See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/222385587

Saving energy in distillation towers by feed splitting

Article  in  Applied Thermal Engineering · June 2002

DOI: 10.1016/S1359-4311(02)00006-6

CITATIONS READS

78 744

2 authors:

Giorgio Soave Josep Anton Feliu

retired Inprocess Technology & Consulting Group
74 PUBLICATIONS   4,706 CITATIONS    21 PUBLICATIONS   190 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

growing my granddaughters View project

All content following this page was uploaded by Giorgio Soave on 31 July 2018.

The user has requested enhancement of the downloaded file.

 Appr,rnu Tlmnnnar,
EucrNnnRrNc

PERGAMON Applied Thermal Engineering22 (2002) 889-896

www. elsevier.com/locate/apthermen g

Saving energy in distillation towers by feed splitting

Giorgio Soave u, Josep A. Feliu b'*

" Via Europa 7, I-20097 S. Donato Milanese, Italy

b
Hyprotech Europe,5.L., Pg. de Gràc'ia 56, E-08007 Barcelona, Spain
Received I October 2001; accepted 24 November 2001

Abstract

It is a common procedure to save energy in industrial distillation towers by preheating the feed with heat
recovered from the bottom product. It will be demonstrated theoretically, and shown by simulation means,
that if, before entering the unit, the feed is split into two streams, and only one of them is preheated, further
savings of energy (up to 50o/o) can be achieved.
The use of a steady state process simulator like Hysys" is precious for a fast and reliable determination
of the optimum split ratio and feed tray.
A typical example of application is given and the use of Hysys" shown. The results for the optimrzation
of the split ratio are provided showing the economical impact of the proposed solution. @ 2002 Elsevier
Science Ltd. All rights reserved.

Keywords: Energy savings; Feed splitting; Process simulation

1. Introduction

The operating costs of chemical productive processes are highly influenced by the downstream
separation units. Fifty to eighty percent of the whole process operating cost can come from the
separation trains. Distillation is the most common unit operation in the chemical process in-
dustries. It has an extensive product history and is still reported as "the method of choice for
many separations, and the method against which other options must be compared" 11,2].
Humphrey [3] estimates for the United States consist of 40,000 distillation columns in operation,
which handle more than 90o/o of all separations for product recovery and purification. The capital

*
Corresponding author. Fax: +349 -32-154-256.
E-mail addresses: [email protected] (G. Soave), [email protected] (J.A. Feliu).

1359-43lll02l$ - see front matter @ 2002 Elsevier Science Ltd. All rights reserved.
PII:
S 1359-431 1(02)00006-6
890 G- soaue, J.A. Feliu I Applied Thermal Engineering 22 (2002) s89-sg6

investment for these distillation systems is reported to be at least 8 x l0e US$. Since distillation is
by far the dominant separation process, the reader is referred to the literature mentioned for a
detailed treatment of its advances [4].
The motivation for research in distillation originates from its serious drawbacks: distillation
columns tend to use huge amounts of energy because of the evaporation steps involved. Typically,
more than half of the process heat distributed to plant operations ends up in the reboilers of
distillation columns [2]. By this, high-level energy is fed at the base of the column and about the
same amount of energy is released at the top, unfortunately at a much lower temperature level.
The difference between the two Gibbs energies can be seen as the necessary energy investment to
reverse the mixing entropy and to separate the components of a given feed by a distillation
process. Often, the energy freed cannot be used for heat integration but is discharged to the at-
mosphere. The amount of energy transformed in the column is unimaginably large as the fol-
lowing approximation shows: Mix and coworkers [5] accounted distillation for about 3uh of the
total US energy consumption and recent reports show that this number has not been undergoing
fundamental changes to date. By combination with recent d,ataof the US DOE from l9g5 (United
States Department of Energy), distillation columns consume2.87 x 1018 I Q.8l million TJ) a year
which is equivalent to a continuous power consumption of 91 GW or to a 54 million tones of
crude oil. Consequently, the impact in absolute numbers of saving or recovering only l% of the
heat used by distillation columns would be tremendous. Despite this potential, distillation re-
search has repeatedly been proclaimed to be a dead area, and some universities have even con-
sidered stopping teaching the basics of distillation. However, there has been renewed interest
during the last years, especially since distillation columns have become a favorite subject in the
process systems engineering field, including the areas of process synthesis (process design), process
dynamics and control [6].
The separation of liquid mixtures by distillation depends on differences in volatility between the
components. The greater the relative volatility, the easier the separation. Al1 separation opera-
tions require energy input in the form of heat or work. In the conventional distiilàtio., op..uiion,
as typified in Fig. I, energy is added in the lorm of heat to the reboiler at the bottom of the
column, where the temperature is highest. Also, heat is removed from the condenser at the top of

Fig. 1. Common industrial column topology.

 G. Soure. J..-1. Feliu I .'lpplied Tlterntal Engineering 22 (2002) 889-896 891

the columrl. \\'here the temperature is lon'est. This frequently results in a large energy-input re-
quiremetrt and lou or erall thermodr namic eftrciencl'. u'hich was of little concern when energy
costs \\'ere lon. \\-ith the dramatic increases in energv costs. complex distillation operations that
oller hieher thermcrdrnamic eliciencr and loner energr-input requirements are continuously
being erplored [-]. Process units built pric-rr tcr l9-,1. the ]ear of the drastic rise in energy costs,
\\ ere generallr designed on a lou capital cost irì\ estn-ìent basis for marimum rates of return.
Ener*qr saving equipment \\'as included in the inr estment if it obviouslv improved the return on
investment. No extensive engineerin*q u-as directed at ener-s)' in the desi-en phase. In the current
period of high-energy costs. economics still dictates how much energy a ne\\' plant design can
conserve. But the incentive to expend more engineering time in the design phase to optimtze fhe
process with maximum energy conservation has increased. Likewise, there is the economic in-
centive to return to older operating plants and retrofit them with additional energy saving
equipment. Similarly, years ago, plant operators had been instructed to minimize off specification
production. They achieved this and reduced the amount of scrutiny and effort needed to operate
the unit by producing a purer product than necessary. This results in an increase in energy usage.
The basic equipment required for standard continuous industrial distillation is shown in the
topology of trig. 1. It contains a feed stream F, a top prorduct stream or distillate D and a bottom
stream B. Vapour flor,vs up the column and liquid counter-currently down the column. The va-
pour and liquid are brou-eht into contact on plates, or packing. Other equipment, a part of the
tra\: section. are a condenser (with its associated heat duty, p.), where part ol the condensate is
returned to the top of the column to provide liquid flow above the feed point (reflux), and a
reboiler (u'ith its associated heat duty, Q.), where part of the liquid from the base of the column is
r apourized an returned to provide vapour flow [8].
Several design alternatives can be proposed in order to decrease the energy consumption of
operating distillation towers. Some of them will require additional capital investment like adding
additional erchan-uers for heat recovery, column revisions, better insulation or column control. In
contrast to this changes, requiring small capital investment, there are other that will require more
erpenditure like vapour recompression or heat pump changes. Nevertheless, some operatin,e
procedure rer isions like reducing reflux ratio, lowering product specification, lowering pumping
Ct-rSIS. lt-rtr erins steam usage and using process heaters might require minimal capital investments
and migirt highh contribute to reduce column energy consumption. If the use of process heaters is
trptintized tas proposed in our contribution) the decrease on waste of energy can be further im-
pror ed.

2. Discussion
F
For the whole paper discussion, it will be assumed that flow rates and compositions of feed and
products are constant, as well as column pressure profile. In such w?y, temperatures and en-
thalpies of produ cls (hp and h) are constant, while feed temperature and feed enth alpy' (h e ) mar
Var\..
\ormally. the operating costs of industrial distillation towers mentioned above are usuallr
dependent on the reboiler heat duty. With a topology like the one described in the Introduction
clraprlgl. reflected into Fig. l, the reboiler heat duty Q, can be calculated from a heat bal:rnce
892 G. Soaue, J.A. Feliu I Applied Thermal Engineering 22 (2002) SS9-596

around the whole tower. The global heat balance keeps into account the condenser duty and the
feed heat content:
Q,: Q,-t B'hu + D'hD - F .ho (l)
From the balance above, it is evident that one direct way of decreasing reboiler duty, Q, ts by
increasing the enthalpy of the feed. To increa.se the feed enthalpy two options are available from
an operating point of view:

l. Preheat feed with external heat: This requires and additional consumption of heat which is
the
equal to the variation of the heat content of the feed so, no advantage at all. Additionally, the
associated increase in condenser heat duty, due to a warmer feed, would give an increase of the
total heat consumption.
2. Preheat îhe feed with the bottom product stream ( as shown in Fig. 2): In that case, the increase of
the feed heat content would suppose a real decrease of e,.

Unfortunately, there is a drawback in the proposed column topology modification because,

while hp increases, so does to a minor extent Q", due to the increase in the minimum reflux ratio,
hence of the real reflux ratio for a fixed number of trays.
In a new thermal balance around the column, both Q, and F . hr are increased. If we use Ap.,,
defined as the increase in condenser heat duty:
LQ,-
- Q'"- Q, (2)
and Qrr,, defined as the heat recovered by preheating the feed:
Qr"r:F'h'o-F'h, (3)
then, a new thermal balance, resulting from the combination of E,qs. (1)-(3) can finally be written
AS:

Q,,: Q|* B.hu + D.hp - F .h',

(4)
Q,,: Q,+LQ,-Q,,,

Condenser

'-r r-
Ill
,_ lL_ To
Rèboilér

Fig. 2. Common topology change when bottom product is much warmer than feed.
 G. Soaue, J.A. Feliu L-lpplietl Tltenttul Ertgineering 22 (2002) 889-596 893

B'B

Fig. 3. Proposed solution: splitting the Èed in su,-h : i\,ri i:13 ilt-rr-heated îiaction keeps the reflux ratio low and
the heated one recovers eners\' liom the reb..ile:.

From the anal.vsis of Eq. t-1) abtrre it can be highli-ehted that the recovered heat Q,.. prevails
over the increase of the condensùr dutr 1Q, and the result is that Q', 4 Q, so, there is an energy
consumption advantage in pre he atins the teed. \evertheless, due to the increase in condenser heat
dutl' then q, > Q, - O-.,. In trthe r \\ rrrds. the decrease in the reboiler duty is not equal to the
recovered heat but clnl1 a iraciit'rn oi it.
The problem that needs tt-r be >c-rir ed is how to take full advantage of the heat recovered from
the bottom product and [1-ru tt] marimize its effect. The proposal presented in this work is to split
the fèed betore it enters ihe heat erchanger in such a way of keeping a proper fraction of the feed
at its original temp-ralure and the rest being heated up with the warmer bottom product. The
objective t-rf the cold teed tiaction is to keep the minimum reflux ratio low so avoiding, or reducing
to a minintunl. the tncrease in the condenser duty by feeding it to a higher column tray. The hot
tiaction is led trr a lo\\'er tray and has the objective of recovering heat from the bottom product.
Fig. -ì shou s the topological changes proposed to the industrial distillation column.
There is a best value of the ratio of the feed flow rates (splitting ratio) that corresponds to the
minimum reboiler duty. A steady state simulator, like Hysys"^n [9,10] is the best tool to find such
an optimum split ratio. In general, the reflux ratio needed for a given separation is close to the
reflur ratio with a cold feed, and the recovered heat from the bottom product is close to that with
a completely preheated feed. The final result is a sharp decrease of the reboiler duty with respect to
both cases. In the limiting case, if the reflux ratio does not increase and the heat recovery is
complete:

o the new reboiler duty would be equal to the reboiler duty with the cold feed minus the recovered
'
?
heat from the wholly preheated feed, and this would be minimum:

Q', - Q,@old feed) - Q,",(wholly preheated feed) : minimum (5)

l. L Some additional considerations

Itr presence of two feeds there exist two values of the minimum reboiler duty: one is calculated
ir.-'nt the minimum reflux rate and condenser duty by a tower heat balance, the other one is
894 G. Soare, J.A. Feliu I Applied Therntal Engineerin7 22 (2002 i 8E9-s9ó

Qn-i'
From the hot feed conditions

From the cold feed conditions

Working point

Fig. 4. Two solutions for minimum reboiler duty.

calculated directly by a treatment symmetrical to that applied for R-1n. It has to be taken the
highest one of the two.
The one associated to the minimum reflux ratio prevails at higher cold feed flou rates. Ir is
constant if the hot feed florv rate is sufficient to recover all the heat from the bottom product.
otherwise (more common) it tends to decrease as the hot feed rate increases since that intproves
the recoverv of heat. From this point of view. it is better to have high hot feed flou' rates. On the
other side. the minimum reboiler dutl' calculated directly, associated to the conditions of the hot
feed, tends to increase u'hen the flori' rate is increased due to the fact that its temperature is de-
creased. From this point of vieu'. it is better to have low hot feed flow rates.
The best working point is located at the point i,'n'here the two lines cross each other" as seen in
Fig. 4. Often, the hot feed flow rate at the u'orkin-e point is not enough for a complete recover)' ol
heat from the bottom product and this would sli-qhtl.v decrease the actual gain.
Hot and feed streams have to be introduced at a different column tray. As wrong feed tray
location would increase the reboiler duty, optimum feed trays for the cold and hot streams are to
be found. The results obtained by running a process simulator are ideal to find such an optimum.

3. Case study

The results of a case study follow below to show the validity of the alternative tower topology.
It is proposed to separate a common industrial propane-benzene mixture. A usual feed to such a
distillation tower is 80% benzene purity at 20 oC and 15 bar. The separation has to achieve a
99.9% benzene recovery at the bottom product andgg.gohpropane purity at the distillate product.
To be consistent in the case study when results will be compared, all three column topolo-uies
shown later are run with the same tray section (30 theoretical stages) and at the same operating
pressure. Just the feed location varies from case to case. To simulate a classical industrial situa-
tion, the same heat exchanger confi-euration (that is characterized br, the same L- . .1 coefficient)
has been used in both preheated simulations to represent a single heat erchanser availability in the
plant.
The simulated results with a non-preheated column u'ill be considered as the base case to
compare to the alternative proposals of preheating the feed.
Fig. 5 shows graphically the results obtained lrom several runs of the process simulator. The
common solution of preheating the whole of the feed that enters the column sa\.es about 1600 MJ/
h in the reboiler as expected but, on the other side. it increases the condenser duty and,, conse-
quently, the dimensions of the upper section of the tower.
 G. Soaue, J.A. Feliu I Applied Tlrcnttal Engineerirtg 22 (2002) 859-596 895

I +cóndàhsr I
| +n.u"it..
Feed split optimization | -----coxPrenexedcara
I
I
I R€b Pr.-hèatea carc I
| - - - Conccnscr gasc case I
I netoitcr sarc casc I

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Fraction of cold feed

Ftg. -t. Results of the optimization procedure. For the case being studied, optimum is found at a 42'% of feed being left
ctrld.

Table 1

Results of the comparative study (Fig. 5)

Case studv name Non-preheated flow (%) Condenser duty (MJ/h) Reboiler duty (MJ/h)
Base case {Fig. 1) 100 312.6 2611.0
Fuìl preheating (Fig. 2) 0 892.2 1042.0
Partial preheating (optimum) (Fig. 3) 42 3t 3.5 664.2

81' b1'passin,s half of the feed through the heat exchanger, keeping it cold, and feeding it to an
upper section of the column it is possible to decrease the energy demand of the condenser down to
\.alues \\.ithout preheating, and, at the same time, to decrease further the reboiler duty. A detailed
studl can be run using Hysys"' process simulator to prove that an optimum split ratio exists. For
the case being studied. 42o/n of the feed should be left cold. Working with this optimum split rario.
it is possible to save 2000 MJ/h of reboiler duty with respect to the base case and about 400
additional \IJlh u'ith respect to the fully preheated feed (Table 1).
It is possible to optimize further the energy consumption by looking for the optimum location
of the leed trar s. This fine-tuning of the optimum solution can just add an additional 5'14, re-
duction and it is not shou n in this paper. The results show the values obtained when feeding the
1,-"'p column section at tral l0 and the belou.section at tray 22.The base case and the full pre-
heated leed examples were run by feeding the column af tray 12.

-1. Conclusions

\\'ith respect to the common solution of preheating the whole feed with the bottom product.
heating only part of the feed:
896 G. Soat'e, J.^-1. Feliu Llpplied Tlterrttol Ertgineerirt, -'l r )()02 t 889-896

1. Decreases the reboiler dutl'.

2. Decreases the condenser dutl (keepin-u the same as the base case u.ithout preheating).
3. Increases thus the capacity of the tou'er (the vapour and liquid trallcs alon_s the touer are de-
creased), allowing to revamp overloaded towers.
4. Requires negligible investment for the introduction ol a bypass to the preheater and the
changes of the feed trays.
5. Requires some runs of Hysys" steady state simulator to find the best split ratio (and. option-
ally, the best feed trays location).
6. The described concept can be also applied to the reduction of condenser duty of cold towers
(i.e. demethanizers) by pre-cooling part of the feed by the distillate stream.

References

[l] H. Kister, Column malfunctions becoming extinct or will they persist in the 21st century? in: R. Darton (Ed.),
Distillation and Absorption '97, Vol. one of IChemE Symposium Series No. 142, IChemE, Rugby, lJK, l99l.
[2] J. Kunesh, H. Kister, M. Lockett, J. Fair. Distillation: still towering over other options, Chem. Eng. Progr. 9l (10)
(19e5) 43-s4.
[3] J. Humphrey, Separation processes: play'ing a critical role. Chem.Eng.Progr. gl (10) (1995) 3141.
[a] J. Tojo, Destilación Avanzada. In'n'ited speech to "Primer Encuentro Universitario de Aplicaciones Hyprotech".
Valencia, July 2001.
[5] T. Mix, J. Dweck, M. Weinber-s. R. Armstrong, Energy conservation in distillation, Chem. Eng. Progr. l4 (1)
(tet8) 4e.
t6] S. Skogestad, Dynamics and control of distillation columns: a tutorial introduction, in: R. Darton (Ed ).
Distillation and Absorption '97. Vol. one of IChemE Symposium Series No. 142, IChemE, Rugby, UK, 1997.
[7] J.D. Seader, Continuous Distillation Operations, in: R.H. Perry, D.W. Green (Eds.), Chemical Engineers'
Handbook, sixth ed., McGrau. Hill. Neu., York, 1984.
l8lH.Z. Kister, Distillation Design. McGrarv Hill. Nerv York. 1992.
[9] Hysys v.2.4.2 Ljser's Guide" AEA Technolo-sr'-H1'protech. Calgar1,. 2001.
[10] Hysys v.2.4.2 Steady Sate Modeling. AEA Technologl'-H1'protech. Calgarl. 2001.

View publication stats