Anda di halaman 1dari 9

Chemical Engineering and Processing 49 (2010) 367–375

Contents lists available at ScienceDirect

Chemical Engineering and Processing:

Process Intensification
journal homepage:

Purification process design in the production of styrene monomer

Ismael Díaz a,b,∗ , Paul Langston b , Gabriel Ovejero a , María Dolores Romero a , Eduardo Díez a
Grupo de Catálisis y Procesos de Separación (CyPS), Departamento de Ingeniería Química, Facultad de CC. Químicas, Universidad Complutense de Madrid Avda.
Complutense s/n, 28040 Madrid, Spain
Chemical and Environmental Engineering, The University of Nottingham, University Park, Nottingham NG7 2RD, UK

a r t i c l e i n f o a b s t r a c t

Article history: An economic study of two alternatives for styrene monomer purification based on equilibrium staged
Received 11 December 2009 processes is presented. For this purpose, the solubility of t-butylcatechol (TBC) in styrene and the
Received in revised form 22 February 2010 styrene–water liquid–liquid equilibrium were experimentally determined. In terms of economics, a com-
Accepted 6 March 2010
bined process extraction–distillation shows worse results than a conventional distillation process, for the
Available online 12 March 2010
styrene purification, but this process could be suitable in a high energy cost scenario. Measured activity
coefficients NRTL model and UNIFAC predictive model were used for the process optimization.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction But organometallic compounds are very sensitive to the pres-

ence of polar compounds which can react with them. For example,
Nowadays, styrene–butadiene derived rubbers (SBR) are widely water reacts with n-butyllithium to form lithium hydroxide and
used in tyres, toys and adhesives production. Because the synthesis butane. In this way, the initiator decomposes, so it is not longer
of these products has risen since the 1960s, the demand for natu- effective for starting polymerization. Therefore, it is necessary to
ral rubber derived products has been replaced. The development reduce impurities to less than 1 ppm for both monomers.
of synthetic rubber processing forced manufacturers to develop Styrene is usually passed over an alumina bed, which adsorbs
new purification processes for both monomers (styrene and buta- traces of water and t-butylcatechol (TBC) [2,3]. This last compound
diene). Purification is needed because, although both monomers is added to avoid the thermal autopolymerization (via radical
are produced with a purity above 99%, traces of polar compounds mechanism) of the styrene during transport and storage, because
can deactivate the initiator of the polymerization reaction. it is an effective free radical scavenger in presence of oxygen. But
In the SBR production, both monomers usually react through the problem is that the lifespan of alumina beds is very short, so
an anionic mechanism, called “anionic” or “living” polymerization, they must be replaced and the waste must be treated; this makes
because the ends of the polymer chains are charged, so they can the process quite expensive [4].
react with other compounds. But the anionic intermediates, which Due to the difference in normal boiling points of the three
propagate the polymerization, need an initiator. Organometallic compounds involved (water, styrene and TBC), the purification of
compounds are usually employed for this purpose because they can styrene can be performed through a distillation process. As styrene
react easily with the monomers to form the intermediate carban- thermal polymerization rate becomes important at temperatures
ions. One of the most common initiators is n-butyllithium, which higher than 50 ◦ C [5,6], the distillation process must be carried out
decomposes to form Li+ and CH3 CH2 CH2 CH2 − (Fig. 2). These last at reduced pressure [7]. The main problem in distillation is the very
carbanions react with olefins to form various polymerization inter- high energy consumption, because TBC is heavier than styrene so
mediates, as can be seen in Fig. 1 [1]. a large amount of the monomer purified must be evaporated and
then condensed. For this reason, this paper proposes a new hybrid
purification system combining distillation and liquid–liquid extrac-
tion in such a way that evaporation of the whole styrene stream
is not necessary. To accurately design or simulate both systems, a
∗ Corresponding author at: Grupo de Catálisis y Procesos de Separación (CyPS),
reliable thermodynamic model is needed. Therefore, experimen-
Departamento de Ingeniería Química, Facultad de C.C. Químicas, Universidad Com-
tal phase equilibrium data of the binary systems styrene–TBC and
plutense de Madrid, Avda. Complutense s/n, 28040 Madrid, Spain.
Tel.: +34 91394411. styrene–water have been obtained and fitted to the NRTL [8] activ-
E-mail address: (I. Díaz). ity coefficient model. These data have been compared with those

0255-2701/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
368 I. Díaz et al. / Chemical Engineering and Processing 49 (2010) 367–375

Fig. 1. Scheme of the anionic polymerization.

predicted by the UNIFAC [9] group contribution model. Economic and melting point (Tfusion ), must be known. In the case of low
feasibility studies have been carried out for both purification sys- temperatures, the difference between ideal and real equilibrium
tems, showing noticeable differences. The calculations of capital composition of the mixture is about 85%.
investments and operational costs were performed with the help of The experimental results presented herein have been also com-
ASPEN ICARUS® software and information in the literature [10–13]. pared with the prediction of UNIFAC model. In terms of activity
coefficients of TBC in solution, the UNIFAC group contribution
2. Experimental measurements of equilibria method predictions, as well as NRTL equation fitting, are shown in
Fig. 5. It can be seen that the lower the temperature, the higher the
2.1. Materials deviation from UNIFAC model. At −23 ◦ C the difference between
the experimental and the UNIFAC predicted value is about 80%.
Styrene was used in analytical grade and was purchased from
Sigma–Aldrich. TBC was obtained from Fluka in HPLC grade. Milli-Q
water was used from a Millipore on site generator.
2.3.2. Styrene–water system
Fig. 6 shows the experimental equilibrium curve of the biphasic
2.2. Procedure
region, along with the values obtained from NTRL fitting, those pre-
dicted by UNIFAC method, and the literature values [15]. It can be
To study the styrene–TBC system, a magnetically stirred tank
seen that, again, the NRTL model adequately fits the experimental
(25 mL), placed in a thermostatically controlled bath was used
data. UNIFAC predictions are quite good for the aqueous phase, but
(Fig. 3). Initially the tank was filled with pure styrene and small
not for the organic phase. The experimental results are in reason-
amounts of TBC were subsequently added, until a solid phase began
able agreement with literature data [15]. The differences between
to appear. Once this second phase had appeared, the system was
them are assumed to be related to the analytical apparatus used.
left until equilibrium was reached (minimun time 2 h) and a sample
of the liquid phase was taken in order to analyze TBC content. Each
sample was analyzed using a PerkinElmer A/S gas chromatograph,
equipped with a 60 m × 0.32 mm × 1 ␮m DB1 capillary column and
a flame ionization detector. Since TBC is solid at ambient condi-
tions (melting point 50 ◦ C), solubility experiments in styrene were
carried out in the range −23 to 30 ◦ C. To study the styrene–water
system, liquid–liquid equilibrium data were obtained from 3 to
80 ◦ C, in the same thermostatically controlled bath previously
employed. In this case, the stirred tanks were bigger (250 mL) and
were equipped with a syringe needle fixed onto the bottle plug, in
order to take sample of both bottom (aqueous) and upper (organic)
phases, minimizing experimental errors. All the experiments pre-
sented herein, were carried out at least four times analyzing each
sample three times. Experimental errors for all measurements were
below 10% (standard deviation/average value × 100).
The samples of the aqueous phase were analyzed by total
organic carbon analysis (SHIMAZDU VCSH). The styrene compo-
sition was calculated from the total organic carbon measured in
the aqueous phase. The water content in the organic phase was
analyzed by Karl–Fisher titration (Mettler DL32).

2.3. Thermodynamic modelling

2.3.1. Styrene–TBC system

Fig. 4 shows the experimental solubility data, along with the
supposed ideal behaviour of the mixture (solid line). Strong devi-
ations from ideality are observed, which are expected from the
polar structure of catechol. As suggested by Prausnitz et al. [14],
the ideal solubility could be written as ln x2 = −fus S/R(Tfusion /T − 1),
where only pure component properties, entropy of fusion (fus S)

Fig. 3. Thermostatically controlled bath: (1) water bath; (2) temperature controller;
(3) magnetic stirrers; (4) tanks for liquid–liquid equilibrium; (5) plug with fixed
Fig. 2. Initiator deactivation because of moisture. needle.
I. Díaz et al. / Chemical Engineering and Processing 49 (2010) 367–375 369

Fig. 4. Experimental solubility of TBC in styrene.

2.3.3. Binary interaction parameters

Table 1 shows the binary interaction parameters of each mix-
ture, obtained from the NRTL fit. These values have been obtained
by fitting the experimental data to a maximum-likelihood objec-
tive function in the form (1) by means of a typical Britt–Luecke
non-linear fitting algorithm:

NP  2 NC  2
Test,i − Texp,i xest,j,i − xexp,j,i
Q = ⎣ +
T,i x,j,i
i=1 j=1

yest,j,i − yexp,j,i
+ ⎦ (1)

where NP is the number of points fitted, Test is the estimated value of

temperature, Texp is the experimental value of temperature,  T,i is
Fig. 5. Activity coefficient of TBC in styrene solution. the standard deviation of the temperature data i, NC is the number
of components in the mixture, xest is the estimated value of liquid
composition, xexp is the experimental value of liquid composition,
 x,i is the standard deviation of the liquid composition data i, yest is
the estimated value of vapour composition, yexp is the experimental
value of vapour composition and  y,i is the standard deviation of
the temperature data i.
The average deviation in terms of activity coefficient for the fit
presented in Fig. 5 is 8.2%. The fitting of the T-x-x data for the system
styrene–water results in an average deviation (in terms of liquid
composition of styrene) of 2.4%.
The values in Table 1 have been implemented in ASPEN PLUS®
commercial software [16] to investigate the process economics.

3. Economic evaluation of two-distillation process

Once a reliable thermodynamic model had been developed, it

Fig. 6. Styrene–water liquid–liquid equilibrium. was used in the styrene purification design. The assumed specifica-
tion was 30,000 tonnes/year of styrene monomer in 8000 working
Table 1
hours (3750 kg/h), with a final composition of both water and
NRTL binary interaction parameter regressed.
TBC below 1 ppm. First of all, we adopted a conventional two-
 ij = aij + bij /T (K) Styrene (i)–TBC (j) Styrene (i)–water (j) distillation column scheme. This process is the simplest from an
aij −15.8 −175.1 engineering point of view, but it has the disadvantage that the
aij 8.3 148.3 energy consumption is high, which is an important issue for the
bij 5446.5 10,068.1 future both environmentally and economically. Hence, an alterna-
bij −2401.9 −4874.9
˛ 0.2 0.3
tive combined extraction–distillation process, which reduces the
energy consumption, is investigated in Section 4.
370 I. Díaz et al. / Chemical Engineering and Processing 49 (2010) 367–375

Fig. 7. Process with two vacuum distillation columns in series.

All economic assessments were performed in terms of minimum

“economic cost” (EC ) [17]:

Ec = Cv + Ct + (ir + im ) × FC (2)

where Cv are the process variable costs, mostly annual utility con-
sumption, Cf are the annual fixed costs, i.e. maintenance and wages,
FC is the fixed capital investment, ir is the fixed capital recovery rate
applied to FC , and im is the minimum acceptable rate of return on
FC . Cf is assumed to be 10% of FC , and ir + im is assumed to be 25% of
FC hence EC expression can be rewritten:

Ec = Cv + (0.35) × Fc (3)

Hence, in terms of economics, the lower the value of EC the

In the distillation process shown in Fig. 7, the water is removed
first through the top of the first column (COLWAT) and TBC is
Fig. 8. Total heat consumption as a function of total pressure of the columns (points
removed next at the bottom of the second one (COLTBC). Hence,
near coincident for different values of pressure in COLWAT).
the time that free inhibitor styrene goes through the process is
minimized. In other words, the number of unit operations in which
styrene goes through without TBC is minimized, so minimal fouling However, the vacuum applied in the columns is also costly.
due to thermal polymerization of styrene is expected. Industrial vacuum is usually carried out by means of ejectors. In
Firstly, the design was carried out by using the experimen- this case we have considered air ejectors, whose operational cost
tal thermodynamic information previously related (NRTL model). is directly related to the amount of vapour to be removed from the
Then, the results were compared with those obtained using the columns. The amount of air required is calculated as in Perry [18],
UNIFAC thermodynamic model. One of the aims of this work is to based on 1% leakage of total inlet flowrate.
quantify the economic accuracy of choosing a good thermodynamic Thus, it was necessary to perform an economic study of both
model. capital and operating costs not only as a function of the differ-
The pressure of the system is an important aspect which was ent thermodynamic models proposed, but also as a function of the
taken into account, because it was observed that the lower the column pressures.
working pressure in the columns, the lower the energy consump- When using the UNIFAC thermodynamic model, it was nec-
tion. The pressure of the first column is not so critical, because most essary to redesign the columns. In this case, the separation was
of the energy is consumed in the second column, where styrene is
evaporated, as shown in Fig. 8. Several authors [5,6] report that the
thermal polymerization mechanism is important at temperatures
above 50–85 ◦ C. Therefore, the pressure range studied herein is lim-
ited to 120 mmHg. Fig. 9 shows the vapour pressure of styrene as
a function of temperature. The column temperatures allowed low
pressure steam in the reboilers and cooling water in the condensers.
Air cooling would be an alternative. A summary of unit utility costs
used here is shown in Appendix A.
Using the experimental equilibrium data, the required purity
was achieved with: first column (COLWAT) with 4 theoretical
stages (feed in stage number 2) and second column (COLTBC) with
3 theoretical stages (feed in stage number 2). It was observed that
reflux ratio is not a sensitive factor for this separation, because of
the low number of theoretical stages. It seems that the addition or
removal of one theoretical stage is more significant than an increase
or decrease of the reflux ratio. Thus, the relux ratio was fixed in a
low value of 0.01 (minimum reflux ratio is very small), in order to
minimize operational costs. Fig. 9. Vapour pressure of styrene (ASPEN PLUS database).
I. Díaz et al. / Chemical Engineering and Processing 49 (2010) 367–375 371

Table 2
Capital (D ) and operating cost (D /y) summary for two-distillation column process.

Component name NRTL UNIFAC

20 mmHg 120 mmHg 20 mmHg 120 mmHg

Capital investment (D )
Column (COLTBC) 112,400 111,700 187,175 112,600
Column (COLWAT) 79,900 110,500 234,530 113,800
Condenser (COLTBC) 75,500 63,900 75,500 63,900
Reboiler (COLTBC) 70,900 70,900 70,700 71,300
Reboiler (COLWAT) 69,200 65,600 78,200 64,900
Ejectors (COLTBC) 28,900 12,500 28,900 12,500
Reflux pump (COLTBC) 27,300 27,300 27,300 27,300
Reflux pump (COLWAT) 20,300 22,600 22,600 22,600
Condenser accum. (COLWAT) 16,800 15,000 16,800 15,000
Condenser accum. (COLTBC) 12,500 13,300 12,500 13,400
Condenser (COLWATa ) – – – –

Total capital (D ) 513,700 513,300 754,200 517,300

Operating costs (D /y)

Reboiler (COLTBC) 49,488 46,888 49,480 46,888
Condenser (COLTBC) 7952 7520 7520 7520 Fig. 11. Cost summary for two-distillation column process for the design with NRTL.
Ejectors (COLTBC) 5680 4080 5680 4080
Reboiler (COLWAT) 4448 13,472 4464 13,424 Fig. 11 shows the variation of the process costs with the pres-
Condenser (COLWAT) 176 176 176 176
sure. It can be seen that these are larger at 120 mmHg, mainly
Operating costs (D /y) 67,700 72,100 67,300 72,000 because of energy consumptions. For the experimental thermo-
Economic cost (D /y) 247,000 252,000 332,000 253,000 dynamic model, the difference in EC with pressure are small. The
choice of column pressure is more likely to be made on grounds of
Neglected because design results in a very small condenser.
safety and control.

4. Economic evaluation of combined

possible with 4 theoretical stages for TBC separation and with 7
extraction–distillation process
theoretical stages for water separation.
In both cases, UNIFAC and NRTL, structured packings type BX To design this process, a material separation agent has been
of SULZER were selected, attending to specialized literature that added. The aim is to eliminate the bottom product (TBC) by an
recommends using packings instead of trays when working under extraction process in order to avoid the complete evaporation of
vacuum [19]. Structured packing has higher efficiencies for systems the styrene stream. The selection between different extractants was
containing styrene (HETP = 8 ) than random packing. made taking into account the relative affinity for TBC and styrene.
Table 2 shows the economic results for the different designs. Usually, TBC is sold as a saturated solution containing 85% TBC
The major product flowrate, styrene, occurs at the top of the TBC and 15% methanol [20,21]. The reported specific value of solubil-
column, hence there is a high energy demand in its reboiler and ity of TBC in methanol is 500 g/100 g [22]. The problem of using
condenser. The reboiler utility is more expensive. The vacuum cost methanol as extractant is that it is miscible with styrene, so the
is small in comparison with energy consumptions. It can be also binary methanol–styrene mixture would be homogeneous and not
seen that the lowest EC is obtained at 20 mmHg but it is not sensitive suitable for liquid–liquid extraction.
to pressure. UNIFAC overstimates capital costs by up to 15%. There A compound which forms two liquid phases with styrene is
are negligible differences in operating costs. Fig. 10 highlight that, water, but the problem of water is that it can only dissolve 2 g of
as expected, COLTBC reboiler is the most significant operational TBC per litre [22].
cost. So, taking into account the previous considerations, a suit-
able solvent for the proposed separation system, could be a
methanol–water mixture. This binary solvent can form two liquid
phases with water, while keeping a great affinity for TBC. In the
design of the process it was observed that with a feed of mixture
50/50 (w/w), the styrene can be purified, minimizing the consump-
tion of methanol.
In the proposed scheme (Fig. 12), the inlet stream goes directly
to the L–L extraction unit where the TBC is removed with the
methanol–water solvent. The extract outlet stream is then car-
ried to a simple distillation unit (COL1, 3 theoretical stages) where
TBC is obtained from the bottom and the methanol and water top
stream is recirculated to the extraction unit. Although the styrene
is impurified with methanol and water (raffinate outlet stream) in
the extraction unit, these two last compounds can be easily sep-
arated from styrene in a distillation column (COL2, 7 theoretical
stages with a reflux ratio of 0.35), where pure styrene is obtained
as a bottom product. The distillation column works at 120 mmHg
of pressure in order to avoid thermal polymerization. Low pressure
steam is used in the reboilers and cooling water is used in the con-
denser for COL1. COL2 requires using chilled water because of the
Fig. 10. Operating cost summary for two-distillation column process at 20 mmHg. lower temperatures.
372 I. Díaz et al. / Chemical Engineering and Processing 49 (2010) 367–375

Fig. 12. Combined process extraction–distillation for the purification of styrene monomer.

The influence of the number of theoretical stages (NE ) in the

extraction column and extractant flowrate (FE ) is the most impor-
tant in terms of process EC . Therefore, the first optimization study
investigated variations of NE and FE maintaining 7 stages in COL2.
Then, for the best pair of variables NE and FE , the optimization of
COL2 was carried out by the classical analysis of calculating opera-
tional and capital costs for different pairs of reflux ratio-number of
stages required.
The extraction unit can operate with high NE and low FE , high
capital low operating costs, or vice versa. In the latter case the
operational costs of methanol would be higher which is the most
significant variable for the extraction–distillation process.

4.1. Extraction optimization

Random packings of INTALOX ceramics rings of 1 in. of diameter

Fig. 13. Dependence of capital cost with NE at different FE .
were assumed. For this kind of column an efficiency of 2 stages/m
and a total capacity of 20 m3 /h m2 is assumed [18].
Capital (Fig. 13), operating (Fig. 14) and EC (Fig. 15) costs have 4.2. Final column (COL2) optimization
been calculated for different of NE –FE . The minimum in EC deter-
mines the optimum design. It was necessary to manipulate the Finally, the economic optimization COL2 was carried out in
reflux ratio in COL2 to achieve purification. For each simulation, terms of the classical representation of operating cost and capi-
the COL2 inlet composition was different so, for a column of 7 tal as a function of the reflux ratio (Fig. 16). Values of 115 kg/h for
theoretical stages, the reflux ratio employed should be different
(Table 3).
Table 3
Fig. 13 shows that the capital costs strongly depend on the size of
Reflux ratio used in COL2 to achieve purity specifications.
the extraction column employed. The influence of extractant flow
is smaller. Operating costs (Fig. 14) have a minimum value for an Extraction column theoretical stages
extractant flow of 115 kg/h. Lower flowrates require higher recircu- 10 15 20 25
lation flows so the energy costs in COL1 increase. Higher extractant
Extractanct flow (kg/h)
flow also results in a high cost of the methanol used. The influence 90 – – 1.25 0.86
of NE over the operating costs is important too, showing that larger 100 – 0.82 0.54 0.45
extraction columns need less extractant to carry out the separation 115 0.80 0.45 0.35 0.33
(Fig. 15). 130 0.60 0.40 0.31 0.28
I. Díaz et al. / Chemical Engineering and Processing 49 (2010) 367–375 373

Table 4
Capital (D ) and operating cost (D /y) summary for the combined
extraction–distillation process.

Component name NRTL UNIFAC

Capital investment (D )
Column (COL2) 133,100 133,100
Column (COL1) 110,500 110,600
Extraction column 96,000 96,000
Condenser (COL1) 66,300 62,100
Reboiler (COL1) 65,100 65,800
Reboiler (COL2) 63,500 65,300
Condenser (COL2) 53,800 53,800
Reflux pump (COL1) 22,700 22,700
Reflux pump (COL2) 22,700 22,700
Condenser accumulator (COL1) 16,500 16,500
Condenser accumulator (COL2) 15,000 15,000
Ejectors (COL2) 12,500 12,500

Total capital (D ) 677,700 676,100

Fig. 14. Dependence of operating costs NE at different FE . Operating costs (D /y)

Methanol 62,560 32,640
Reboiler (COL1) 16,437 14,253
Reboiler (COL2) 15,373 15,953
Condenser (COL2) 4628 5072
Air for ejectors 2903 2903
Condenser (COL1) 2278 2037

Operating costs (D /y) 104,000 72,800

Economic cost (D /y) 341,000 309,000

extractant flow and 15 theoretical stages were fixed for the extrac-
tion column. The optimum now is the design with the minimum
EC,COL2 (EC,COL2 = Cv,COL2 + 0.35·FC,COL2 ). Thus, a column of 6 theoreti-
cal stages operating at a mass reflux ratio of 0,53 would determine
an EC,COL2 around 130,000 D /y.

4.3. Final design, comparison with UNIFAC model

The economic assessment was repeated using the UNIFAC ther-

modynamic model. The optimum was obtained with: FE = 60 kg/h,
NE = 15 and RRCOL2 = 0.77. The capital cost is similar to the NRTL
model and the operating cost are 30% lower as shown in Table 4.
It can be seen that the highest operating cost is due to methanol
(Fig. 17). That can be minimized by further purification of the dis-
tillate stream in COL2, obtaining pure methanol to be storage and
reused in the process. Thus, the operating cost would be reduce in
50,000 D /y but the capital cost (because of the distillation column
Fig. 15. Dependence of EC with NE at different FE (the maximum in the figure is the required) would increase in 200,000 D so the EC would increase too.
minimum EC , the optimum design).

Fig. 16. EC as a function of the reflux ratio for COL2.

Fig. 17. Operating cost summary for extraction–distillation combined process.
374 I. Díaz et al. / Chemical Engineering and Processing 49 (2010) 367–375

Table 5
Operational costs of different processes.

Process Operating cost (D /y) Capital cost (D ) EC (D /y)

Adsorption (only accounted adsorbent material cost) 200,000 – –

Distillation 67,000 51,3000 24,7000
Combined extraction–distillation (methanol outlet stream valuation not carried out) 104,000 678,000 341,000
Combined extraction–distillation (all outlet methanol reused in the process) 50,000 878,000 357,000

5. Conclusions could be feasible. For the case where most of the methanol would
be purified and recirculated in the process (assuming zero cost of
The purification of styrene monomer has been studied in order methanol), the combined process would be even more appealing in
to seek a new process with lower costs than the current method terms of operating cost. The operational costs of all processes pre-
(adsorption onto alumina). Economic assessments (capital and sented here are much lower than the current adsorption process.
operating) of two alternatives have been undertaken showing the There are significant differences in the final results of EC for the
lowest costs for a conventional distillation process. Experimen- proposed processes when using different values of mutual solubili-
tal measurements of the phase equilibria were undertaken which ties of the various binary mixtures (experimental values or UNIFAC
significantly improves the accuracy of the economics. Table 5 com- predicted values).
pares the EC of the two processes proposed with the current Future research should investigate improving the com-
purification process. bined extraction–distillation process with other extractants and
The relative high price of alumina (1000 D /tonne) as well as the improvements by complex distillation designs (divided wall
price of solid waste treatment, make the current process expen- columns, thermal coupling, etc.).
sive. A consumption of 200 tonnes/y of alumina is estimated for
The two-column distillation process is the most economic. Since Acknowledgment
this process has a high energy usage, which may become more
significant, a combined extraction–distillation process with lower Financial support from Spanish Education Ministry (AP2006-
energy use is proposed. 02300) for a short stay in a foreign institution is acknowledged.
This combined process is feasible and reduces energy consump-
tion by 38%, however, the costs are higher mainly due to the
consumption of methanol. The capital needed is also higher than Appendix A.
conventional distillation because the required equipment would
be larger. In a future scenario of very high energy costs this process See Table A1.

Table A1
Utility costs.

Utility description Price

Cooling water (21 ◦ C, D /m3 ) [23] 0.0145

Chilled water (10 ◦ C, D /m3 ) [24] 0.225
Steam (690 kPa, D /tonne) [23] 8.18
Compressed air (D /m3 ) [11] 0.0117
Methanol (D /tonne) [25] 136

Table B1
Stream data for the two-distillation process at 20 mmHg (NRTL model).

Feed stream Bottom COLWAT Distillate COLWAT Bottom COLTBC Distillate COLTBC

Temperature ( C) 25 46 9 47 10
Pressure (mmHg) 20 20 20 20 20
Mass flow (kg/h) 3750 3745 5 5 3740

Mass frac
Water 0 0 0.225 0 0
Styrene 1 1 0.775 0.925 1
TBC 0 0 0 0.075 0

Table B2
Stream data for the optimized combined extraction–distillation process (NRTL model).

Feed stream Fresh Extractant to Raffinate Extract Bottom COL1 Distillate COL1 Distillate Bottom COL2
extractant extraction column stream stream COL2

Temperature (◦ C) 25 25 60 31 23,5 95,2 73 26 87

Pressure (mmHg) 760 760 760 760 760 760 760 120 120
Mass flow (kg/h)3750 115 415 3806 358 58.9 300 71.0 3735

Mass frac
Water 0 0.5 0.372 0.001 0.424 0.943 0.322 0.043 0
Styrene 1 0 0.043 0.985 0.049 0 0.059 0.19 1
TBC 0 0 0 0 0.001 0.006 0 0 0
Methanol 0 0.5 0.586 0.014 0.525 0.051 0.619 0.767 0
I. Díaz et al. / Chemical Engineering and Processing 49 (2010) 367–375 375

Appendix B. [10] M.S. Peters, K.D. Timmerhaus, R.E. West, Plant Design and Economics for Chem-
ical Engineers, 5th ed., McGraw-Hill, New York, 2003.
[11] IchemE education subject group: http://ed.ichemeorg/.
See Tables B1 and B2. [12] A.M. Gerrard, Guide to Capital Cost Estimating, 4th ed., IChemE, UK, 1988.
[13] R. Sinnot, G. Towler, Chemical Engineering Design, 5th ed., Elsevier, Amster-
References dam, 2009.
[14] J.M. Prausnitz, R.N. Lichtenthaler, E. Gomes de Azevedo, Molecular Thermody-
namic of Fluid Phase Equilibria, 3rd ed., Prentice Hall, New Jersey, 1998.
[1] H.L. Hsieh, R.P. Quirk, Anionic Polymerization, Principles and Practical Appli- [15] W.H. Lane, Determination of the solubility of styrene in water and of water in
cations, Marcel Dekker, New York, 1996. styrene, Ind. Eng. Chem. Anal. Ed. 18 (1946) 295–296.
[2] E.R. Moore, Anionic process design for rapid polymerization of polystyrene [16] Aspen Technology, Aspen Plus 2006 User Guide, Cambridge, MA, 2006.
without gel formation and product produced there from, U.S. Patent [17] E. Díez, P. Langston, G. Ovejero, M.D. Romero, Economic feasibility of heat
2003/0073792, 2003. pumps in distillation to reduce energy use, Appl. Therm. Eng. 29 (2009)
[3] UOP company: 1216–1223.
[4] M.J. Rivero, Ph.D. Thesis, Universidad de Cantabria, Spain, 2002. [18] R.H. Perry, D.W. Green (Eds.), Perry’s Chemical Engineer’s Handbook, 7th ed.,
[5] C.C. Liao, S.H. Wu, T.S. Su, M.L. Shyu, C.M. Shu, Thermokinetics evaluation McGraw-Hill, New York, 1997.
and simulations for the polymerization of styrene in the presence of various [19] H.Z. Kister, Distillation Design, McGraw-Hill, New York, 1992.
inhibitor concentrations, J. Therm. Anal. Calorim. 85 (2006) 65–71. [20] Rhodia manufacturers:
[6] C.C. Chen, Y.S. Duh, C.M. Shu, Thermal polymerization of uninhibited [21] ECEM chemical distributor:
styrene investigated by using microcalorimetry, J. Hazard. Mater. 163 (2006) [22] ACROS chemical distributor:
1385–1390. [23] Aspen Technology, Aspen ICARUS 2006 User Guide, Cambridge, MA, 2006.
[7] P. Schmaus, W.G. Mezder, Procedure for the distillation of vinylaromatic [24] A.C. Dimian, Integrated Design and Simulation of Chemical Processes, McGraw-
monomers in the presence of oxygen and 4-tertiary-butylcatechol as polymer- Hill, Amsterdam, 2003.
ization inhibitors, U.S. Patent 2002/0040845, 2002. [25] ICIS pricing report:
[8] H. Renon, J.M. Prausnitz, Local compositions in thermodynamic excess func-
tions for liquid mixtures, AiChe J. 14 (1968) 135–144.
[9] A. Fredenslund, R.L. Jones, J.M. Prausnitz, Group-contribution estimation of
activity coefficients in nonideal liquid mixtures, AiChe J. 21 (1975) 1086–1099.