You are on page 1of 24

Computers &em. Engng, Vol. 15, No. 12, pp. 809-832, 1991 0098-1354/91 $3.00 + 0.

00
Riated ia Great Britain Pcqml-ion Press plc

HEAT RECOVERY BETWEEN AREAS OF INTEGRITY


S. AHMAD~ and D. C. W. Htn
Centre for Process Integration, UMIST, P.O. Box 88, Manchester, M60 IQD, U.K.

(Received 14 February 199f;final revision received 18 July 1991; receivedfor publication 29 August 1991)

Ahatraet-Process plants are often divided into logically identifiable regions, each having associated
processing tasks. Such regions are necessary when considering the practical features of plant design such
as operational flexibility, safety, plant-layout, etc. The regions are therefore more generally described as
“areas of integrity”. This paper introduces the subject of heat recovery between areas of integrity. While
this is new to the heat integration literature the problem is well known in practice.
It is straightforward nowdays to predict the minimum overall energy consumption for such heat-
integrated systems. However, this can be achieved by several different schemes for heat flows between
the areas. The task then becomes identifying those schemes which offer maximum area integrity when
seeking minimum energy, with say the least number of interconnections between the regions. This paper
develops some understanding of the problem and gives a procedure for finding the required heat flows
between areas. The results are made available as targets, before any network design. Thereafter, a method
is shown which translates the heat flow schemes into actual network designs.

1. INTRODUCTION reduction in operating costs can be significant and


worthwhile. A simple example in Fig. 2 demonstrates
Within a single process there are certain regions this. Areas A and B contain only two streams each
which can be readily identified, for example “front and are separated by a large distance. Long pipe-runs
end,” “back end,” “reaction section,” “separation are required if heat exchange is to be achieved
section” and so on, see Fig. la. Indeed, a whole between the areas (Fig. 2a). By recovering heat only
plant site similarly displays identifiable regions com- within each area (without heat recovery between
posed of the various individual processes and the the areas) and assuming a minimum temperature
utility system(s), (Fig. 1b). Broadly speaking, there- approach of AT,, = 2O”C, the minimum overall
fore, a plant is normally designed to have logically hot utility requirement is (Qd..t+ tQ,iJ, =
identifiable regions, each having defined processing 1400 + 0 = 1400 kW and the minimum overall
tasks. cold utility requirement is (Q,-_& +(Qo,,& =
The existence of such regions is a necessary and 0 + 1350 = 1350 kW (Fig. 2b). However, allowing
practical feature of plant design. The regions are heat recovery between the two areas and treating
usually defined on the basis of several requirements them as a single combined system reduces both
such as operational flexibility (start-up, shut-down, the hot and cold utility consumptions by 450 kW,
control, operational independence, etc.), safety (Fig. 2~).
(emergency shut-down, explosive materials, hygiene, In general, the minimum energy requirement of
etc.) and plant layout (maintenance, access, roads, a system composed of several areas of integrity can
etc.). Such regions are therefore more generally be easily predicted. This simply means treating all
described as “areas of integrity”. the hot and cold streams from all the areas wllec-
Consider now the possibility of allowing heat tively as a single heat recovery problem. However,
recovery between areas of integrity. There is there generally exists a prohibitively large number of
no contradiction implied here because heat inte- heat recovery schemes able to achieve the resulting
gration need not violate area integrity if the “energy target” (Linnhoff and Flower, 1978). Some
correct measures are taken. Such measures typically of these schemes will make considerably more inter-
include identifying those matching schemes between connections between the areas than others. Some
the areas which are appropriate to the type of schemes may not make full use of heat recovery
integrity sought. In addition, auxiliary heaters and within each area before seeking integration with other
coolers, or even buffer tanks, may become necessary areas. Some schemes may make interconnections
in practice. between the areas which seriously compromise their
Although extra capital is implied by using operational flexibility or other desired features of the
such measures to maintain the area integrity, the design.
Consider now some possibilities for network design
TCurrent address: Aspen Technotogy, Inc., 251 Vassar in the example of Fig. 2 to see how heat recovery can
Street, Cambridge, MA 02139, U.S.A. be achieved between the two areas. The additional
810 S. AHMAD and D. C. W. HUI

(4 recovery? What are the energy penalties if any of


these heat flows are disallowed? In the case of our
I example in Figs 2 and 3, it can be shown that only
one match with a duty of 450 kW is needed between
areas A and B to obtain full heat recovery (Fig. 4a).
Importantly, the location of this match can be deter-
mined without trial-and-error or exhaustive search.
After this match is placed there are a number of
possible networks to complete the design. One such
solution is shown in Fig. 4b.
These kinds of results are made available by
a procedure to be developed in this paper. The
procedure works in two stages. First, the necessary
heat flows are found between the areas to achieve
minimum overall energy subject to a specified value

n
of ATti. Some degree of design flexibility, however,
is also sought to allow the engineer to steer towards
!3 “choosing” a good solution. For example, after find-
k ing the few essential heat flows, what are the different
m
choices for the additional heat flows to complete the
B
a solution? How are alternative solutions generated?
3
How are individual preferences handled regarding
enforced or disallowed heat flows? These features
are incorporated in the first stage of the procedure.
Next, each heat flow may require more than one
Fig. 1. Examples of identifiableregions in: (a) a process;
and (b) a whole site. match. The second stage of the procedure finds these
matches and, again, the designer is informed of the
choices available.
capital cost required here by the long pipe-runs would The procedure operates at the targeting stage,
have to be compared with the energy cost savings. using only the stream data, and makes the results
To begin design using conventional techniques the known before any network design. When used
engineer could develop an initial network using the interactively, the procedure allows the designer to
Pinch Design Method (Linnhoff and Hindmarsh, enforce certain preferences, typically based on a
1983) such as in Fig. 3a. This initial minimum energy judgement of those schemes which are felt to be
network has two matches between the areas (shown practically viable in a given project. From a more
in bold), but is subject to some evolution to trade academic viewpoint, or at least a somewhat more
off the energy and capital requirements. The number mathematical approach, the procedure can automati-
of inter-area matches can be reduced from two cally generate all the possible heat flow schemes for
to one by sacrificing 150 kW of the energy savings, minimum energy.
(Fig. 3b). The subject of heat integration between areas
The engineer now faces the question of whether a of integrity is further complicated by the fact
better network may have been possible using a differ- that heat exchange can be performed either by direct
ent design strategy. What is really the correct amount transfer or indirect transfer. That is, either by the
of heat transfer between the areas? How many direct exchange of heat from a process stream in
matches are required to achieve this? Luckily, only one area to a process stream in another area as in
two areas were considered here, but if three or more Fig. 5a, or alternatively, using a heat transfer
areas of integrity were examined, the problem would medium such as steam or hot oil to convey the heat
have been much more difficult. indirectly (Fig. 5b). Normally, direct heat transfer is
Although the subject of heat recovery between viable within a single area of integrity since the
areas of integrity is known to be often addressed streams and equipment therein are subject to similar
in practice, and usually with good results, there or related operational constraints. If this were not
seems to be virtually no treatment of it in the the case the region would be better regarded as more
literature. For the design community in general, than one area of integrity. For heat integration
no systematic procedures have appeared which tackle between regions, however, indirect heat transfer
the problem. is sometimes more practical and cheaper in terms
This paper considers a number of questions of the capital employed. This is often the case
towards understanding heat recovery between areas when considering heat recovery between different
of integrity. For example, how does the designer individual processes on a plant site and where the
know which few interconnections are essential heat recovery over long distances is better effected
between the areas to achieve a scheme for full heat through indirect transfer via the utility system.
Heat recovery between area8 of integrity 811

ATmir, = 2OT ATmh = ZPC


Area A
950 lcw

\
1400 kW

Long
Piimns
QOkW
Area B

W/D

20.0
12oYJ 3ooc 4
5.0
1350 kW
,‘“I
[

Fig. 2. Example problem of heat recovery between two areas of integrity: (a) streams in areas A and B;
(b) without heat recovery between areas; and (c) with heat recovery between areas.

Area

Area a (kW/‘=C)

A 2.5

B 20.0

A 20.0

B 5.0

90” 300
I I
(Temperatures: T; Duties: kw)
Fig. 3. Design for heat recovery &tween the areas in Fig 2: (a) minimum energy design; and (b) evolved
design.
812 S. - and D. C. W. Hm

00

Area ,110” ix &W~-3


A 110-4 2.5
* 0 & I 50”
B 20.0
170”
A 20.0
450 ,
1200 30-m 5.0
B I
’ 90”

1100 D (kWPC)
Area I
A 2.5

B 20.0

A 20.0

B 5.0
150 300
’ 90”

(Temperahues: T; Duties: kw

Fig. 4. An alternative design for heat recoverybetween the areas in Fig. 2: (a) essentialmatch@); and
(b) completed minimum energy design.

A typical example of this in practice is heat inte- The development in this paper is largely based
gration between the crude unit and vacuum unit in a on allowing direct heat transfer between areas of
refinery. integrity. A case study of area integrity in a given

(a)

Area A AreaA

Direct Indirect
Heat Transfer Heat Transfer

Fig. 5. Heat recovery between areas using: (a) direct heat transfer; and (b) indirect heat transfer.
Heat recovery betweenareas of integrity

process is used to demonstrate the procedure. How-


ever, the paper will also show that with a little
refinement the methods are also readily applicable
to the case of indirect heat transfer between areas of
integrity.

2. EXTENSIONS OF PREVIOUS WORK

In the absence of virtually any previous work


which explicitly addresses area integrity, this section
briefly discusses those techniques which display
potential for obtaining useful results to the problem.
Of the existing methods that could be adapted to
tackle the question of heat recovery between such
areas, the mathematical programming approaches
and the pinch technology approach are probably
the most obvious. Both types of methods have
their advantages and disadvantages, and both need
to make assumptions to simplify the problem. Never-
theless, some useful guidance can be offered to
Fig. 6. Overall procedurefor designingheat recoverynet-
the designer by these techniques towards a viable works using mathematicalprogrammingtechniques.
solution as well as alternative solutions.
2.1. Mathematical programming needs to become:
Mathematical programming methods for heat
exchanger network design have developed signifi- min 5 5 Wuyu ,
cantly over about the last 10 yr and are currently at i j

a stage where a solution (or solutions) optimal in where ur/is a binary integer variable (0 or 1) determin-
energy (Papoulias and Grossmann, 1983), number of ing whether a match exists between hot stream i and
units (Cerda and Westerberg, 1983), heat exchange cold stream J There are W hot streams and C cold
area (Saboo et al., 1986; Colberg and Morari, streams. W, is a weighing factor to discriminate
1990) or total cost (Floudas et al., 1986; Yee and against a match between streams i and j. To minimize
Grossmann, 1990) can be generated automatically. inter-area matches, large W, are chosen for these.
An attractive feature of such methods is that they This then formulates the area integrity problem for
readily incorporate decisions for constraints (for- a given overall energy consumption. The MILP
bidden matches or disallowed heat flows) in the model minimizes (subject to WJ the number of
problem definition. stream matches rather than the number of heat
Most mathematical programming procedures for exchangers, so that matches with more than one
heat recovery problems begin by establishing the exchanger (cyclic matches) cannot easily be predicted
minimum energy requirement using the transpor- or avoided.
tation or transshipment model. Subsequently, for a The procedure has some drawbacks regarding
given network energy consumption, mixed integer involvement from the designer when steering towards
linear programming (MILP) is used to evaluate the preferred solutions. It is difficult to assess which
minimum number of heat exchanger units. Various match decisions are responsible for either causing or
network structures may be generated to simul- relieving penalties in the energy target. There is little
taneously achieve the energy and units targets. The knowledge of how the solution is developed thereby
resulting networks are then subject to non-linear making it difficult to know of other choices avail-
programming (NLP) to minimize the total annual able for completing the solution, or indeed any
cost of energy plus capital in the design. The overall knowledge of more practical solutions which are near
procedure is summarized in Fig. 6. the “optimal’* one. Also, the designer is not told
The formulation can be slightly modified to tackle which matches are essential and those which may be
area integrity. Firstly, the minimum energy require- substituted for other matches. Such features in the
ment is still found using the transportation model. design method are important because of the nature of
For the minimum number of matches the objective how area integrity problems arise. After all, safety,
function has to be changed to minimize the number operability, layout, etc. are not usually amenable
of matches between areas prior to establishing the to the type of cost functions used in mathematical
matches inside each area. In other words, the original programming. In brief, whether or not integrity
objective function in the MILP: among the areas can be maintained despite the heat
integration requires the kind of user-interaction
ming CJJ~, perhaps lacking in this approach. Although not
1 J providing an entirely practical tool, the mathematical
814 S.-andD.C.W.Hm

programml ng
‘ techniques can nevertheless be useful AT,, and shows the minimum overall heat surpluses
to this problem. In particular, the NLP can be used and deficits at any temperature for a heat recovery
to optimize total cost once an acceptable design system. The shaded areas in Fig. 7a reveal pockets
structure has been identified by the designer. where process-to-process heat exchange can take
place. Importantly, the curve also reveals the “pinch”
2.2. Pinch technology location. This is the temperature at which zero heat
The pinch technology approach is driven by a flow is required in the heat cascade. Thus heat
thermodynamic understanding of the heat integration exchange is to be avoided across the pinch in mini-
problem. In recent years the method has gained mum energy designs. Overall, the system acts as a
widespread popularity for being easy-to-use by hand net heat sink above the pinch and a net heat source
and for giving significant physical insight into the below the pinch.
design problem. With this background consider the scope for heat
One of the most important tools of pinch technol- recovery between two simple processes, A and B in
ogy is a diagram of the heat cascade for a set of Fig. 7b. Morton and Linnhoff (1984) have shown that
stream data (Fig. 7a). This diagram is called the at least an amount @,+a of heat can be recovered
“Grand Composite Curve” (Linnhoff et al., 1982). between the processes. That is, the degree of hori-
It is drawn in shifted temperatures to allow for zontal overlap with positive temperature difference

(a)

Process B

Fig. 7. Heaq recovery betweenareas of integrityusing pinch methods: (a) Grand Composite Curve for
a singleprocess;(b) overlapof Grand Compositesfor two processesto show maximumheat recoveryusing
steam; and (c) in some cases, more than one steam level may be requiredfor maximum energyrecovery
between processes.
Heat recovery between areas of integrity 815

between the curves. They further show that if this Curve (such as hot oil or tempered water) then heat
heat recovery is made by using a constant tempera- recovery can be further improved. This can be seen
ture heat transfer medium, say by raising steam at in Fig. 8a by noticing that section XY of the profile
temperature Ti from process A and rejecting it against for process A can in fact feasibly supply heat to
process B, then IQ- is the maximum amount of heat section WZ of profile B. The argument is more clearly
that can be recovered in this manner. (The steam demonstrated for the same processes in Fig. 8b where
heat transfer medium is shown as two different lines the profile of process B is now inverted. This gives a
with arrows in opposing directions to indicate the better reflection of the net heat demand and rejection
utility has different shifted temperatures on the dia- profiles as seen in “countercurrent” exchange for the
gram depending on whether it accepts or rejects heat.) two heat cascades. This reveals that more tempera-
Furthermore, because all the information is provided ture difference actually exists between the processes
pictorially an elegant feature of this approach is to than apparent from the “cocurrent” heat exchange of
be able to identify process and/or utility system these profiles in Fig. 8a.
modifications which can improve the scope for such If the process Grand Composites of the two pro-
heat recovery. cesses (that is, two areas of integrity) overlap in
Morton and Linnhoff (1984) did not, however, temperature, which is very often the case in practice,
address further important enhancements of this tech- then some important observations now follow:
nique which are now considered here.
Firstly, not all problems are as simple as the one (i) the maximum amount of heat recovery
described in Fig. 7b. Sometimes there may need to be between the processes can be achieved by
more than one indirect heat transfer level between the direct heat transfer, Fig. 8;
processes to achieve maximum recovery using con- (ii) various selections can be made for indirect
stant temperature media (Fig. 7~). In other cases heat transfer between the processes, but in
with several areas of integrity there may be several general a sloping profile for the medium is
options available for choosing how to perform the sufficient to achieve the same heat recovery as
heat recovery and all of which are equally effective in a direct heat transfer scheme, Fig. 9a;
energy terms. This will be discussed later. (iii) if a constant temperature heat transfer
If the indirect heat transfer medium is not medium is used then this offers equal or less
restricted to a constant temperature (such as steam) heat recovery than a medium with a sloping
but has a sloping profile on the Grand Composite profile, Fig. 9b.

(a) (b)

Fig. 8. (a) Section XY of process A can transfer heat to section WZ of process B. (b) Full scope
for heat recovery between processesA and B is obtained from countercurrentprofiles for the Grand
Composites.
816 S.-and D. C. W. Hm

04

ProcessA Process A

Process B

Fig. 9. Indirectheat transferbetweenprocessesA and B with a mediumof: (a) slopingtemperatureprofile


(hot oil); and (b) constant temperatureprotile(steam).

For completeness, we can also state that if the Grand instead involving all three steam levels, such as shown
Composites of the two processes do not overlap at all in Fig. 1Oc.
in temperature then: Although the pinch technology approach is in
keeping with some of our original objectives, that of
0) a constant temperature indirect heat transfer allowing user-interaction with the solution develop-
medium offers the same degree of recovery as
ment and giving some physical understanding to the
a sloping profile medium;
problem, it nevertheless has some drawbacks. Firstly,
(ii) indirect heat transfer offers the same degree
it is difficult to establish which heat flows between
of heat recovery as direct heat transfer between
regions are essential and which are optional or can be
the processes.
substituted for others. Next, the Grand Composite
Not apparent from the discussion so far is that Curves do not easily show how to manipulate around
when more than two Grand Composite profiles are disallowed heat flows to find the resulting energy
to be examined according to the above technique, penalties. Furthermore, the selection of heat flows
the problem can become difficult to handle, with can be remarkably difficult if interactions exist among
the possibility of missing schemes offering maxi- them and it can become easy to miss minimum energy
mum energy recovery. For example, Fig. 10a shows schemes.
the Grand Composite profiles of three processes. Before moving. on to an alternative approach to
If steam is used as the heat transfer medium then this problem, a useful direction can be set by briefly
the amount of heat recoverable between processes reconsidering the example in Fig. 10. The maximum
A and B is Z&a, between B and C it is ZQac heat recovery in this example was constrained by
and between A and C it is IQ,,=. However, the using steam levels-the problem was being tackled
total amount of heat recovery is less than for indirect heat transfer. However, the full potential
ZQ,+ZQ,c+ZQ, because the three component for heat recovery actually exists when direct heat
amounts cannot all coexist simultaneously. For transfer is allowed between the areas. Once the direct
example, if all of ZQAc is achieved, it is impossible heat transfer problem is solved, with a knowledge of
to achieve all of ZQAa and ZQX, (Fig. lob). The how much heat should flow between the different
heat flows between the processes therefore display areas, then the designer can choose to achieve these
interaction: the choice of one amount prejudices the flows using indirect heat transfer between the areas.
other amounts. This leaves the designer with the If inadequate profiles are chosen for the indirect heat
worry that perhaps there is another scheme which transfer media then the energy penalties will become
offers more recovery than using all of ZQ, but clear.
Heat recovery between areas of integrity 817

(a)

IQAC

Process B
@AB

IQAC

IQBC I

Fig. 10. Possibilities for indirect heat transfer between three proces ses using different steam levets:
(a) maximum individual amounts of each steam level; (b) choosing the maximum amount of ZQAc here
precludes any amounts of ZQABor IQ=, . and (c) a different scheme with alJ three levels gJves more overaJJ
heat recovery.

With the direction in mind, the following sections show heat is available from the hot streams in areas
consider direct heat transfer between the areas. The A, B and C. The vertical labelling shows heat is
results can be subsequently used for indirect heat required by the cold streams in these areas.
transfer solutions. Importantly, however, the pro- If no heat recovery is allowed between the areas
cedure to be developed will define the correct (Fig. 1 la), then crosses appear in the matrix; the ticks
amounts of heat flow between areas in minimum in the shaded boxes simply show that heat recovery
energy solutions. Later on, the choices for matches is allowed within each area. The minimum overall
able to satisfy the heat flows will also be found. energy consumption of the scheme in Fig. 1 la is
QTsT= Q,, + Qm, + QtimC, where TST abbreviates
3. TARGETS: A NEW APPROACH TO AREA INTEGRITY “totally separated target”. If, however, heat recovery
is allowed between all the areas of integrity then
This section develops a method for finding all the ticks are applied throughout (Fig. 11b). The mini-
necessary and sufficient heat flows between areas of mum overall energy consumption of this scheme is
integrity in minimum energy solutions. The approach evaluated by treating the streams from all areas as a
requires the matrix representation shown in Fig. 11. single heat recovery problem. The energy target for
Here, three areas of integrity (A, B and C) are this scheme is QTIT where TIT abbreviates “totally
considered. The matrix is labelled horizontally to integrated target”.
818 S. AHMAD and D. C. W. HUI

fT.T.T_J

AreaB

Q TST = Q minA + QminB + Q minC

QTIT=Qmin(A+B+C)

Fig. Il. Matrixrepresentation


and energytargetfor: (a) totallyseparatedsystem;and (b) totallyintegrated
system.

Now Q, G &I-ST and the difference Qm= heat flow in any one such cell incurs the energy
Qrsr - Q, defmes the total energy penalty caused penalty shown. It follows that wherever Qn > 0 gives
by disallowing heat recovery between the areas of the location of a heat flow necessary to achieving
integrity. To avoid this energy penalty, not all of the minimum overall energy. In fact, Qn is the minimum
inter-area heat flows shown in Fig. 11b need normally necessary heat flow required in ccl1 i for a minimum
be allowed. To determine which few of these flows are energy solution.
necessary requires some further concepts which are So far, this establishes only those heat flows necess-
now developed. ary in any minimum energy solution, but not all the
heat flows required to complete such a solution. In
3.1. Energy penalty other words, the necessary heat flows may not be all
Rather than considering the energy penalty the suEcient ones for minimum overall energy. This
caused by disallowing all the inter-area heat flows is because each heat flow with Qn = 0 is only un-
(as in Fig. 1 la), it is possible to evaluate the energy necessary when all other heat flows are allowed, but
penalty from disallowing only one inter-area heat this may no longer be true when any of the other heat
flow. For example, in Fig. 12a the individual penalty flows with Qn = 0 are disallowed. A check for this is
Qn = (Qce& - Qnr from disallowing heat flow in the easily made by finding the penalty from disallowing
cell i = CA is determined by forbidding matches all of the heat flows with Qn = 0, while following the
between all hot streams in area C to all cold streams necessaryheat flows with Qn > 0 (Fig. 13a). Ifthis total
in area A. Methods for calculating the “constrained penalty for the remaining cells (Q&x = 0 then no fur-
energy target” Qcxr in such cases are well docu- ther heat flows are required. If (Q&a > 0 then some
mented in the heat integration literature (Cerda et al., of the heat flows with Qn = 0 will need to be allowed.
1983; O’Young et al., 1988). The individual energy
penalties Qn can be found independentlyfor all the 3.2. Energy saving
inter-area heat flows i in the matrix by forbidding Figure 13b now considers which of the heat flows
heat recovery in each cell i but allowing all other heat with Qn = 0 in Fig. 12b are required when (Q& = 0.
flows. Here, one of the heat flows with Qn = 0 is allowed
The individual penalties for the cells in Fig. 12a while disallowing all the other heat flows with Qn = 0.
are now shown in Fig. 12b. Disallowing only the The resulting change in total penalty is termed the
Heat recovery betweem areaa of integrity
820 S. AHMAD and D. C. W. Hm

energy saving of heat flow i, or Q,. This can be done ing all heat Bows except i were allowed. However, we
in turn for all the cells with & = 0. If among these can still think in terms of finding a revised value of
there exists a cell with Qsi = (Q&= then allowing heat Qn for these cells but subject to the one heat flow
flow i recovers all of the total remaining penalty. which has been disallowed. Qpi can therefore be more
Hence, i must be the only additional heat ilow now generally &fined according to the status of the
required. Of course, it is possible that more than one remaining problem. That is, Qti for a given cell
heat flow exists with Qsi = (Q,), and any one of these depends upon which heat flows are disallowed in
can be chosen to complete the solution. other parts of the matrix (Fig. 14b).
The other outcome could instead be that all heat If any of the revised values of Qpi in Fig. 14b for
flows with Qpi - 0 have Qsi < (Qm)R. Hence, two or the remaining cells now show Qpi> 0 then clearly
more of these w-ill be required in addition to the these heat flows did not become necessary until the
necessary heat flows with Qpi > 0. To handle this heat flow elsewhere in the problem had been re-
situation we need to observe that any one heat flow stricted beforehand.
with Qpi = 0 can be disallowed. Doing so will not add The idea of evaluating the performance of a
energy penalty to the problem. This gives a definite cell subject to what is happening in the remaining
stage at which the number of heat flows to be problem can also be applied to the energy saving Q, .
considered can actually be restricted without affecting In general, the energy saving Q, can be defined by
the energy performance of the final solution. This disallowing all heat flows in the remaining problem
is important because the designer can choose accord- except i. In Section 3.2 above, however, the values
ing to individual preferences which of these heat of Qsi were determined subject to having allowed
flows to restrict. Alternatively, this is the stage the necessary heat flows with Qpi > 0. Hence, the
from which multiple solutions can be generated: remaining problem had already been adjusted before
disallowing different heat flows with Qpi = 0 will calculating the values of Q,. Ideally, we should have
now lead to different final solutions, as for example started off the whole analysis by finding Q, for each
in Fig. 14a. cell before allowing necessary heat flows. At that
stage, if any cell had been found with Q, = Q, then
3.3 The remaining problem only this one heat flow i would have been required to
When one of the heat flows in Fig. 14a is restricted achieve minimum energy and the solution is com-
all the other cells which originally had Qpi = 0 will not pleted. Otherwise, all cells have Qsi < QpT and we
necessarily be able to return a zero individual penalty proceed to finding the necessary heat flow with
any longer. This is because Qpi was evaluated assum- Q,+ > 0. The arguement here for energy saving is

(a) @)

.
. . After disallow-
multiple ing one, remaining
. .
solutions cells may no longe
. . have Qpi - 0.
.

QPi
QPi

Fig. 14. (a) There are usually choices for disallowing a heat flow with Qpi = 0. (b) After disallowingone,
the remaining heat flows may no longer show Qpi = 0.
Heat recovery betweenareas of integrity 821

identical to that explained in Section 3.2 except that each area of integrity. No heat flows between the
we are considering a larger remaining problem. areas are yet decided. This initial state can have its
The remaining problem is therefore an important total penalty (Q& evaluated to check if this is zero.
concept determining the reference for calculating If yes, we can stop immediately since there is no scope
energy penalties and energy savings. These values for energy reduction from heat recovery between the
in turn drive the decision-making for which heat areas.
flows are necessary and which can be disallowed. The If (QPT& > 0 then (Q& is evaluated for each cell
remaining problem therefore needs to be treated more i in R to check whether any of these equal (Q&s_
carefully. It can be defined as those cells in the matrix If yes, then only the one heat flow of duty (Q,& is
for which no decision has been made as to whether required to complete the solution and we can stop.
their heat flows are definitely allowed (required) in Otherwise, two or more heat flows arc required and
the final solution or definitely disallowed. In other these are found by calculating (Q&_ Cells with
words, the remaining problem at any stage of devel- (Qr.JR> 0 locate the necessary heat flows in R and
oping a solution constitutes those cells which still these should transfer at least (Q& amount of
need to be determined. Each time a heat flow is heat. Upon allowing such heat flows the remaining
allowed or disallowed towards its place in the final problem R needs to be reset and is now defined to
solution the corresponding cell is removed from exclude these cells since they have been decided.
the remaining problem. In general, therefore, total For the cells in the revised remaining problem
penalty (Q&R 1individual penalties ( Qpi)Rand energy (Qpr)R and (QsdR are calculated to check again if,
savings (Q& are all calculated for cells i which exist given the necessary heat flows, we can now stop
in the remaining problem R. immediately [(QPT)R= 0] or only one further heat flow
is required [(Qsi)R= (Qpr),& If not, the penalties
3.4. Procedure
(Q&R need not be recalculated before proceeding
So far we have seen how energy penalties and because these are unaffected if the remaining problem
energy savings operate within the remaining problem was only changed by allowing heat flows.
to identify which heat flows are required and which At this stage, all (Q& = 0 in R since only necess-
can be disallowed. This understanding can be orga- ary heat flows were allowed in defining R and no heat
nized into a systematic procedure for finding mini- flows have been restricted yet. Any one of these heat
mum energy solutions. flows in R can now be disallowed. Once this decision
The procedure is shown in Fig. 15. It starts with is made the remaining problem R has to be reset
the remaining problem R being defined as R,, which again. Disallowing the one heat flow will cause the
simply consists of only heat recovery allowed within values of (Qn)R for each cell remaining in R to

Need only
matchi

Fig. 15. Procedure.for finding necessaryand sufficientheat flows to allow for minimum solution(s).
822 S. AHMAD and D.C.W.Hm

AH-S.15
AH-15 I5w

=+%tzJzg
no*

E
199
A
MO-
10

AI?%

3;P
Ax-25

Area C
1 AH-123
Area B

(Temperatures: T; Duties: MW)


Fig. 16. Example process with three areas of integrity.

change. So the (Q& values are now recalculated may differ from those shown here depending on the
subject to the disallowed heat flow to see if any type of integrity required by a designer, but the
(QdR > 0. This locates which additional heat flows example given is sufficient to demonstrate the pro-
have become necessary after disallowing the earlier cedure. Before starting this, however, it is worthwhile
one. to examine an analysis which could have been made
The procedure continues in this way, filling in more for this problem without the method in this paper.
and more of the undecided parts of the matrix and Figure 17 shows how much additional heat recovery
reducing the size of the remaining problem. It is above the totally separated target (QTsT) is achievable
essentially a recursive procedure which solves a small by taking different combinations of the three areas.
part of the matrix in one level of recursion and then The designer can soon establish from the energy
applies the same rules all over again to solve the targets in this problem that no two areas taken alone
remaining problem in further levels of recursion. can fulfill as much heat recovery as all three areas
In all cases the procedure is guaranteed to stop with together. However, not all the six possible heat flows
a minimum energy solution. While the procedure between the three areas may be necessary to achieving
tries to minimize the number of heat flows required full heat recovery.
in a minimum energy solution there is no guarantee This is where the procedure developed in this paper
of this in one pass because of the stage where comes into effect (Fig. 18). It is soon found that
the designer chooses which heat flow is disallowed. at least 7.1 MW of heat flow is required from area
Depending upon what choice is made here different C to A in any minimum energy solution for this
solutions with different numbers of heat flows problem and that there will be a choice for other
can emerge. Despite this, the procedure is easily
implemented on a desktop computer and can thus be Table 1. Stream data for pr-s in Fig. 16
used to efficiently generate multiple solutions by T..

back-tracking and disallowing different heat flows at Area stream Type (2, (G5)
the required stage. Used in this fashion it is straight- A 1 Hot 300 60 0.30
A 2 Hot 70 25.00
forward to find all the minimum energy solutions and
to choose the one(s) with the fewest number of heat
A
A :
Cold
Cold
30
3.5
3;
100
0.30
0.25
flows. A 5 Cold 139 140 30.00
B 6 Hot 500 120 0.25
3.5. Example Cold 139 500 0.15
E 8
7 Cold 20 250 0.10
The procedure of Fig. 15 is now applied to the
c 9 Hot 120 119 15.00
example flowsheet in Fig. 16 which consists of three Hot 200 30 0.20
areas of integrity. The stream data are shown in E 10
11 Cold 110 160 0.25
C 12 Cold 200 201 25.00
Table 1. How the areas and their streams are chosen
Heat -very between areas of integrity 823

AdditionalHeat Recovery
above c&T WW

BC,A
t- AB,c

- AC.B

0 @ +
///////N//N//////H/
Totally Separated ’
@ 0 Target
t
Q~sp 69.50hk

ATmin = 10°C
Fig. 17. Analysis of heat recoveriesby combining differentareas in the example process.

heat flows elsewhere required in addition to this. Usually, a better approach is to use the insights
Thereafter, Fig. 19 continues the procedure to choose gained earlier from identifying the heat Aows. For
to disallow the heat flow from area A to C and finds solution 1 in Fig. 19 this means starting the procedure
two resulting solutions. again with the matrix in Fig. 21 b, in which the
Finally, it remains to discover the duties of disallowed heat flows (black cells) are ignored for
the required heat flows in the final solution(s). inter-area matches and the allowed heat flows (white
In minimum energy solutions this can be done by cells) are to be developed further for their matches.
evaluating the individual penalty from forbidding This considerably narrows the search space and
each of the required heat flows in the final solution. assumes that the heat flows identified earlier are
For our example solutions this gives the duties shown indeed the only ones to be accepted in design. This
in Fig. 20. is usually the case if the heat flows have been
carefully selected to consider the type(s) of integrity
required between the areas. For solution 1 in Fig. 19,
4. DESIGN: TRANSLATING HEAT FLOWS
INTO MATCHES
a repeated application of the procedure can give the
matches shown in Fig. 21~.
Once the heat flows between areas of integrity have Although many designs are possible here, the one
been found the next step is to identify some network shown in Fig. 21c comes from first finding matches
design(s) to achieve the scheme. This means finding 6-l 1 and 6-12 to be necessary [(Qpi)R> 0] in Fig. 21 b
the individual streams which should be matched to and then forbidding 9-3,!L4, 9-5 and 9-7 [(Qpi)R= 0]
realize the heat flows. In principle, the matches can all without any penalty. Match 9-8 then becomes
be found by simply using the procedure of Fig. 15 necessary. Finally, matches lo-4 and lG5 can com-
but now assuming each stream to be a separate plete the solution.
area of integrity. In other words, for the example So far, this establishes which streams are to be
in Fig. 16 the procedure can be used starting with matched between the areas of integrity, but not
the initial matrix shown in Fig. 21a. This would how many heat exchanger units are required between
certainly be able to generate valid network designs these streams (that is, the possibility of “cyclic”
but a prohibitively large number of solutions will matches) nor how much heat duty is required for
usually results. This can be a lengthy task. Further- these units. To understand further these questions
more, such an approach would make redundant the about the design requires a brief review of some
useful step of having found the heat flows between recent results regarding the thermodynamics of con-
the areas. straints (or forbidden matches) in heat exchanger
824 S. AHMAD and D. C. W. Hm

start KlDTh = 38.95 MW (Q’ i)R

(QPi h Allow flow C -CA (‘$$R’ 24.95 MW

Fig. 18. The procedureof Fig. 15 applied to the example process. Heat flow CA is found necessary.

networks. The relevant work has been reported by energy penalty can be decomposed and attributed to
Linnhoff and O’Young (1987) and is discussed in the the amounts of:
next section.
. cold utility above the pinch, Q,;
4.1. The components of energy penalty + hot utility below the pinch, Q,,,;
l process-to-process heat exchange across the
Figure 22a shows in outline that if a minimum
pinch, Q,.
energy solution is to be achieved for a given of AT,,
in any heat recovery problem then zero heat flow is A procedure exists for quantifying Q,. QPhand Qrr
required across the pinch. This means having: (Linnhoff and O’Young, 1987). What is important
here is that these three components are additive
l no cold utility above the pinch;
in their contribution to the total energy penalty
- no hot utility below the pinch;
caused by disallowed matches. For convenience, the
l no Ijrocess-to-process heat exchange across the
locations of these violations of the minimum energy
pinch.
target are also shown on the grid diagram of the
Many options normally exist in a network design streams in Fig. 22~.
for satisfying these conditions. However, disallowing
certain heat flows or matches in a heat recovery 4.2. Avoiding energy penalty

problem can prevent any of these conditions from The basic understanding of energy penalty is now
being fulfilled in the completed design. As a result, developed further for the area integrity problem.
forbidden heat flows or matches can incur energy The heat flows or matches determined by the pro-
penalty over the target amount (Fig. 22b). The total cedure of Fig. 15 for a complete design are the ones

(Q&R (%i) R DeIete any 0 penaltyflow

Two possible solutions

Solution1
(%‘i) R Allow flow B - C (QPT)R = 6.1. MW

Solution2

Fig. 19. Completion of the procedure thereafter results in two possible solutions if heat flow AC is chosen
to be disallowed.
Heat recovery between areas of integrity 825

Solution 1 heat exchanger on the relevant side of the pinch.


If the cross-pinch component of energy penalty (Q,,)
for the match is non-zero then a heat exchanger is
required for the streams on both sides of the pinch.
Thus, the possibility of cyclic matches which span
the pinch is detected. Similarly, if a near-pinch
exists in the problem then the possibility of cyclic
matches which span this (Ahmad, 1985) can also be
examined.
6.1 MW Figure 23 shows the penalty components of the
matches for the example in Fig. 21~. This gives the
exchangers between areas as shown with their duties
Solution 2
in Fig. 24a. The rest of the design simply amounts to
matching wifhin each area the remaining streams on
each side of the pinch (Fig. 24b). The Pinch Design
Method (Linnhoff and Hindmarsh, 1983) is sufficient
for this part. The design achieves the hot utility target
of 30.55 MW.

25.45 h%w
(a)

Fig. 20. The duties of the requiredheat flows in the final


solutionscan be found from their finai QR values.

required in a minimum energy solution. Disallowing


such heat flows or matches will incur energy penalty.
More specifically, the energy penalty will be com-
posed of the components Qpc, Qp,, and Qpr, in the
locations of the network shown in Fig. 22~. There-
fore, in order to avoid the energy penalty means the
required heat flows or matches should transfer, on OJ)
each side of the pinch, the duties shown in Fig. 22d.
That is:
total duty = Qpe+ !&, + ZQ,, ,
of which:
duty above pinch = Qp+ Qpp,
duty above pinch = Qp,,+ Qm.
Also, Fig. 22c and 22d together show that in order
to avoid the Qp,, component of the energy penalty
needs an amount 2Q, of duty to be incorporated in Cc)
the required heat flow or match. \
These results are perfectly general. They apply
to any single heat flow or match, or any groups of
these.

4.3. Design procedure


When applied to individual matches, the results in
Section 4.2 are central to understanding how to
identify the heat exchanger duties in a completed
network design for an area integrity problem. Dis-
allowing each match independently, while allowing
all the other required matches, will give the amount Fig. 21. The matchesfor the networkdesigncan be found
of overall heat duty required by pairing the streams by repeating the procedure of Fig. 15 assuming each stream
to be a separate area of integrity. This can be started: (a)
in that match. If the above pinch or below pinch or
with none of the heat flows found earlier, or better still (b)
both components of energy penalty (Q, or Q,,) for using the heat flows found earlier-this gives the matches in
the match are non-zero then this will necessitate a (c) for solution 1 in Fig. 10.
826 S. AHMAD and D. C. W. HUI

QHmin

Qcmin+Qpc + Qpp+ Qph

(a)

Fig. 22. Components of energy penalty: (a) minimum energy system; (b) penalty components in a system
with constrained (forbidden) matches; (c) penalty components in the grid diagram; and (d) components
of heat duty for required matches.

Qph = 10.0 MW

,Qph=6.1MV

Qph - 1.45 h%W


\
Qpc = 24.0 hfW

Fig. 23. Components of heat duty for the matches found in Fig. 21~.
Heat recovery between areas of integrity 827

LcGzi 149’ a! (Mw/W


3ooo I
A q 0.3
25.0
0.25
15.0
0.2

0.1
0.25
25.0

wa 149- !a? (MIw/“c)


A 0.3
A 25.0

B 0.25
C 15.0
C 0.2

A 0.3
A 0.25
A 30.0
B 0.15
B
0.1
C 0.25
C 25.0
1390

Design for QJ-Jmin = QTIT= 30.55 MW


5 inter-area matches.
J
Fig. 24. (a) Matches with their duties for the solution in Fig. 23. These are all the inter-area matches needed
here for minimum energy. (b) Completed minimum energy design. The remaining matches are within each
area.

Once a completed minimum energy design is causing a small increase (4.95 MW) in energy wn-
obtained it can be evolved to simplify the network sumption but the network structure is appreciably
structure and to balance the overall capital require- simplified.
ment against energy consumption. The existence of The final network design is reinterpreted back
heat load “loops” and “paths” in the network allows in the flowsheet in Fig. 25b. It is interesting to note
evolution towards minimizing the total cost (energy that inter-area exchangers do not necessarily require
plus capital) of the design (Linnhoff and Hindmarsh, two pipe-runs between the areas.
1983). For example, in Fig. 24a matches 6-11 and
24 can be removed using the loops they lie on 4.4. Design for indirect heat transfer
(involving the coolers). Match 10-l 1 is cyclic (a loop) Suppose that heat exchange between the areas of
and can be reduced to a single exchanger. Then, integrity is to be achieved using indirect heat transfer.
match l&5 can be removed since it lies on the path As discussed in Section 2.2 this situation can restrict
between the heater on stream S and the cooler on the heat recovery possible between the areas if a
stream 10. The results of such an evolution are shown constant temperature medium such as steam is used.
in Fig. 25a. Two exchangers between the areas Given this, the designer will want to identify those
removed, as well as another two within the areas, schemes which offer better approach to the target
828 S. - and D. C. W. HUI

(a)

After Evolution
a? (Mw/“C)
0.3
25.0
0.25
15.0
0.2

0.3
0.25
30.0
0.15
0.1
0.25
25.0

QH = 35.50 MW with 3 inter-area matches

(b)

Area B I
Fig. 25. (a) After evolutionof the minimumenergydesign, the network is simplified with some additional
energy consumption. (b) The final design is translated back into the dowsheet for the example
process. I

heat flows identified earlier. This needs to he balanced energy design. Prior knowledge of the required heat
against the use of a reasonable number of utility flows between the areas makes it straightforward
levels and few indirect heat transfer matches between to identify the duty of each level on the diagram,
the areas. Thus there is a tradeoff between heat otherwise this would have been very difficult to
recovery and design compltxity. arrive at.
These features can be examined in the example The steam level (ZQ,& has a duty of only
problem considered in Figs 16-25, but. now using 1.45 MW, noticeably smaller than for the other
steam levels to convey heat between the areas. levels, and can be removed with little penalty. This
Solution 1 in Fig. 20a shows the heat flows for will cause the levels (ZQ,& and (ZQcs) to have
minimum energy which are to be approached using their duties reduced by 1.45 MW. This is because
steam levels. How these heat flows distribute can be removing (ZQ& = 1.45 MW, which goes into the
found by overlapping the Grand Composite Curves pocket for area C, requires for heat balance that
for the three areas (Fig. 26a). The diagram offers a (ZQ& is reduced by 1.45 MW out of the pocket
better understanding of the indirect heat transfer (Fig. 26b). Similarly, (ZQ,-e) is reduced by 1.45 MW
opportunities and immediately reveals that five steam in the pocket for area B. Next, it can be seen that
levels should be sufficient here to achieve a minimum ZQca = 4.65 MW can now actually be placed at the
Heat recovery betweenareas of integrity 829

600”

5000
a B
A
2‘l.OMW

4.6sMw
C
r35Mw

__-- AreaA

J
Fig. 26. Indirect heat transfer using steam levels for the example in Fig. 16: (a) five steam levels are n&cd
for minimum energy solution 1 in Fig 20a. Level (IQ,,& can he eliminated with little energy penalty (h)
to give four steam levels.

same temperature level as (ZQ,-J2. Thus three steam for 30.0 MW of hot utility to area A. How the various
levels emerge as this stage. Next, (ZQc-), = 3.5 MW utilities relate and where to locate them among
has a much smaller duty than the other levels and can the integrated areas is easily seen from the Grand
now also be eliminated. The resulting configuration Composite Curves.
with only two levels for indirect heat transfer is For each separate area, the relevant indirect heat
shown in Fig. 27a, with MP steam at 220°C for transfer steam levels with the required duties can
(ZQ& =24.0MW and LP steam at 110°C for be included in the design grid and a network devel-
(ZQc.Jz = 10.0 MW and ZQce = 4.65 MW. oped using the Pinch Design Method (Lint&off and
The remaining utilities to be placed will provide Hindmarsh, 1983). The networks for the separate
net heating and cooling to the process. This is areas are then linked using the steam levels (Fig. 27b).
straightforward since the outstanding profile of each This overall network design may be further evolved
area not satisfied by indirect heat transfer will need to remove small exchangers.
external utilities. However, some judgement is advis- In summary, the analysis reveals how the heat
able because some steam levels for indirect heat flows found from the targeting stage can be re-
transfer could be chosen to coincide with external interpreted on the Grand Composite Curves for an
utilities in order to be practical. For example, if indirect heat transfer solution. This representation
MP steam is available here as an external utility, allows the engineer to evolve away from minimum
say between 210 and 240°C then Fig. 27a shows it energy towards a more practical solution while being
could be used as the steam for (ZQ& as well as made aware of any penalties incurred. Importantly,
S. AHMAD and D. C. W. Hui

Indirect Heat Transfer Solution

0.2

0.25
12.5
201-&_@ a e
25.0
1.0 24.0

Fig. 27. (a) Next, ZQcs can be moved to the same temperatureas (ZQ& and the small duty of (ZQ,-_,),
can be eliminated. This leaves two steam levels. (b) A network design using the two steam levels.

this activity can be carried out on the curves before the solution individual concerns for the type(s) of
developing a network design. integrity sought between the areas.
The procedure given here offers certain insights
5. DISCUSSIONAND CONCLUSIONS regarding area integrity. In particular, these insights
show how to identify:
The basic purpose of this paper is to introduce into
the research literature the problem of heat recovery (a) the few heat flows and matches necessary
between the areas in any minimum energy
between areas of integrity. The problem is frequently
solution; and
encountered in industry and provides a challenging
topic for investigation. The paper opens the subject (b) those heat flows and matches which are
su&%ent to complete such solutions.
by proposing a new and systematic procedure for
designing minimum energy networks which feature Importantly, the method shows that rather than
few interconnections between the areas of integrity. treating area integrity as a problem of enforcing
The development focusses on obtaining a thermo- constraints (forbidden matches) between areas, the
dynamic understanding of the problem and appreci- approach is to fmd which constraints should be
ating the design options available to the engineer. relieved to get minimum energy. If the design scheme
These aspects are important for ensuring design is to be steered away from minimum energy for
flexibility and allowing the engineer to build into practical reasons then the options and consequences
Heat recovery between areas of integrity 831

are made clear. The choices are indicated for the heat (Q& = Energy saving from allowing only heat
flows or matches to select and thereafter the designer flow i in the remaining problem R, kW
Q, = ~~l~$tegrated targetenergyconsump-
can decide whether these will be achieved through
direct or indirect heat transfer. QmT= Todlly separatedtargetenergyconsump-
The subject of area integrity is open to a number tion, kW
of improvements and extensions. Future develop- Qpe= Energy penalty component due to cold
utility above the pinch, kW
ments a;e excepted as follows:
Qph= Energy penalty component due to hot
utility below the pinch, kW
(i) incorporation of capital targets. These can be
Qm = Energy penalty component due to process-
optimized with the energy target to predict to-process heat exchange across the pinch,
minimum total cost solutions; kW
(ii) separate methodologies for area integrity in R = Remaining problem of heat flows awaiting
new design and retrofit. New designs should to be determined
& = Initial remaining problem containing all
be able to choose the best areas of integrity
heat flows between areas but no heat flows
to define whereas retrofits need to improve within areas
designs based on existing or predesignated T = Temperature, ‘C
areas; T. = Stream supply temperature, “C
(iii) multiple utilities with heat and power. Area TT = Stream target temperature, “C
AT,, = Minimum approach temperature for heat
integrity can become an overriding concern exchange, “C
when dealing with a whole plant site contain- WV = Weighting factor to discriminate against
ing several processes. Integration between the a match between hot stream i and cold
processes using the utility system as a heat stream j
viJ= Binary integer variable, 0 to 1, determining
transfer intermediary can offer remarkable
whether a match exists between hot stream
results but this has to be balanced against i and cold stream j
power generation on the site;
(iv) process modifications to improve heat REFERENCES
recovery and area integrity. How areas
relate to each other in terms of creating Ahmad S., Heat exchanger networks: cost tradeoffs
in energy and capital. Ph.D. Thesis, University of
further opportunities for heat recovery and
Manchester, U.K. (1985).
capital reduction depends on how the areas Cerda J. and A. W. Westerberg, Synthesizing heat
can be modified. Process changes may exchanger networks having restricted stream/stream
also exist which actually enhance integrity matches using transportation problem formulations.
between the areas. A technique for finding Chem. Engng Sci. 38, 1723-1740 (1983).
Cerda J., A. W. Westerberg, D. Mason and B. Linnhoff,
beneficial process changes could be extremely Minimum utilitv usage in heat exchaneer network
valuable. synthesis--a transport&on problem. Chem. Engng Sci.
38, 373-387 (1983).
While the above developments represent substantial Colberg R. D. and M. Morari, Area and capital cost
strides to make, this paper provides an important targets for heat exchanger network synthesis with con-
beginning. It is felt that area integrity will become a strained matches and unequal heat transfer cofficients.
more important part of heat integration understand- Compurers them. Engng. 14, 1-22 (1990).
Floudas C. A., A. R. Ciric and I. E. Grossmann, Automatic
ing as industry makes more such problems known to
synthesis of optimum heat exchanger network configur-
the research community. ations. AZChE JI 32, 276290 (1986).
Linnhoff B. and J. R. Flower, Synthesis of heat exchanger
Acknowledgement-The authors would like to thank networks: I. Systematic generation of energy optimal
Professor Bodo Linnhoff for useful suggestions in the networks. AZChE JI 24, 633-642. (t978).
preparation of this manuscript. Linnhoff B. and E. Hindmarsh. The Pinch Design Method
for heat exchanger netwo&s. Chem. Engni Sci. 38,
745-763
-- f\----,-
19831.
NOMENCLATURE Linnhoff B. and D. L. O’Young, The three components of
cross pinch heat flow in constrained heat exchanger
CP = Heat-capacity flowrate, kW “C - ’ networks. AZChE Meering, New York (1987).
H = Enthalpy, kW Linnhoff B. et al., User Guide on Process Integration for
~QAB= Heat recovered between areas A and 0 rhe Eficient Use of-Energy. IChemE, U.K. (1982).
using indirect heat transfer medium, kW Morton R. J. and B. Linnhoff, Individual process improve-
(QrrT)i = Constrained energy target from disallow- ments in the context of site-wide interactions. ZChemE
ing heat flow i, kW Annl. Res. Mrg, Bath (1984).
QHmio(Q,-& = Minimum hot (cold) utility target, kW O’Young D. L., D. M. Jenkins and B. Linnhoff, The
Q,, = Minimum energy target for streams in area constrained problem table for heat exchanger networks.
A, kW ZChemE Symp. Ser. No. 109, pp. 75-l 15 (1988).
(Qpi)* = Individual energy penalty from disallow- Papoulias S. A. and I. E. Grossmann, A structural
ing only heat flow i in the remaining optimization approach in process synthesis-II. Heat
problem R, kW recovery networks. Computers &em. Engng 7, 707-721
(Qpr)R = Total energy penalty from disallowing all (1983).
heat flows in the remaining problem R, Saboo A. K., M. Morari and R. D. Colberg, RESHEX-
kW an interactive software package for the synthesis and
832 S. AHMAD and D. C. W. Hm

anaiysis of resilent heat exchanger networks-II. Discus- Yee T. F. and I. E. Gr oesmann, Simultaneous optimization
sion of area targeting and network synthesis algorithms. models for heat integration--II. Heat exchanger network
Computers &em. Engng 10, 591-599 (1986). synthesis. Computers &em. Engng 14, 116>1184 (1990).

You might also like