You are on page 1of 7

0263-8762/97/$10.00+0.

00

IChem E

Institution of Chemical Engineers

OVERALL SITE OPTIMISATION BY PINCH


TECHNOLOGY
By *B. LINNHOFF and **A. R. EASTWOOD (MEMBER)
*Department of Chemical Engineering, University of Manchester Institute of Science and Technology
** LinnhoffJ March Ltd, Manchester
The importance of integration in process plant design has long been recognised. This is evideoced for example, by the existence
of complex heat and power networks in modern prodoction sites. These networks represent site-wide integration. They usually
allow process heating duties in several processes to be effected at low cost by using turbine pass-out steam after obtaining
a credit for power generated in a central turbine. When looking for energy saving in retrofit projects, an installed heat and
power network of this type usually introdoces a counter-prodoctive element. A redoction in steam heating duties leads to a
reduction in turbine pass-out steam which, in turn, leads to a reduction in power generation. This makes optimisation of the
process alone and optimisation of the process in the site context two rather different tasks. More complications are given
through other more general reflections across a site. A modem processing plant soch as ethylene, for instaoce, is usually linked
into the overall site and interacts with other plants through several utilities, not just back pressure steam. This makes it
Detessary to not only understand the fundamental process demands but also to analyse these demands in the site context.
Process licensors will usually prefer to consider standard designs for individual processes. The best project, on the other hand,
will almost invariably be site dependent. This artide highlights the distiDttion between the application of process integration
techniques-in particular pioch technology-within an individual process and their wider application in the overall site context.
The "counter prodoctive" aspect of heat savings in the context of isntalled heat and power networks is given attention. In
addition to energy, the artide discusses yields, flexibility and capital cost savings.

INTRODUCTION
Over the last four years Pinch Technology has been
widely used for the integration of process plants. Virtually every sector of the process industries has benefitted.
This experience clearly teaches that integration must be
undertaken in a total site context.
This paper considers how process plants interface with
utility systems. It describes the development of an optimum retrofit scheme in the context of an overall site.
Individual process improvements are not taken in isolation. Rather, their total impact is considered in a
genuinely site wide context.
The paper takes as its example an ethylene process.
This is because ethylene processes are not only complex
but interact with the site utility system in a complicated
way. The process employs a large number of separate
utilities. These utilities are closely coupled with the
overall factory site.
In terms of ethylene technology the paper adds to
previous advances 1,2.
Figure 1 is a simplified schematic of a "typical"
ethylene plane showing the following, not uncommon,
process features:
Pyrolysis furnace and quench steam raising
Principal mechanical drives
-raw gas compressor
-propylene refrigeration set, Fl
--ethylene refrigeration set, F2
-boiler feed water pump
"Cold box" with steam and fridge demand (several
levels)
Other process steam demands (several levels)

Import steam from central power station (several


levels)
Figure 2 is an overall site plan in which the ethylene
plant is shown in relation to other plants in the factory
as well as to the central boiler house and power station.
Again, this type of situation is typical of many sites in
a wide cross-section of industries. Generating pressure
steam at 120 bar (VHP) is raised in the main boilers and
all of this steam is fed to the turbo-generators in the
power station. Steam is extracted at three levels (HP, MP
and LP) and distributed around the factory; there is no
condensing of steam in the power station. Site electricity
demand is satisfied by import from the grid in addition
to self-generation.
Figure 3 represents a more detailed look at the
ethylene plant steam system. We can see that condensing
steam turbines are employed on both the raw gas
compressor and the propylene refrigeration compressor
drives. Total steam for heating duties around the site
amounts to 270 tons/hour distributed as shown. The
central boiler house supplies 210 t/h of steam to the
turbo generators to raise 16MW of electrical power.
Quench steam raised in the ethylene plant amounts to
150 t/h.
Taking energy savings as the first objective in our
retrofit, we must try to:
Reduce hot utility requirements in all plants where
cost-effective
Increase power generation
Achieving either or both of these objectives should
result in reduced operating costs. There is, however, a

Institution of Chemical Engineers


S138

JUBILEE SUPPLEMENT-Trans IChemE, Vol 75, December 1997

S139

OVERALL SITE OPTIMISATION BY PINCH TECHNOLOGY

HP STEAM
:MP STEAM
LP STEAM

n II

FRIDGE SETS
BFW
PUMP

QUENCH
STEAM

RAW GAS
COMPRESSOR

ETHYLENE

t-o'--+-<I"PRODUCT
I
I
I

FUEL

IL

...J

Figure I. Ethylene plant schematic.

basic conflict between these objectives in that a reduction


in steam heating duty will result in a reduction in turbine
extraction steam and consequently a reduction in backpressure power generation.
PRELIMINARY ASSESSMENT
First we set out the economic criteria for possible
retrofit projects. This often includes ranking of the
various means of obtaining shaftwork.
The marginal cost of VHP steam generation in our
example, which is based on European conditions, is
7.50 per ton and the cost of imported electricity is 28
per MWh. The condensing turbines in the ethylene plant
can generate 220 kWH per ton of VHP steam which
means that the marginal cost of "condensing shaftwork"
is 34/MWh. In producing shaftwork, then, we have the
following ranking in terms of cost:

(This is representative of many European sites. In the


USA, the ranking between options (2) and (3) is becoming less clear-cut. None of the methodology to be
discussed in the present article depends on this ranking.

t1PORTED

'n"'"

FUEL

16HW

WENCH STEAM

(I) Back pressure steam turbine (least expensive)

HP USERS
(TOTAL SITE I

(2) Imported electricity from the grid


(3) Condensing steam turbine (most expensive)

II ELECTRICITY
t1PORTED
{7
UEL

' " ELECTRICAL


POWER

M.P I--_ _--=-_---JL...-..--_---JL.-.;-_-:-....I..l

lP I----......:.-----"-=---L---'--....I..l

MP USERS
!TOTAL SITE I

LP USERS
llOTAl SITE)
270 t/h TOTAl

CONDENSATE.

Figure 2. Overall site plan.

Chern Eng Res Des, Vol. 65, September 1987

Figure 3. Ethylene steam system.

JUBILEE SUPPLEMENT-Trans IChemE, Vol 75, December 1997

S140

LINNHOFF and EASTWOOD

After studying
each process carefully....

-12%

-5%

-10%

....and each process


is made to achieve target
Figure 4. Analysis of separate plants.

Rather, it is part of the approach to establish the ranking


up front, whatever it may be.)
Given the case above, our objectives are to reduce or
eliminate the use of condensing turbines and introduce
electric motors and/or back pressure turbines where
possible.
Let us now assume that careful studies have been
performed on the separate process units to minimise
energy consumption by process improvements including
better heat recovery, improved housekeeping and whatever else may be appropriate. In pinch technology terms
let us assume that all projects are known to bring the
individual plants "on target" (Figure 4). The result of
such a thorough, and site-wide, campaign of process-byprocess improvements may be that the total steam
demand for process heating could be reduced from 270
t/h to 240 t/h, a saving of 14% at the MP level and 11 %
at the LP level (Figure 5). However, we wish to retain
as much as possible of the existing steam system, reducing the condensing steam component in preference to
back-pressure, in line with our economic criteria.
In the revised scheme shown in Figure 5 we have
reduced the condensing steam to its minimum value in
the raw gas compressor drive and introduced an electric
helper motor to make up the deficit in shaftwork. Fuel
to the boiler house is reduced by 24% (assuming constant boiler house efficiency) but imported electricity is
increased by 7MW to cater for the helper motor (4 MW)
and the reduction in output of the turbo-generator (3
MW).
The reduction in main boiler house steam raising is
worth 375 per hour and the increase in import electricity amounts to 197 per hour. The net saving with
this combination of projects is therefore 178 Per hour
or 1.44 million p.a. The associated capital cost of 2.2
million means that the overall project payback period is
approximately 18 months. This could be considered
quite typical for overall benefits to be achieved from a
process-by-process campaign of improvements.

FURTHER IMPROVEMENT
The system shown in Figure 5, although offering
reasonable savings at good payback, still includes
significant amounts of shaftwork derived from condensing steam. We would ideally like to replace the propylene
compressor (FI) turbine with an electric motor. We
cannot, however, treat this turbine in isolation as closer
inspection of the steam system reveals. Figure 6 shows
the steam from the ethylene compressor (F2) turbine is
fixed by the duty of the compressor at 80 t/h. Process
requirements for LP steam are given as 80 t/h and the
import of LP steam from the power station turbine is at
a minimum value of 20 t/h determined by the characteristics of the power station turbo-alternator. To replace
F I condensing turbine by an electric motor would mean
that 20 t/h of LP steam would be surplus to demand
would have to be vented if not being condensed in the
turbine. Further savings would therefore appear to be
strictly limited because we have run up against this
operability constraint.
At this point we need a thermodynamic tool which
will address the complex interaction of heat and power
inherent in the overall plant complex in a rigorous and
systematic way.

USE OF THE GRAND COMPOSITE CURVE


The ideal tool with which to analyse the process/utility
interface and interactions of heat and power networks is
the grand composite curve3 which is now well accepted.

,, , ,
~

<'RJEL "
"-24"--

',-'

QUENCH
STEAM

35
HP I - - -........~~r---- ......~-:--,.....-.........

M.P USERS
!'I.P

1-----;--......-r------''--:---;-...I.o1

LP

1 - - - - - ' - - - -........-7---L--..L-...I.o1

CONOENSATE

~,
~~~/

LP USERS

240 t/h TOTAL

Figure 5. Steam system after implementation of heat recovery projects.

Chern Eng Res Des, Vol. 65, September 1987

JUBILEE SUPPLEMENT-Trans IChemE, Vol 75, December 1997

S141

OVERALL SITE OPTIMISATION BY PINCH TECHNOLOGY


1600

F/01
POWER
STATION

1400

1200

1000
~-,,--,--aOtfh

TEMP.

20 fh

(OC)

L.P. HEADER

//

800

Cannot replace F1 by IMPORT

//

600

/
/

without venting LP steam.

/
/

400

/
/

Figure 6. L.P. Steam constraints.

HP
PIP

Figure 7 illustrates the grand composite curve for the


overall site above the process "pinch" which occurs at
about 80C in this example. This grand composite curve
is shown by the broken line upon which we can overlay
the primary utility for the site, namely flue gas from the
combustion of hydrocarbon fuels. The figure shows the
existing flue gas line which combines the flue gases of the
central boiler house and ,he ethylene pyrolysis furnace.
The diagram further shows that at high process temperatures flue gas is required for process heating (i.e. in
the pyrolysis furnace), whereas at lower temperatures
VHP steam raising can be employed, with turbine passout steam being used at successively lower temperatures
1600

1400

1200

1000
TEMP.

rOc)

800
/
/

600
/

/
/

400

VHP STEAM RAISING


F===9==~;;';";;~~HP5====-t'
LP

__""-SITE GRAND
__
COMPOSITE

I---,Mc.;;.P_-::-::t.. / SHAFT.. ------.


WORK
_r'"
44MW

<~' PROcESS PINCH

100

200

300

HEAT IMW)

Figure 7. Existing heat and power profile against grand composite


curve.

Chem Eng Res Des, Vol. 65, September 1987

'

~I
EXCESS 0

HEAT

-..

,..-----"

__ /

,/

_~J--'

100

200

300

HEAT IHWI

Figure 8. More accurately.

for process heating. Total shaftwork of 44 MW produced by the steam system is also indicated.
This diagram is thermodynamically incorrect since it
does not show the correct interaction of process and
utilities but it does give a clear picture of the power
cycles. The derivation of a more accurate utility line is
not discussed in this article but is reproduced as Figure
8. We can now see more clearly just what is happening
at the existing process/utility interface.
Now Figure 8 indicates sub-optimality in two ways:
1. The excessive use of heat caused by the lack of process
intergration (i.e. the overshoot of the utility profile on
the grand composite curve). This is equivalent (in
terms of heat duty) to the 30 t/h of steam which we
saw earlier could be saved by improved process heat
recovery within the process plants plus the unrecovered heat in the flue gases below the existing
stack temperatures.
2. The poor fit of steam levels against the grand composite curve. The curve indicates an opportunity to
use more LP steam in preference of HP and MP
steam. The process actually requires 20 t/h HP, 100
t/h MP and 120 t/h LP-the same total of 240 t/h but
in a different distribution.
Now, the revised steam distribution will allow us to
produce more shaftwork from the same total amount of
steam and the increased LP demand will avoid the
problem of minimum flow in the back end of the power
station turbine. This will give us more flexibility in our
revised steam system design. Figure 9 illustrates such a
better matched process/utility interface. Note that the
flue gas line is now slightly steeper than before because
of the reduced steam raising at the boiler house.

JUBILEE SUPPLEMENT-Trans IChemE, Vol 75, December 1997

S142

LINNHOFF and EASTWOOD


1600

COOLING

WATER

~--r--20.C

11.00

SUB-AMBIENT
/

FOR

ETHYLENE PLANT

o'C

1200

GRAND COMPOSITE

F1

mo
TEI'1PI"C)

,,;oJ'"

800

600

400

200

LP
/

MP

H ,,/
.1

,/
,,

/,/'

,/

F2

J-'-..J

.....\
100

200

HfAT IMWI

Figure 9. Improved matching to process.

SUB-AMBIENT CONSIDERAnONS
The sub-ambient region of an ethylene plant is extremely complex involving perhaps six levels of refrigeration in addition to the primary cold utility of cooling
water and/or air cooling. The six levels of refrigeration
are achieved by using two separate but interlinked
systems.
Fl, propylene refrigeration system
(10 to -40C)
F2, ethylene refrigeration system
( - 60 to - 100C)
Each system may extract heat from the process at
three levels and, in addition, there will be a temperature
level at which the propylene system extracts waste heat
from the ethylene system. These interactions are illustrated in Figure lOin which the refrigeration systems are
placed against the sub-ambient grand composite curve of
the ethylene process. The "pinch" at -40C is, technically, a cooling water pinch since it is formed after
maximising the use of cooling water in the pinch analysis
(cooling water being the least expensive cold utility). We
shall see that it is the interaction of the two refrigeration
systems which allows increased flexibility in design.
Figure 11 shows the existing refrigeration levels on the
grand composite curve (chilling duties represented by
solid lines, waste heat rejection by broken lines). The
existing compressor duties are 4 MW and 7 MW for
propylene and ethylene respectively. Again we see suboptimality caused by:
I. Larger than necessary chilling duties shown by the
overshoot of each refrigeration level on the grand
composite. These can be reduced by improved process
heat (or cold) recovery.

Figure 10. Interaction of fridge systems.

2. Temperature levels not properly matched with process requirement.


By optimising heat recovery we can save about 15% on
compressor shaftwork, producing a 1.5 MW saving in
total shaftwork.
~------------~-Fl HEAT

REJECTION

+20

I
I
Fl FRIDGE
LEVELS

-20

..~--:..::=.=j=./
HEAT
-- ----1---w' --==--------------'=--- F2
REJECTION

F2 FRIDGE
LEVELS _

_80

-100

Fl LOAD

F2 LOAD

EXISTING

4-0 MW

7-0 MW

11-0
100

POSSIBILITY A

3-5 MW

6-5 MW

POSSIBILITY B

4-0 MW

6-0 MW

10-0

POSSIBILITY C

3-0 MW

7-0 MW

10-0

Figure II. Sub-ambient grand composite curve for ethylene.

Chern Eng Res Des, Vol. 65, September 1987

JUBILEE SUPPLEMENT-Trans IChemE, Vol 75, December 1997

Sl43

OVERALL SITE OPTIMISATION BY PINCH TECHNOLOGY


IMPORT

I IPOWER
V ~MW

FUEL

HP I-----L......:....---,,-------L---,r---..L...j

MP I - - -

--'-_~I

~o

80

LP ~--.....:....----------lL...--.L.-..L...t

....

,"' "

The raw gas compressor turbine arrangement is unchanged since we are already at minimum condensing
flow rate but the remaining 35 tlh of condensing steam
from the propylene refrigeration system are eliminated.
VHP steam raising is reduced by a further 35 tlh from
the level in Figure 5 (85 tlh reduction on the existing
steam raising) but import electricity is increased by 4
MW over our earlier design to cater for the new motor
and the slight reduction in turbo alternator output (an
overall increase in imported electricity of 11 MW over
the existing operation). Total project savings are therefore 330 per hour or 2.6 million annually. With an
estimated capital expenditure of 3.3 million, project
payback is in the order of 15 months. This compares
with the apparent maximum savings identified earlier of
1.44 million at an 18 month payback (Figure 13).
In terms of power generation, compare the two
schemes in Figure 5 and Figure 12 to the existing design:

~'"

t Ih TOTAL
(UNC HANGED I

2~0

Figure 12. Improved steam system.

On the other hand, increased product yield is often


more important to us than energy saving and a trade-off
is often possible between the two. In this example, we
can increase yield by 2%, the energy penalty in refrigeration being a saving of only 10% rather than the 15%
calculated above. If this 10% saving (1.0 MW) is distributed uniformly between the two refrigeration systems,
we arrive at Possibility A in Figure II with the shaftwork
of each compressor reduced by 10%.
By also induding the FI/F2 interaction in our analysis
we can obtain the same overall saving of 1 MW but
arrange the total saving to be on either the ethylene
machine alone (Possibility B) or on the propylene machine alone (Possibility C).

FINAL DESIGN
Figure 12 shows our improved design (Possibility C in
Figure 11) following the overall site analysis and the
sub-ambient consideration described above. These factors have allowed us to introduce two elements of
flexibility which avoid the earlier constraints in our
retrofit projects.
The increased LP steam demand has increased the
potential for power generation (per ton of HP steam)
and allowed us to eliminate the condensing turbine on
the propylene compressor without penalty. The
flexibility within the two refrigeration systems has further allowed us to minimise the capital cost of the new
electric motor introduced on the propylene compressor.
Chern Eng Res Des, Vol. 65, September 1987

Difference from
existing design
Figure 12
Figure 5
-3 MW
0
-4 MW
-12 MW

Back pressure generation


Condensing generation
Saving in refrigeration
systems
Import

+7MW

-IMW
+1] MW

Alternatively, we have increased the back-pressure


power generation per ton of VHP steam at the boiler
house. Total back-pressure power generation is now 24
MW from 125 tlh at the boiler house compared with 24
MW from 210 tlh in the existing design and 21 MW from
160 tlh in the earlier project which involved in-plant
optimisation only.
Also, we must not overlook the fact that we have
improved product yield by 2%, a feature which we have
not included in our economic analysis but which is likely
to be of considerable value.

5
SAVINGS
$10' p..

//L"'/
/'

'" '"
//
''''
"';'"
/

// /

,v

'"

,.

PROJECT A BY
INOIVIDUAL PLANT
OPTIMISATION

18 MONTH
AVERAGE
PAYBACK
5
INVESTMENT
$ 10'

PROJECT B OFFERS LARGER SAVINGS


AT A SHORTER PAYBACK!

Figure 13. Comparative economics.

S144

JUBILEE SUPPLEMENT-Trans IChemE, Vol 75, December 1997


LINNHOFF and EASTWOOD

A COMMENT ABOUT EXPANSION PROJECTS


This case study relates to the retrofit design of an
existing chemical plant complex. In this example, the
total site approach has reduced boiler house steam
raising from 210 to 125 t/h whereas the plant-by-plant
approach only reduced steam raising from 210 t/h to 160
t/h. This could have a tremendous impact on factory
expansion plans in a case where the existing boilers were
already at full capacity and further increase in steam
demand would necessitate the purchase of additional
boiler plant.
The significant improvement in energy and yield described earlier could well be accompanied by a substantial capital cost saving if factory expansion plans would
have involved the purchase of additional boiler plant.

CONCLUSIONS
Retrofit projects of individual process plants aimed at
improved yield, energy savings, expansion, etc. will
increase profits in an overall complex but there could be
better profit improvement still if interactions are properly understood. For example, the reduced steam demand inherent in an energy saving project can be
counter-productive since it often leads to reduced shaft-

Improve profitability (reduced utility


consumption, increased yield, flexibility)
on each process by any legitimate means.

I
Determine optimum levels of site
services from grand composite.

I
Exploit interaction between different
systems (e. g. propylene and ethylene
refrigeration I.

I
Model cogeneration and site service
systems into process consumers (not
based on design detail but based on
target curves l.

I
Design of projects (treating process and
service systems simultaneously l.

Figure 14.

Table I. Experience of total sites

Cost Savings

Industry
Petrochemicals
Inorganics
Chemicals
Pharmaceuticals
Resins
Pigments
Steelworks
Foodstuffs

30% of imported fuel


Improved co-generation
30% of total energy
30% of total energy
20-40% of total energy
25% + debottlenecking
15% total energy
50% increase in power
generation
25% of total energy

Project payback
12-24 months
2-3 years
9-16 months
15 months
2-2! years
up to 3 years
15 months
2 years
2 years

(Courtesy: Linnhoff March)

work availability. Also, process steam savings may add


unwelcome constraints to central steam system improvement.
To obtain best overall site strategies we recommend
the use of grand composite curves of the whole site in
order to determine the optimum levels of utility supply.
This applies to both hot and cold utilities. Multi-level
systems (such as ethylene refrigeration) often introduce
an element of flexibility which we can exploit to good
effect.
Figure 14 is a logic diagram of the design philosophy.
Such a systematic approach is vital to optimum
retrofit design. There is no substitute known to us for the
grand composite curve. It combines all the elements of
process heating and cooling, the process/utility interface
construction and allows us to assess the opportunities
for co-generation within the overall factory. At the same
time we can systematically address the trade-offs inherent in the system between capital and energy, between
yield and energy, and between flexibility and energy. No
other approach allows us to investigate all these tradeoffs both separately and in combination. Table 1 contains just a few of the many examples of the total site
integration studies that have been carried out over the
past few years and are known to us. The table demonstrates the scope revealed by such studies.

REFERENCES
I. Linnhoff B. and Vredeveld D. R., 1984, Chern Eng Prog, July.
2. Linnhoff B. and Witherell W. D., 1986, Oil & Gas J, April 7.
3. Linnhoff B. et al. 1982, User Guide on Process Integration for the
Efficient Use of Energy. (I, Chern E)

ADDRESS
Correspondence on this paper should be addressed to Professor B.
Linnhoff, Department of Chemical Engineering, UMIST, PO Box 88,
Manchester M60 IQD.
The manuscript was received 26 June 1987 and accepted for publication.

Chern Eng Res Des, Vol. 65, September 1987

You might also like