. The new match pairs hot stream 4 from
forces is justified. component D, which has a supply temperature of 160C,
with stream 5 from component A. The objective is to re-
An alternative approach to retrofit economics duce the load on the heater on stream 5, which has a sup-
Silangwa (4) found that the assumption of constant area ply temperature of 100C and is currently heated using ex-
efficiency tended to be conservative. He suggested that a changer E1. The outlet temperature of the hot stream leav-
better relationship could be obtained by assuming that the ing exchanger E1 is 174C (which is above the supply
new area is added ideally. This results in a curve that is temperature of hot stream 4). It is obvious that the new
parallel with the area vs. energy plot. match must be positioned prior to E1, at the cold end of
The questions now are: Can we justify this ap- stream 5. Since the temperature field dictates the match
proach? Does this open up the way to handling alter- position, the match will be close to being an ideal match.
native technologies? So, we can expect the retrofit economics to again follow
Consider the simple heat-recovery system illustrated in the area vs. energy plot.
Figure 5. If the performance of this system was positioned The use of the assumption of ideal placement is a valid
on an area vs. energy plot, it would fall on the plot itself. approximation, and the concept of area efficiency is not
This is because the heat recovery matches are vertically necessary. The targeting curve is given by the area vs. ener-
aligned and recovery is achieved in a pure countercurrent gy plot starting at the current energy consumption.
manner. If additional heat-recovery surface was added to Most significantly, in the retrofit the focus is on the new
this system, the resulting performance would also fall on exchangers. Consequently, an area vs. energy plot derived

48 www.cepmagazine.org November 2002 CEP


Component D Component Component Component A
second part is only economically justi-
C B fied if plate-and-frame units are used.
30 84 174 327
C E5 E1 1 The hot utility saving resulting from
these projects would be 7.3 MW. How-
45 121 160 160 220 ever, we see that by integrating all of the
C E3 4 E2 2 components, a saving of 10.6 MW with
15.13 a 0.6-yr payback may be possible.
60 110 220 100 253 300 The identification of structural
C E4 3 5 H
H changes involving two independent
3.0 components is usually easily achieved
35 164 through inspection. The situation in
6
which four components are to be
9.03 linked is harder to resolve it could
80 125 140 188 300 be done by inspection, or the stream
7 9 H
cascade tables introduced by Amidpour
7.88 9.6 22.4
(5) could be used to systematically de-
60 170
8 velop a solution. Here we will identify
6.6 structural changes by inspection, guid-
ed by the process composite curves or
Figure 7. Example problem decomposed into four components.
(perhaps more clearly) by the driving-
force plot (Figure 8). The driving-force
plot displays the temperature separa-
on the basis of the new exchanger technology can be used tion of two composites as a function of temperature. Both
to develop the relationship for the retrofit economics. representations indicate that the available temperature
driving forces are small over the entire process. This ob-
Example viously strongly influences the plant structure.
This example was originally developed by Tjoe and The first step in deriving the new structure is to identify
Linnhoff (2) and subsequently used by Polley and Amid- the sources and sinks of heat that are to be matched. In-
pour (1) to demonstrate retrofit design technique. Here it is spection of Table 2 leads to the following conclusions:
updated to current energy costs ($3/MBtu) and capital
costs (using the equations in the previous articles). The
changes in utility and capital costs do not change the result Table 1. Projects yielding a 2-yr payback
of the retrofit for the shell-and-tube design. The existing (shell-and-tube design).
network is shown in Figure 7. The projects with a 2-yr pay- Component or Combination Energy Savings, MW
back (1) are listed in Table 1. A 2.22
Based on the new economics, the amount of energy AB 1.92
saved (by the 2-yr-payback projects) remains at around
4.7 MW, and the redesign required a new match between AC 3.52
stream 4 and stream 5. Increased heat recovery and AD 4.70
changes in the temperature driving force also required BC 0.48
modifications to exchangers E1, E5 and E3. The capital BD 0.10
cost of the heat-recovery equipment was estimated to be
ABCD 4.88
$0.7M. By undertaking two projects, one involving the
linkage of components A and D and one involving the Source: (1).
linkage of components B and C, it is possible to save 5.18
MW with a 2-yr payback, whereas the traditional analysis Table 2. Identified projects for a plate-and-frame design.
indicated that it was only possible to save 4.88 MW.
Combination Utility Saving, Payback, Temperature
When plate-and-frame exchangers are considered, other MW yr Approach, C
options emerge (Table 2).
ABCD 10.6 0.6 6
One choice is a project involving a structural change (Full System)
linking zones A and D, which would save 4.7 MW with a
0.2-yr payback, plus a change involving linking zones B ABD 7.9 0.4 6
and C, which would save an additional 2.6 MW with pay- AD 4.7 0.2 14
back of 0.4 yr. The first part of this project is identical with BC 2.6 0.4 6
the solution derived for the shell-and-tube exchangers. The

CEP November 2002 www.cepmagazine.org 49


Heat Transfer

with stream 5 (zone A) would be made at the supply end


500 of this cold stream (temperature 100C). A match with
stream 9 would also be made at the supply end (tempera-
400 ture 140C). Given these temperatures, working from the
supply end of stream 4, the match with stream 9 should
Hot Temperatures, C

come before the match with stream 5.


300 Now lets consider the transfer between stream 3 (zone
C) and stream 9 (zone B). This match must be made at the
Hot vs. Cold x=y supply end of stream 3. The temperature span of the match
200
is similar to that of the existing match between stream 2
and stream 9. Given the close temperature driving forces
100 throughout the process, this indicates that the match must
be made by splitting stream 9.
The match between streams 4 and 9 will also have to in-
0
0 100 200 300 400 500 volve splitting stream 9.
Finally, lets consider linkage between zones A
Cold Temperatures, C
Minimum Temperature Difference = 5C (stream 1) and B (stream 9). Because of the close tem-
perature approaches, this should be made at the hot end
Figure 8. Driving force plot for example problem. of the streams.
The new structure for the plant is shown in Figure 9.
Existing matches are shown in green. New matches are
shown in orange.
Stream 1 With the structure determined, the next step is to size
Stream 6 the new exchangers incorporated in that structure. Ref. 1
Stream 4 outlines the procedure for doing this. In heat-balance
terms, the additional heat-recovery equals the reductions
Stream
in the amount of heat rejected in the coolers (which can
7 be considered the heat source) and the amount consumed
in the heaters (which can be considered the sink). It can
Stream 5 only be achieved where there is a path between source
and sink. The paths available in the new network are
Stream 2 shown in Figure 10.
Stream The changes to the network are:
Stream 9 3 new heat-recovery match between streams 1 and 9
Stream
8 enhancement of heat-recovery unit E1 (due to in-
creased recovery and/or reduced temperature driving force)
enhancement of heat recovery unit E5 (due to reduced
temperature driving force)
Figure 9. Structure for retrofitted plant. new heat-recovery match between streams 4 and 5
new heat-recovery match between streams 4 and 9
enhancement of heat-recovery unit E3 (due to reduced
temperature driving force)
A new match between stream 4 (zone D) and stream 5 new heat-recovery match between streams 3 and 9
(zone A) would significantly reduce the load on the heater enhancement of heat-recovery unit E4 (due to reduced
on stream 5. temperature driving force).
A match between stream 3 (zone C) and stream 9 The enhancement of existing units can, depending on
(zone B) would reduce the load on the heater on stream 9 the changes, be achieved in a variety of ways:
by around 2.6 MW. the use of heat-transfer devices
The load on the heater on stream 9 needs to be reduced replacement of tube bundles
further; around 3 MW must be transferred from either zone replacement of exchangers
A or zone D. Given the small temperature driving forces, it installation of supplementary units next to existing ones.
is probable that both transfers will be needed. At first glance, these changes look quite complex. How-
Lets now consider the transfers from stream 4 (zone ever, the complexity can be substantially reduced by using
D). Any new matches must be made at the supply end of 3-stream plate-and-frame exchangers. One possible solu-
the stream (the supply temperature is 160C). A match tion is shown in Figure 11. Multi-stream units are shown

50 www.cepmagazine.org November 2002 CEP


full benefits of compact heat exchanger technology, it is
E5 E1 essential that the application of that technology is consid-
ered from the very outset of design. This is why the
Stream 1 incorporation of that technology into process integra-
Stream 6 tion/synthesis techniques is important. CEP
E3 Stream 4

Stream In Part 1, (September, pp 3237) there was an error in the


7 sample calculation illustrating how to determine the overall
heat-transfer coefficient (U). In Step 5 (p. 34), the value of x/k
Stream 5
for 5-mm thick stainless steel should be 0.000189 h-ft2-F/Btu
E2 and not the 0.00189 h-ft2-F/Btu reported. Otherwise the calcu-
E4
Stream 2 lation is correct, giving an overall coefficient of 49 Btu/h-ft2-
Stream F. Fortunately, the typo has no bearing the U-values calculated
Stream 9 3 in the examples given. The online version of the article is cor-
Stream
8 rect and can be downloaded from www.cepmagazine.org.

Figure 10. Network paths used for increasing energy recovery. Literature Cited
1. Polley, G. T., and M. Amidpour, Dont Let the Retrofit Pinch
Pinch You, Chem. Eng. Progress, 96 (11), pp. 4348 (Nov. 2000).
2. Tjoe, T. N., and B. Linnhoff, Using Pinch Technology for Process
Retrofit, Chem. Eng., 93 (8), pp. 4760 (Apr. 28, 1986).
Stream 1 3. Tjoe, T. N., Retrofit of Heat Exchanger Networks, PhD Thesis,
Stream 6 Univ. of Manchester Institute of Science and Technology, Manch-
ester, U.K. (1986).
Stream
4 4. Silangwa, M., Evaluation of Various Surface Area Efficiency Crite-
ria in Heat Exchanger Network Retrofits, MSc Dissertation, Univ.
Stream
of Manchester Institute of Science and Technology, Manchester,
7
U.K. (1986).
Stream 5 5. Amidpour, M., Applications of Problem Decomposition in Process
Integration, PhD Thesis, Univ. of Manchester Institute of Science
and Technology, Manchester, U.K. (1997).
Stream 2

Stream 3
Stream 9 GRAHAM T. POLLEY (Phone: +44-1229-585-330; Fax: +44-1229-585-708;
Stream
8 E-mail: gtpolley@compuserve.com) is retired from the Univ. of Manchester
Institute of Science and Technology, Manchester, U.K., where he was the
director of the Centre for Process Integration. To promote the application of
process integration technology, he has developed the website
www.pinchtechnology.com. His research activities have involved
Figure 11. Retrofit using multi-stream exchangers.
condensation heat transfer, boiling heat transfer, heat-recovery-system
design and the development of process and equipment design
methodologies. He is the current president of the U.K. Heat Transfer
by touching matches; enhanced units are shown in green. Society. In 1990, he was awarded the Moulton Medal by the Institution of
Chemical Engineers for his work on oil refinery revamping, and in 1992, he
Note how the use of multi-streaming has reduced the com- and Dr. Nasr were awarded the Ackrill Trophy by the U.K. Heat Transfer
plexity of the stream split on stream 9. Society for their work on heat-transfer enhancement. He holds BTech, MSc
and PhD degrees in chemical engineering from Loughborough Univ. of
Concluding thoughts Technology, and is a member of AIChE.

By understanding exchanger technology and knowing CHRISTOPHER HASLEGO is a design engineer with the Process Technology
how to properly specify and size compact exchangers, Div. of Alfa Laval (5400 International Trade Dr., Richmond, VA 23236;
one can realize significant energy savings. However, the Phone: (804) 236-1318; Fax: (804) 236-1360; E-mail:
benefits of the technology are not restricted to energy chris.haslego@alfalaval.com). His focus is heat transfer in the inorganic
base chemicals industry. He received a BS in chemical engineering from
savings or to savings in equipment capital cost. They ex- West Virginia Univ. A member of AIChE, he also maintains a website
tend to reduced pipe complexity, smaller footprints and dedicated to chemical engineering entitled The Chemical Engineers
potentially large savings in installation cost. So, we will Resource Page at www.cheresources.com.
end as we started, by stressing that in order to exploit the

CEP November 2002 www.cepmagazine.org 51

Das könnte Ihnen auch gefallen