Professional Documents
Culture Documents
1 Introduction
In the design of energy systems, First Law "losses" describe tween energy and capital costs requires a certain inefficiency in
energy flows across the system boundaries; Second Law Second Law terms to achieve finite rate processes and finite
"losses" describe lost potential. The latter form of loss is equipment sizes. Numerical relationships between the efficien-
generally accepted as a more useful concept for the engineer cy of processes and their speed, and thus equipment size and
seeking clues for design improvement. capital cost, are outside equilibrium thermodynamics. Once
Second Law losses are usually calculated in terms of irrever- again, inefficiencies and losses are therefore, at least partly,
sible entropy gain or exergy loss. They are evaluated for any inevitable.
system in nature with equations that are well documented and
unambiguous. Early work by Keenan (1932), Bosnjakovic The conclusive study of any practical system would have to
(1935), Rant (1956), Grassman (1959) and many others did involve quantifying and interpreting these three effects. Such
much to popularize Second Law Analysis based on interpretation is difficult. There are no systematic or unam-
equilibrium thermodynamics in terms of available energy, en- biguous procedures known which distinguish practically
tropy or exergy. avoidable losses from those that are inevitable. For point
Somehow, industrial application has lagged behind 3 - capital costs - the technique of thermoeconomics has been
academic enthusiasm. With the exception of perhaps low suggested (Tribus and Evans, 1962; Evans and Tribus, 1965;
temperature systems and power cycles, industrial processes El-Sayed and Evans, 1970); but it is probably fair to say that it
have almost defied improvement through Second Law inter- applies to the optimization of a system of given configuration
pretation. The analysis seems difficult to utilize in terms of rather than to genuine synthesis.
practical design information to be derived from it. This then makes the use of Second Law Analysis in design
This paper suggests that there are three main reasons for very much a skill-based exercise. The analysis itself is exact
this apparent lack of practical results. These are: science. The interpretation unfortunately is not.
Over the past 10 years or so a new design method has emerg-
1 Equipment Choice. Safety, materials, and other limits ed in the chemical industry. This method, "pinch
constrain the choice of practically available equipment. Thus, technology," initially related to heat exchanger networks. It
inefficiencies and losses are often partly inevitable. sets rigorous heat recovery targets for overall networks (sub-
2 Network Interactions. Complex systems usually com- ject to a minimum temperature difference) and identifies, in
prise a number of different unit operations with multiple inter- true synthesis, network structures which achieve the target.
connections and nontrivial interactions. Thus, in the network More recently, pinch technology has been extended to incor-
context losses are rarely caused where they occur and the porate heat pumps and combined heat and power cycles. In
magnitude of a loss rarely relates to the overall savings addition, it now offers capital-energy cost tradeoff optimiza-
available if the loss was avoided. tion during synthesis.
3 Capital Costs. The need for economic tradeoffs be- This paper explains the basics of pinch technology and
demonstrates its relationship with Second Law Analysis. It
shows that the exergy losses in an optimal design developed by
pinch technology are rigorously inevitable. The methods are
Contributed by the Advanced Energy Systems Division and presented at the
1986 Winter Annual Meeting, of THE AMERICAN SOCIETY or- MECHANICAL
introduced with reference to heat exchanger networks. The
ENGINEERS. Manuscript received by the AES Division, September 1985; revised principles are then applied to heat and power systems to
manuscript received May 26, 1989. enable a parallel treatment.
162.5° 50°cr;
0.34 0.15 1.63
STEAM
1
105°c
UTILITIES :
1 25°
172.5°
25"
CW
2.31
Steam-. 5.75 MW 105°c
CW : 9.25MW
Fig. 1 Example heat exchanger network |80°c
Q l = exergy loss(MW)
2 Traditional Second Law Analysis of a Simple Heat utilities usage: 2 2 % less than
Exchanger Network base case.
Fig. 3 Modified heat exchanger network
Consider the heat exchanger network (HEN) shown in Fig.
1. The required temperature changes are achieved by using
heat exchangers between process streams, a heater supplied avoided if the heat exchanger was completely replaced by ideal
with steam at 250 °C and a cooler supplied with cooling water Carnot engines and/or heat pumps, which would exploit the
at 25°C. The network performs the duties described in Table temperature differences in the exchanger while maintaining all
1. It should be noted that the smallest temperature difference process stream temperatures in and out. The rest of the net-
occurring in the network (ATmin) is 10°C and occurs in ex- work and the process would stay as they are. Alternatively, the
changer 2. This is an important parameter for network design. loss could be reduced by reducing the temperature differences
The question is, "Is it possible to reduce the hot utility con- in the heat exchanger. However, this approach would require
sumption?" The traditional approach to exergy analysis (for altering the configuration of the process and the network.
example, King et al., 1972), would be to quantify exergy losses Consequently, because of network interactions, there will be
for each item of plant. Such an analysis gives the results shown changes to the temperatures and conditions in other pieces of
in Fig. 2, where the boxes symbolizing equipment are propor- equipment at the same time.
tional in size to the losses. For the cost data given in Table 1 Let us apply the first approach to the example. The largest
the annualized cost of this network would be M$ 2.10. loss is in the cooler. A heat engine, such as a Rankine cycle,
In a single heat exchanger the exergy loss is due to heat could conceivably be used to recover some of the lost work.
transfer across a finite temperature difference. There are two However, for a practical engine with finite temperature dif-
possible approaches to reducing this loss. First, it could be ferences in the evaporator and condenser, only about 1.0 MW
Nomenclature
120°c 70°c
•037 C1 3.0 35-
171.6°
STEAM
0-
215°c =L 192.5* 150° 50°c
-180
160°
?105°c
1.91 CW m-
r/To=4.44MW 1425"
|80°c Q ] = exergy loss. MW
Fig. 4 Exergy losses in the modified network A1Q5-
M--=ZJM
DATA FOR PROPOSED
— 35 — - 4.0--
HEAT TRANSFER:
+ initial temperature
difference
+ heat capacity
flowrates
IS TEMPERATURE DIFFERENCE NO
CP- •CP
(MW/-C) 020 0.10 0 0.10 0.20 Q 3 0 ( M W A )
SUFFICIENT ?
Fig. 6 Application of problem table algorithm to example problem
' YES 'r
HEAT TRANSFER dQ NO HEAT TRANSFER
FEASIBLE FEASIBLE for a better design but not in any sense directly. The improved
after dQ, temperature design, which resulted from changing the topology of the heat
difference to be recovery network, could have been achieved by mere inspec-
re-established on grounds
of heat capacity
tion aided by repeated simulations. Furthermore, we have no
indication as to whether our improvement is the best possible.
Perhaps, further improvements are yet to be discovered?
There are still exergy losses spread throughout the system (Fig.
Fig. 5 Algorithm for anlayzing heat exchanger network problems 4). To be complete, a design method should be capable of
identifying the point beyond which the losses involved are in-
evitable, or rather, beyond which practical improvement is a
of work could be recovered, assuming that a suitable working genuine impossibility.
fluid exists for the temperatures in question. A more com- We had two approaches. The first was based on exact
plicated system, using two cycles to match the temperature- replacement of individual pieces of equipment by different
enthalpy profiles better, would improve the recovery to 1.4 items (engines, heat pumps) without changing their operating
MW. The capital cost necessary for this reduction in exergy parameters or the network configuration. The second was
loss would be considerable. The necessity of finite driving based on changes to the operating parameters of existing
forces and realistic equipment sizes has a strong bearing on equipment including alterations to the network configuration.
improvements that are practically possible. The two approaches have conflicting effects. For instance,
Let us now apply the second approach. Although the cooler although the second approach achieves a saving of hot utility,
shows the largest loss, it is difficult, by inspection, to conceive the scope for work recovery at the cooler has been reduced.
of design changes to the overall network which would reduce The first approach, which we might call the "textbook" ap-
the temperature differences in the cooler. Concentrating next proach, tends to require complicated, expensive and even
on exchanger 3, which has the second largest exergy loss, it is unrealistic equipment. It is only practical when obvious op-
quite easily possible by inspection to reduce its temperature portunities exist for savings, based on known power recovery
difference in a new network configuration as shown in Fig. 3. or similar equipment. It is also focused on individual units
The new losses are seen in Fig. 4. As Fig. 4 shows, the losses in rather than integrated systems.
all other items of equipment have been reduced alongside that The second approach tends to be more difficult, but is prac-
in exchanger 3. Similarly, the overall requirement for heating tical. It aims at genuine synthesis of integrated systems. It
and cooling has decreased. The overall benefit is larger than utilizes practical equipment. Nevertheless, when large losses
that in exchanger 3 alone. This illustrates the effect of network exist, there is obviously always a case for the first approach
interactions. The annualized cost of the modified network not to be forgotten. In this paper we will use the second ap-
would be M$ 1.82, a saving of 12.9 percent. proach while remembering the first for use where appropriate.
We accept the design in Fig. 3 as the best improvement we As will become apparent, the methods developed allow a
can find by inspection. Exergy analysis has guided the search seamless connection between the two approaches.
(MW/°c
(b) Cold composite curve
Fig. 7 Composite curves
8
3 Exergy Analysis With a Difference - Pinch Tech- Consider load against level.
a
nology Do not degrade levels prematurely.
3.1 Energy Targets. In the case of heat exchanger net- This sounds sensible but vague. Let us then develop the first
work design, there now exists a design method which theme into a logic structure as shown in Fig. 5 (Linnhoff,
minimizes the overall exergy loss for a predetermined trade- 1983). First, to enable heat to be transferred, there needs to be
off between energy and capital costs. The target levels of ex- a sufficient temperature difference. Next, heat can be trans-
ergy loss, and therefore energy use, can be calculated prior to ferred until the temperature difference has become insuffi-
design. cient. In other words, there is a clear hierarchy. First we need
Pinch technology was derived by Linnhoff and his co- level. Level enables heat transfer to take place. Heat transfer
workers (1982) following his work on exergy analysis. It stems gives load. Then load re-establishes level, and so on.
directly from exergy analysis as treated by him (1983). The ex- The second theme helps to develop this logic structure into
ergy principles are expressed in a series of "themes." Those an algorithm for several sources and sinks of heat.
which are relevant to heat exchangers are: "Premature" degradation of levels is heat exchange at ex-
Qc= 6-0
_ f-|
IMW) IMW)
Fig. 8 Combined composite curves Fig. 9 Composite curves for A7"min = 10°C
2000
= 1500-
a
1000-
500-
Fig. 11 The pinch principle: system division and three forms of
cross-pinch heat transfer
ly zero for the minimum feasible energy input. In Fig. 8 there ^5> 50°
C2
is a point of closest approach between the composite curves. ULZBj \Jtj3j \225j
These two temperature points are one and the same, called the
"pinch." L _ J : Heat load (MW)
At the pinch, the process can be divided into two sub- Fig. 12 Grid representation of example problem
systems each of which is in heat balance with its respective
utility. Above the pinch only hot utility is required. The use of
cold utility in this region only increases the hot utility demand amount of the transfer written beneath the lower circle. With
(Fig. 11). Similarly, below the pinch only cold utility is re- this representation any design can be developed using any se-
quired. If heat is transferred from a hot stream above the quence of hot and cold stream matches and any design change
pinch to a cold stream below the pinch, the subsystem heat is possible without reordering or rerouting streams.
balances require that both hot and cold utility loads are in-
However, the main benefit of this representation is that
creased by the amount transferred. A design with no heat
when the pinch temperatures are marked in, heat exchangers
transfer across the pinch gives the maximum energy recovery
transferring heat across the pinch are easily identified (see ex-
(MER) possible.
changers 1 and 2 in Fig. 13). Finally, the grid is a valuable tool
From this thermodynamic analysis a design method has when studying the operability and flexibility of a process
been developed which guarantees networks having the target (Linnhoff and Kotjabasakis, 1986).
energy recovery. Briefly, the design is treated in two parts,
above and below the pinch. Remembering that heating is 3.5 Example Problem. To illustrate the use of pinch
allowed only above the pinch and cooling below, it is usually technology, the grid representation and the design procedure,
possible, by inspection, to design two sub-networks to achieve let us now continue assuming that AT",,,;,, = 10°C does repre-
the targets. In both parts, matches are made between hot and sent the optimum choice for lowest overall cost. It follows that
cold streams in such a way that the number of equipment Qh = 2.5 MW is our energy target. It may be worth noting
items is minimized. After all the process to process matches that this does not so much represent a thermodynamic target
are made, matches to and from the utilities are used to com- as an economic target: while it is possible to achieve designs
plete the solution (Evans and Tribus, 1965). Recognition of with Qh < 2.5 MW, it would not be cost-effective to do so.
the pinch division and the rules for utility placement make a Designs which achieve Qh = 2.5 MW at ATmia = 10°C are
foolproof method of designing networks achieving MER. predicted to be economic optima. Pinch technology reconciles
rigorous thermodynamics (composite curves) with economics.
3.4 The Heat Exchanger Network Grid. An essential Comparing Qh = 2.5 MW with the original network (Fig. 1)
aspect of pinch technology is the simple way in which heat ex- and the network improved by exergy analysis (Fig. 3), we see
changer networks are represented. A simple grid (Linnhoff et that although exergy analysis reduced the value of Qh from
al., 1982) is used, as shown in Fig. 12, which illustrates the ex- 5.77 MW to 4.41 MW, it should be possible to reduce Qh still
ample problem with the original design. The hot streams run further to 2.5 MW for the same A:rmin.
from left to right at the top of the diagram and cold streams To start the design procedure, we establish from the com-
run "countercurrent" at the bottom. A heat interchange is posite curves that the pinch temperature corresponds to hot
shown as two circles connected by a vertical line with the streams at 180°C and cold streams at 170°C. The grid
70'
120V
0-34 m
155°
25°
1-32 -CW
PINCH
analysis has been performed on the new network and the
180°
results are shown in Fig. 15. The total oT0 has been reduced to
3.58 MW. This is an improvement of 0.86 MW over the design
H1
245°
o -0 155° ^ 1 4 0 ° / ^ , 80°.
!
-^°J 105;
achieved by traditional Second Law Analysis and an improve-
ment of 1.4 MW over the base case.
The total annual cost for the final design is M$ 1.70,
HH) representing a saving of 19 percent on the original base case.
The target figure was M$ 1.65 showing that the targeting pro-
120 70c cedure was accurate within 3 percent. This accuracy is by no
C1 means unusual (Linnhoff and Ahmad, 1989).
^15^202.5^
L2^i L6.5i
-o
[22.51
rirV-^
£5 57.5,
Q
LL5]
50'
C2 4 Pinch Technology and Exergy Analysis
4.1 Exergy Analysis Using the Problem Table. Pinch
170° technology was derived from exergy analysis principles. In-
!--HeaHoad(MW)
(a) Grid representation
deed, it represents an expression of those principles in a
powerful but simple stand-alone design tool for heat ex-
changer networks. The techniques make exergy analysis ap-
plicable for genuine synthesis, with the added advantage of ac-
M counting for equipment choice, network interactions, and
STEAM 245° 180°
economic trade-offs, as described in the Introduction.
In the Introduction it was stated that the use of pinch
215V7A 202.5°/ 170° 57.5° ^ 50°,
technology led to designs whereby the exergy losses were
"inevitable." While it is easy to say so conceptually, it is more
180° 140° difficult to quantify this statement. We will now attempt ex-
actly that.
Figure 16(a) shows the composite curves for the example
J^V 70'
m problem with ATmin = 0°C. The composite curves predict the
utility requirements. The key observation is that we do not
need a network design to predict oT0 as well. Rather, aT0 can
155°
be evaluated from the equation
AEx (sources and sinks) + oT0 = 0 (1)
UTILITY USAGE
4 3 % less than base case We know all stream data and, with the composite curves, we
105°
also know the utility heat flows ahead of design. The exergy
(to) Traditional representation change for a constant temperature heat source or sink is given
Fig. 14 Maximum energy recovery design (MER) by
AEx = A / / [ l - r o / 7 1 (2)
representation is divided at the pinch. The separate networks for nonconstant temperature, the expression
are started at the pinch, with "essential matches," and pro-
ceed away from it (Linnhoff and Hindmarsh, 1983). The dEx = dH[l-T0/T]
resulting designs are then combined to produce the overall can be integrated (Linnhoff, 1983), assuming constant heat
MER network shown in Fig. 14(a). For the traditionally mind- capacity, to give
ed, the design is also shown in a conventional sketch in Fig.
14(b). It is probably fair to say that this design would have AEx = A / 7 [ l - r o / A r L M ] (3)
been difficult to find by inspection. Equations (2) and (3) can now be applied to each source and
For comparison with the previous designs, an exergy sink or to the segments of the composite curves. This is
\=r
f /
Steam
245-215
Hot composite
-0.21
-1.22
Cold composite
—
Qc 215-180 -1.28
—
2.56
180-120 -7.06 3.53
120-105 -1.36 0.85
105-80 -0.46 1.15
80-70 _ 0.36
70-50 _ 0.42
|T CW - 0
E= -11.59 E=+8.87
E = -2.72
Qhmin
of To = 2-72
A Tmin =0°
"Qc H
(MW)
Fig. 17 "Energy Availability" diagram
T1 Qhmin AQ Qcmin
-AW
the pinch is that work is converted into additional load on the
cold utility (Linnhoff, 1986).
-PINCH 5.2 The Grand Composite Curve. So far the heat ex-
changer network has been considered in isolation, with the
assumption of fixed utilities. The principles of appropriate
placement have been derived assuming that the system above
or below the pinch can provide or accept all heat transfers
Qcmin from and to utilities necessary for the overall heat balance. An
H alternative diagram derived from Fig. 6 is helpful in consider-
Oin=Qhmin-(AQ-AW) ing these aspects in more detail. Figure 19 shows the derivation
Fig. 18 Appropriate placement of a heat engine
of the "grand composite curve." Here the hot and cold
streams are combined and the net heat surplus or deficit for
each interval is plotted on a temperature-enthalpy diagram.
The resulting graph shows that heat can be recovered from
dicative of aT0. We think more is gained by the correct scale section A-B to C-B and from section E-F to G-F. This leaves
for temperatures than is lost by an incorrect scale for exergy the external heating requirement D-C and cooling requirement
losses. D-E. These requirements are served through the hot utility
K-L and the cold utility M-N.
5 Applying Pinch Technology to Combined Heat and Similarly to the composite curves, the grand composite
Power curve shows "areas" of heat transfer K-L-A-B-D-K and D-F-
G-N-M-D which resemble exergy losses oT0. If we were to
5.1 Appropriate Placement. The application of pinch construct the grand composite curve for a vertical axis in terms
principles to the integration of heat engines and heat pumps of the Carnot factor, then the relationship would be exact, in
with heat exchanger networks has led to a small number of perfect analogy to Fig. 17. The important difference between
fundamental design principles which are expressed in rules for the composite curves and the grand composite curves is that
"appropriate placement." the latter allows us to distinguish readily between avoidable
Consider Fig. 18. A heat engine absorbs heat at high and inevitable exergy losses when we introduce different
temperatures and rejects it at lower ones. If an engine were utilities.
"placed" such that its source was above the pinch temperature Consider again Fig. 19. It is obvious that the hot utility K-L
and its sink below, it would effectively transfer heat across the is hotter than necessary. Consequently, most of the area K-L-
pinch. The effect on the overall energy balance would be the C-D-K resembles exergy losses which would be avoidable
same as if the engine were not integrated at all. Therefore, given a better (i.e., lower temperature) choice of hot utility. By
engines must be integrated either entirely above (Fig. 18) or contrast, the exergy losses resembled by the area A-B-C-A
entirely below the pinch. Cycles which require high input would not be affected, these losses are inevitable.
temperatures, such as gas turbines, will normally be integrated Similarly, cold utility could be used at higher temperatures
above the process pinch. Rankine cycles are conveniently avoiding most of the exergy losses resembled by area D-E-N-
placed either above or below, depending on the pinch M-D. The losses resembled by area E-F-G-E remain
temperature. The (First Law) efficiency of "appropriately inevitable.
placed" engines can be said to approach 100 percent (Town- In pinch technology, the grand composite curve is not
send and Linnhoff, 1983). drawn to assess inevitable exergy losses as such. Remember
A heat pump is an engine running in reverse and its place- that pinch technology, as a manifestation of Second Law
ment is reverse. A heat pump should be placed across the Analysis which works explicitly in terms of temperatures and
pinch so that it can pump heat from the process heat source heat loads, has made the very evaluation of exergy losses per
below the pinch to the sink above. A heat pump integrated se avoidable!
wholly above the pinch simply alters the energy balance by us- Instead, grand composite curves are used to select utilities.
ing work to replace an equal quantity of heat. This is usually Consider Fig. 20(a). In the top sketch, the grand composite
an uneconomical proposition. Below the pinch the situation is curve of a process with a high temperature external heating re-
worse. The net overall effect of a heat pump integrated below quirement is "matched" by a flue gas line. The construction
Tc
1/
iTTm-^
losses with the process in Fig. 20(b). Subject to network in-
teractions, choice of equipment, and economics (see Introduc-
tion), the areas shown shaded in Fig. 20 resemble inevitable
exergy losses. Not qualitatively, but precisely.
rrr^ 6 Discussion
It may be appropriate here to briefly emphasize that Pinch
Technology is not new. It has a proven track record. It has
emerged over the last 5-10 years from within chemical
engineering. It led to significant results even early on (Linn-
hoff and Vredeveld, 1984) and has since been proven in over
500 studies worldwide (University of Manchester, 1985-1986).
Some companies have publicly testified to its results with
average energy cost savings of 50 percent at capital cost reduc-
tion in grass roots designs (Linnhoff and Vredeveld, 1984;
Steinmetz and Chaney, 1985). These results were typically
achieved in projects relating to modern state-of-the-art designs
previously thought to be optimized. Results in older plants
have reached 90 percent plus energy cost savings and payback
times measured in days. Such impact is perhaps more easily
understood if the following simple facts are borne in mind: