You are on page 1of 13

Separation and Purification Technology 290 (2022) 120821

Contents lists available at ScienceDirect

Separation and Purification Technology


journal homepage: www.elsevier.com/locate/seppur

Pervaporative recovery of aroma compounds in the production of


non-alcoholic beers: Incorporation of different condensation strategies into
the conceptual design of the process
Danilo Alexander Figueroa Paredes a, 1, Ramiro Julián Sánchez a, b, 1, Micaela Magariño c,
José Espinosa a, b, *
a
INGAR (CONICET-UTN), Avellaneda 3657, S3002GJC Santa Fe, Argentina
b
Universidad Nacional del Litoral, Facultad de Ingeniería Química, Santiago del Estero 2829, 3000 Santa Fe, Argentina
c
Núcleo TECSE, Facultad de Ingeniería, Universidad Nacional del Centro de la Provincia de Buenos Aires, Olavarría, Argentina

A R T I C L E I N F O A B S T R A C T

Keywords: The study carried out in the present work focused on the conceptual design of the aroma recovery stage of from a
Beer lager beer by pervaporation with a PDMS membrane taking into account the condensation step.
Aroma recovery First, using laboratory-scale experiments, the performance of the Pervatech PDMS membrane was measured in
Pervaporation
terms of aroma recovery, ratio of higher alcohols to esters (A/E ratio), ethanol flux and overall flux for different
Partial condensation
operating conditions. In addition, model was obtained that adequately characterized the fluxes through the
Total condensation
Conceptual design membrane as a function of temperature and permeate pressure. This model allowed estimating the membrane
area requirements for different operating conditions.
Secondly, the possibility of using two partial condensers in the condensation stage to approximate the A/E
ratio of a lager beer was explored through simulations. Influence of operating conditions and presence of CO2 and
air on values for the A/E ratio, the overall fractional recovery of aroma compounds and the economics of the
process were modeled. Both the case of mixing condensates obtained from the two condensers and that of an
optimal combination of condensates were taken into account in order to approximate the A/E ratio of the lager
beer.
The best alternative for this variant was the one corresponding to the operating conditions of 30 ◦ C tem­
perature and a pressure of 2.13 kPa for the pervaporation stage, with two partial condensers in series operating at
− 5 ◦ C and − 46 ◦ C, respectively. In this way, two condensates would be obtained which, when combined in a
0.43/1 ratio for subsequent reincorporation into the dealcoholized beer, would result in a product maintaining
the (A/E) ratio of the original beer. Under these conditions, a recovery of 14.1% of aromas would be achieved
with a total cost of 0.1070 USD/L.
Finally, the state-of-the-art alternative consisting of two total condensers condensers in parallel working out of
phase was also considered. Results indicate that, if properly designed, the A/E ratio obtained at lab scale level
would coincide with that achieved after total condensation and thus, the adjustability of the ratio of higher
alcohols to esters would decrease with respect to that achieved when partial condensers are used. Although the
use of a total condenser results in more expensive designs, this alternative is very advantageous in terms of
overall aroma recovery and process feasibility.
From the results obtained, the importance of developing conceptual models that make use of both experi­
mental runs of the aroma recovery process at laboratory scale and simulations of the condensation process from
parameters corresponding to the vapour-liquid equilibrium of the multi-component system can be deduced.

* Corresponding author at: INGAR (CONICET-UTN), Avellaneda 3657, S3002GJC Santa Fe, Argentina.
E-mail address: destila@santafe-conicet.gob.ar (J. Espinosa).
1
Danilo A. Figueroa Paredes and Ramiro J. Sánchez contributed equally to this work.

https://doi.org/10.1016/j.seppur.2022.120821
Received 16 February 2022; Received in revised form 8 March 2022; Accepted 9 March 2022
Available online 12 March 2022
1383-5866/© 2022 Elsevier B.V. All rights reserved.
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

1. Introduction Table 2
Sub-tasks involved in the whole process.
Paz et al. [1] address the recovery of beer aromas by pervaporation Dealcoholization task with aroma recovery via nanofiltration/pervaporation
with PDMS (polydimethylsiloxane) membranes. The authors begin by
aroma recovery stage – step 1 (pervaporation)
demonstrating the profound deterioration of the aromatic profile pre-concentration stage – step 2 (nanofiltration)
brought about by dealcoholization in the case of five non-alcoholic constant volume diafiltration stage – step 3 (nanofiltration)
commercial beers. The percentage reduction in the concentration of re-dilution stage – step 4
the two esters and the three alcohols studied, depending on the beer, aroma addition stage – step 5
CO2 level adjustment – step 6
varies between 35% and 100% as shown in Table 2 of the cited manu­ Alcohol valorization task through distillation
script. Performing experiments at a feed temperature of 25 ◦C and alcohol-rich fraction recovery at variable reflux ratio – step 1
vacuum levels of around 10 mbar (Pervatech 030705) and 14 mbar water-rich fraction recovery at constant reflux ratio – step 2
(PERVAP 4060), the authors characterized the separation performance Water usage minimization task
mixing of the water-rich fraction from the distillation stage with deionized water
of each aroma compound in terms of flux, mass recovery and enrichment
make-up
factor. During experiments, the maximum permeate amount collected,
measured as a percentage of the feed, is about one percent for the Per­
vatech membrane and 0.5 percent for the Sulzer membrane. compounds are obtained by pervaporation of the original beer using a
Catarino et al. [2] studied the aroma extraction from beer by per­ (POMS/PEI) composite membrane, the aroma depleted beer is subjected
vaporation, using a polyoctylmethylsiloxane/polyetherimide (POMS/ to dealcoholization in a spinning cone distillation column with water
PEI) composite asymmetric membrane. A response surface methodology vapor as stripping agent. After dealcoholization, the beer is blended with
(RSM) was adopted by the authors to describe the influence of relevant the extracted aroma compounds and with a fraction of original beer to
operating conditions; namely, feed temperature, feed velocity and achieve a non-alcoholic beer with an ethanol content lower than 0.5% v/
permeate pressure on process performance. The responses considered v and with a good flavor profile as the ratio between each aroma con­
were the permeate flux, the aromas/ethanol selectivities, the ethanol centration and ethanol is close to the original beer. It is noteworthy that
concentration and the ratio between higher alcohols and esters con­ the dealcoholization process itself brings a deep deterioration of the
centrations on the permeate, being the last one an important indicator aroma profile as shown in Table 3 [4]. This result is in line with those
for evaluating the alcoholic or fruity character of a beer. A desirability shown in Paz et al. [1], Brányik et al. [5] and Sánchez et al. [6].
function was used to find the optimal conditions for multiple responses In our previous work [6], we addressed the conceptual modeling
and to solve the trade-off between them. The optimal operating condi­ concerning the dealcoholization of beers with simultaneous permeate
tions were 12.4 ◦C for feed temperature, 0.45 ms− 1 for feed velocity and valorization through a hybrid process composed of a membrane unit and
1.0 mbar for permeate pressure. A desirability of 0.54 was achieved, a distillation unit. Regarding the modeling of the dealcoholization task
meaning that 54% of the goals were satisfied. For these conditions the through nanofiltration, we used the model developed by Avlonitis et al.
permeate flux was 7.26E-05 kgm− 2s− 1; the high alcohols selectivity [7] to describe the dependence of the average flux on the trans­
ranged from 1.31 to 3.39; the esters selectivity ranged from 14.46 to membrane pressure and the degree of concentration. The referred model
17.10 and the higher alcohols/esters ratio was 1.07. includes the estimation of the difference of the osmotic pressure
Olmo et al. [3] analyzed the separation performance of a PDMS considering a linear dependence of this with the solids concentration,
membrane commercialized by Pervatech in terms of flux and enrich­ which in turn depends on the degree of concentration. From the
ment factor for three typical aroma compounds; namely, ethyl acetate, experimental data, the parameters B and Φ necessary to model both the
isobutyl alcohol and isoamyl acetate. The time evolution of these vari­ flux and the osmotic pressure difference were obtained (see Figs. 2 and 3
ables is shown for the case of two lager beers by operating the lab scale in [6]). Its incorporation in the conceptual modeling of the nano­
setup at 25 ◦ C (feed temperature) and 0.1 kPa (permeate pressure). The filtration unit allowed to adequately estimate the time evolution of the
recirculation flow rate was set to 1.8 L min− 1 corresponding to a average flux during the complete dealcoholization task (see Figure 7 in
tangential velocity of 1 ms− 1. In the mentioned manuscript, selectivities [6]). The scaling-up of the process took into account the pressure loss
with respect to ethanol were also roughly predicted by a model based on issue for the case of spiral wound type modules [8]. For the case of the
the Hildebrand solubility parameters for the polymer and the species in distillation task, on the other hand, the minimum energy demand was
the solution. The theoretical approximate calculation of relative selec­ estimated resorting to pinch-based methods [9,10] capable to deal with
tivities from solubility parameters could help, according to the authors, highly nonideal azeotropic mixtures [11,12]. Fig. 1 shows the variable
in the design of the process and the selection of the membrane in order to reflux policy corresponding to the optimal design for the hybrid process
comply with specific separation requirements. during the main cut of the distillation step. It is noteworthy that the
Catarino and Mendes [4], in accordance with their previous study at conceptual model of each unit operation incorporated a detailed cost
lab scale, proposed a novel industrial process for producing non- model in order to capture the operation and investment costs. Since the
alcoholic beer with a corrected natural flavor profile. While the aroma system was designed to operate batch wise, the presence of recycles

Table 1
Evolution of the mass balance, the steam requirement and membrane area corresponding to design D8 [6]. Pseudo steady state is achieved after seven batches.
Batch Membrane Beer Permeate Mass Deionized Distillation Mass Bottom Recovered Recovered Recycled Mass Steam
area (m2) (kg/ (kg/ fraction water (kg/ Feed (kg/ fraction (kg/ ethanol water (kg/ distillate fraction (kg/
batch) batch) (wt.%) batch) batch) (wt.%) batch) (kg/batch) batch) cut (kg/ (wt.%) batch)
batch)

0 27.5 3000 4179.2 2.5 1207.5 4179.2 2.5 3811.5 92.5 250.0 25.3 10.1 813.2
1 33.7 3000 5019.0 2.1 526.8 5044.3 2.2 4721.3 101.6 274.7 16.7 10.4 953.7
2 35.4 3000 5248.1 2.1 362.3 5264.7 2.1 4944.4 102.6 277.4 12.2 10.6 980.7
3 35.8 3000 5306.6 2.1 319.4 5318.8 2.1 4999.4 102.5 277.3 11.8 10.6 987.0
4 35.9 3000 5318.8 2.1 310.9 5330.6 2.1 5011.5 102.6 277.5 11.3 10.6 988.4
5 35.9 3000 5322.4 2.1 309.1 5333.7 2.1 5015.5 102.7 277.6 10.2 10.7 988.4
6 35.9 3000 5324.6 2.1 308.6 5334.8 2.1 5016.6 102.5 277.2 10.7 10.7 988.1
7 36.0 3000 5325.1 2.1 312.3 5335.9 2.1 5017.6 102.6 277.3 10.7 10.7 987.1

2
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

Table 3
Operating conditions, permeate and CO2 fluxes and ethanol mass fraction in the permeate corresponding to aroma recovery step with Pervatech PDMS membrane.
Model parameters: xethanol = 0.259 g g− 1,PWO /l = 7.29 mol m-2h− 1 kPa− 1, EapW = 18.3 kJ/mol. R2 = 0.995.
Run # Temp. Permeate Permeate/CO2 flux Ethanol mass fraction Jethanolexp JPcalc Jethanolcalc Error in ethanol flux Error in permeate flux
(g/g)
-2 − 1 -2 − 1 -2 − 1 -2 − 1
(C) pressure (kPa) (gm h ) (gm h ) (gm h ) (gm h ) (%) (%)

A 40 1.60 823/589 0.254 209 824 213 2.1 0.2


B 40 2.13 775/577 0.280 217 758 196 9.5 2.1
C 30 1.60 481/471 0.259 125 492 127 2.2 2.2
D 30 2.13 367/314 0.270 99 409 106 6.8 11.4
E 20 1.60 245/145 0.240 59 203 53 10.4 17.0
F 20 2.13 121/67 0.250 30 97 25 16.9 19.8

condensation variant will be considered. In the latter case, duplication of


the total condenser is considered (Fig. 2(b)) in order to allow the semi-
continuous production of the aroma rich-concentrate [4].

2. Problem statement

The maximum operating time allotted for the aroma recovery task
via pervaporation with a hydrophobic membrane (stage 1 in Table 2) is
set at 4 h per batch, one batch per day with a time horizon of 240 days
per year. The maximum quantity of permeate to be obtained at this stage
represents 1 % of the batch of beer to be dealcoholized (3000 kg) and
will be added together with an equivalent quantity of fresh beer in stage
5 (see Table 2) to the dealcoholized beer. For the condensation of the
permeate under vacuum, the alternatives of total condensation at a
temperature below the freezing point of the ethanol-water mixture or
the use of two flash condensers will be considered. In both cases, the
conceptual model for the whole process will be built from experiments at
lab scale (pervaporation) and simulations (condensation system). The
performance of each alternative will be measured in economic and
Fig. 1. Evolution of the reflux ratio during the constant composition stage
quality terms, with the latter assessed through the ratio of higher alco­
corresponding to the optimal design for the hybrid process considering the
recycle of the water-rich residuum at the end of the distillation stage to the next
hols to esters achieved after the aroma addition task.
batch of the dealcoholization task via nanofiltration.
3. Materials and methods
required obtaining optimal operating and investment costs in the pseudo
stationary state [13,14]. Table 1 shows the evolution towards the 3.1. Materials
pseudo stationary state for the design D8 (optimal). The preference for
the alternative that contemplates recycling the contents of the boiler at The lager beer (before the carbonation stage) used in the experiments
the end of the distillation step to the membrane unit was based on was provided by a local beverage company. The same lager beer, with an
environmental aspects through the life cycle analysis of the process ethanol content of 4.7% by volume, was employed in Sánchez et al. [6]
[15,16] and the convenience of recovering the original content of for the dealcoholization task with the commercial nanofiltration mem­
mineral salts lost during the dealcoholization task previously carried out brane NF99 HF (Alfa Laval, Sweden).
[17]. Furthermore, the mentioned recycle minimizes the usage of The commercial membrane used in pervaporation experiments was
foreign water during the nanofiltration task, which is frequently Pervatech PDMS (Pervatech BV, The Netherlands), a hydrophobic
forbidden in oenological practices [18–20]. membrane of polydimethylsiloxane (PDMS) adopted by other authors in
In the present work we propose to incorporate the task of aroma the recovery of aroma compounds of beer [1] and in the production of
recovery to the already considered tasks of dealcoholization and alcohol-free wines and grape spirits [24].
permeate valorization through the hybrid membrane/distillation pro­
cess. Fig. 2 shows the scheme of the process carried out batch wise. 3.2. Equipment
Table 2, on the other hand, summarizes the tasks involved in the whole
process. After dealcoholization, as suggested in Catarino and Mendes The pervaporation experiments were accomplished in a CELFA
[4], the dealcoholized beer is blended with the extracted aroma com­ Laboratory P-28 (CM-Celfa Membrantrenntechnik AG, Switzerland) unit
pounds and with a fraction of original beer to achieve a low-alcoholic provided with a pressure vessel (0.5 L) with double wall cylinder and a
beer with an alcohol content lower than 1% v/v and with enhanced membrane cell (Fig. 3). This equipment allows the use of flat membrane
aroma profile. From experiments at lab scale with a commercial PDMS discs of 75 mm diameter with an effective area of 28 cm2. Recirculation
membrane and a lager type beer, an emphasis will be given to the flow rate is provided by a gear pump (Scherzinger, Germany) allowing
conceptual design of the process considering different condensation flow velocities from 0.1 to 6 m s− 1. Temperatures in the desired range of
strategies [21,4]. Special attention will be paid to the influence of CO2 [20–40 ◦ C] can be tested through an external thermostat (Ecoline
permeation [21,22] through the pervaporation membranes and the air Staredition E103, Lauda, Germany). To allow continuous permeate
leakage [21,23] in the design and aroma recovery performance of the sample collection, two traps in parallel, cooled with liquid nitrogen, are
condensation system under vacuum. In addition to the alternative con­ coupled to the permeate exit. A three-stage chemistry-design diaphragm
sisting of pervaporation units followed by two partial condensers oper­ pump (MD 4C NT, Vacuubrand, Germany) is used to maintain the
ating under vacuum as shown in Fig. 2(a), the pervaporation/total desired permeate pressure in the desired range of [1.60–2.13 kPa]. This
unit, provided with a nanofiltration membrane, was also used for the

3
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

Fig. 2. Sketch of the dealcoholization process with permeate valorization [6] including the aroma recovery task through pervaporation followed by: (a) two partial
condensers in series; (b) two total condensers in parallel.

4
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

3.3.4. CO2 measurement


The CO2 dissolved in the original beer and retentate samples were
determined using an Orbisphere 6110 analyser (HACH) with automatic
defoamer injection. The samples were placed in 500 mL bottles to be
inserted into the analyser. The carbon dioxide is calculated as a function
of temperature, pressure and thermal conductivity in a working range of
1.5–10 g/kg.

3.4. Conceptual modeling CM

3.4.1. Pervaporation model


Fig. 3. Membrane unit equipment for pervaporation and nanofiltration To describe the overall permeate flux in terms of temperature and
experiments. permeate pressure the following assumptions are made:
Firstly, the molar flux of water jW (molm− 2h− 1) can be expressed in
dealcoholization task as explained in our previous work [6]. The dia­ terms of permeance and driving force from Equation (1) [26]:
filtration stage was performed with the aid of the diaphragm metering ( )
PW ( )
pump Mytho HL10 shown in Fig. 3. jW = xW γ W pfW − yw p (1)
l
( )
3.3. Experiments and analytical methods
In Equation (1), PWl is the water permeance (mol m-2h− 1 kPa− 1),

3.3.1. Experiments xW and yw are feed and permeate water molar fractions, γW and pfW are
Pervaporation experiments were performed in batchwise mode. The the activity coefficient and saturated vapor pressure (kPa) for water in
membrane conditioning was performed following the guidelines pro­ feed and p is permeate pressure (kPa). While activity coefficients are
vided by the suppliers. The recirculation flow velocity selected was 0.45 estimated using the Wilson method, vapor pressures are calculated
m s− 1 in order to achieve a high value of Reynolds number avoiding resorting to the extended Antoine equation. Parameters for pure com­
polarization of both concentration and temperature. Permeate samples ponents and the binary ethanol-water mixture were taken from the
were weighed (Precisa Junior 50) and stored for subsequent analysis. Hysys database [27].
( )
The average values of the overall permeate flux were estimated by
considering the weight of each permeate and the duration of each Secondly, the temperature dependence of the water permeance PlW
experiment until a permeate amount corresponding to 1% of initial feed can be calculated from an Arrhenius type equation, see Equation (2)
was collected [1]. [26]:
( ) ( ) ( ( ))
3.3.2. Ethanol content PW
=
PWO
exp
− EapW 1

1
(2)
Ethanol content of permeates was determined by a colorimetric l l R 303.15 T
method used in our previous work [6,] based in a reaction of ethanol In Equation (2), the reference permeance (PWO /l) (mol m-2h− 1 kPa− 1)
with dichromate ion in presence of acid sulfuric and buffer acetate. A and the water activation energy of permeation EapW (kJ mol− 1) are both
“Spectrum SP2000UV” spectrophotometer at a 570 nm wavelength was the model parameters. R is gas constant (J mol− 1 K− 1) and T is tem­
used and the calibration curve was constructed based on Beer’s law to perature (K).
correlate ethanol concentrations with absorbance. In addition, the Thirdly, ethanol composition in permeate is assumed to be constant
content of ethanol was measured from samples of original beer and in the range of operating conditions studied.
retentate by incorporating a colored compounds removing procedure Under the above assumptions, the overall mass flux J is calculated
according to Sumbhate et al. [25]. Colored samples were mixed with using Equation (3) where MWW is molecular weight of water (g mol− 1)
activated carbon for 1 h and then centrifuged. Ethanol content of and YW is water mass fraction in permeate.
permeate and supernatant samples was measured by duplicate and the
average result was considered. jw *MWW
J= (3)
YW
3.3.3. Aroma composition As will be seen in the Results Section, from the total flux model as a
Aroma compounds from permeate samples were measurement ac­ function of operating conditions it will be possible to estimate in a
cording to the methodology reported by Sánchez et al. [6]. Acetalde­ straightforward manner the corresponding membrane area
hyde, ethyl acetate, propanol, isobutanol, isoamyl acetate and isoamyl requirements.
alcohol concentration was evaluated using Headspace Capillary Gas
Chromatography and Flame Ionization Detection (FID). The analysis 3.4.2. Condensation model
was performed in a gas chromatograph (Perkin Elmer) equipped with a
capillary column J&W Scientific (DB-5) (Standard polysiloxane (5%- 3.4.2.1. Condensation of the pseudo-binary ethanol-water system in the
Phenyl)-methylpolysiloxane 0.53 mm × 60 m × 1.5 µm) and a flame presence of inert gases. For the design of the condensation of the
ionization detector (FID). The temperature was raised from 45 to 150 ◦C permeate stream obtained in pervaporation, the model developed by
at a rate of 40 ◦ C min− 1, with initial and final holds of 1.1 min and 0.6 Brazinha and Crespo [21] for the condensation of aromas in a hydro-
min, respectively. The detector temperature was 150 ◦ C while the alcoholic mixture was used, which contemplates the presence of com­
injector temperature was 100 ◦ C. The chromatograph used nitrogen as ponents such as carbon dioxide and air. The model is described here for
carrier gas with a flow rate of 30 mL min− 1 and the Split was 25 mL the first partial condenser (although the conclusions can be applied to
min− 1. The quantification was performed by calibrations curves using the second partial condenser). The following assumptions are taken into
external standards (Sigma-Aldrich) of each analyzed compound. Aroma account:
content was measured by duplicate.
- The system operates in steady state.
- Thermodynamic equilibrium is reached in the partial condensers.

5
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

- The inert gases flow rate is constant throughout the downstream


circuit, assuming that the inert gases are not captured in the ni1 n’i
condi1 = = 1 − i1 (13)
condensers. ni0 n0
- For the vacuum conditions adopted, the vapor transport in the
downstream circuit has no diffusive resistance. n’i1 x1i .γi1 .piv (T1 )
= (14)
- The contribution of aroma compounds to the permeate pressure is ninert pperm − x1w .γ w1 .pwv (T1 ) − xet1 .γ et1 .petv (T1 )
negligible given the low concentration of aromas with respect to Equation (14) is valid for all condensable components. Introducing
ethanol and water. this equation into Equation (13) gives an expression for the condensed
- The temperature remains constant throughout the condenser. fraction of each component that incorporates the inert gases flow rate,
the permeate pressure and the partial pressures of each component
The partial pressure of the components can be expressed according to (Equation (15)):
Equation (4):
ninert xj1 .γj1 .pjv (T1 )
pi1 = x1i .γi1 .piv (T1 ) (4) condj1 = 1 − j . w w w et et et
j = w, et (15)
p
n0 perm − x1 1 .pv (T1 ) − x1 .γ 1 .pv (T1 )

In Equation (4), pi1 is the partial pressure of component i in the first Applied to the pseudo-binary system, Equation (15), which is also of
condenser, piv is the saturation vapor pressure of component i evaluated a general nature, requires the calculation of the activity coefficients from
at condenser 1 temperatureT1 , γ i1 is the activity coefficient of component the corresponding binary interaction coefficients (see Appendix).
i estimated from the Wilson equation (see Appendix) at the operating Both the compositions of ethanol and water in the condensate and
conditions prevailing in the condenser. vapor stream and the condensate fraction are obtained by iteratively
In turn, the sum of the mole fractions of ethanol and water in the solving the system of equations (4)-(15) for a chosen value of condensing
condensate is approximately 1 given the low concentration of aroma temperature and operating pressure. The energy requirement of the
compounds in the condensate (Equation (5)). condenser is then calculated from the energy balance while the heat
xw1 + x1et ≅ 1 (5) exchange area is estimated by considering typical values for the overall
heat transfer coefficient of condensers operating under vacuum (see
On the other hand, the mole fraction of water on an inert-free basis in Appendix).
the condenser vapor stream can be calculated from the water and In order to avoid freezing of the condensate, a lower bound of the
ethanol flow rates according to Equation (6): condensing temperature is set using the Flick [28] correlation for the
w freezing point of ethanol-water solutions. As well, the calculation of the
nẤ1
yw1 = (6) infiltration air flow rate, which depends on the volume of the equipment
n’w1 + n’et1
under vacuum, was incorporated in the iterative scheme [29].
Having obtained the mass fluxes of the components to be condensed,
the CO2 flux and the membrane area A as described in the previous 3.4.2.2. Condensation of the aromas. After converging the flash equa­
section, the molar flow rates of water, ethanol and CO2 entering the tions for the pseudo-binary system in the presence of inert gases, the
condensation system can be calculated as shown in Equations (7), (8) condensate fraction for each aroma is estimated from Equation (15) and
and (9), respectively. approximating the activity coefficient of each aroma compound to its
corresponding activity coefficient at infinite dilution, as shown in
Jw
nw0 = .A (7) Equation (16):
MW w
ninert xk1 .γ k∞1 .pkv (T1 )
J et condk1 ≅ 1 − . (16)
net0 = .A (8) nk0 pperm − xw1 .γw1 .pwv (T1 ) − x1et .γet1 .petv (T1 )
MW et
k: aroma compound.
J CO2 The activity coefficients at infinite dilution of each component γk∞1
nCO2 = .A (9)
MW CO2 were estimated from the Wilson equation (see Appendix) assuming that
the behavior of each aroma depends only on temperature, pressure and
According to Brazinha and Crespo [21], the inert gases flow rate is a
the composition of the majority components (water and ethanol).
determining variable for estimating the condensation performance. For
To achieve convergence, an iterative calculation is proposed that
the system studied, the inert gases flow rate (ninert ) is composed of the
molar flow rate of CO2 and the molar flow rate of infiltrated air nair (its starts from an estimate of the condensed fraction of an aroma condk1 that
calculation will be detailed later in this section) (Equation (10)). allows the calculation of the flow rate nk1 and its composition xk1 from
Equations (17) and (18). With the value of the composition xk1 it is
ninert = nCO2 + nair (10)
1 . The
possible to calculate a new value of the condensed fractioncond’k
Once the molar flow rates have been established and applying the iterative procedure continues until convergence of the error function
mass balance in the partial condenser and neglecting the contribution of given by Equation (19).
the aromas, Equations (11) and (12) are obtained:
nk1 = condk1 .nk0 (17)
nw0 + net0 + ninert ≅ nL1 + nV1 + ninert (11)
nk1
xk1 = (18)
nw0 = xw1 .nL1 + yw1 .nV1 (12) nL1
nL1 : Total molar flow rate of the condensate at the condenser out­
f = condk1 − cond’k
1 = 0 (19)
let;nV1 : total molar flow rate of the vapor stream at the condenser outlet.
Defining the condensate fraction of component i in condenser 1 by At convergence, nk1 is recalculated.
Equation (13) and considering that the ratio between the vapor flow rate Then, from the mass balance, the molar flow rates into the next
of a component and the inert gases flow rate is equal to the ratio between condenser are obtained.
the partial pressure of the component (see Equation (4)) and the partial
pressure of the inert gases, Equation (14) is obtained. 3.4.2.3. Condensation under freezing conditions. It is reasonable to

6
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

assume a total condensation of ethanol, water, and aroma compounds if adjusted by minimizing least squares error of overall permeate flux.
the condensing temperature reached in the condenser is far below of the Table 4, on the other hand, reports the concentrations of the aromas
freezing point of ethanol-water mixture. In such a case a parallel and their percent recovery with respect to the amount present in the
configuration of heat exchangers, operated in a cooling/heating cycle, original beer. The ratio of higher alcohols to esters (A/E), which de­
should be adopted in order to allow a continuous recollection of the termines the alcoholic or fruity character of the beer, was also added. It
permeate (Fig. 2(b)). On the other hand, a condensing temperature of is noted that the values obtained do not correspond to the value for lager
− 70 ◦ C, which is high enough to prevent the freezing of CO2 in the beer, i.e. (A/E)lager = 6.2. Note that results for runs E and F are not re­
condenser, is selected. ported in Table 4 as the low fluxes prevailing in those operating con­
ditions would lead to high membrane areas to perform the aroma
recovery task within the stipulated 4 h.
3.5. Economic analysis
Table 5 shows the simulation results of different feasible designs of
the condensation task at industrial scale for the operating conditions
The cost related to aroma recovery step is estimated following the
corresponding to run B for the alternative of partial condensation (Fig. 2
same methodology employed in Sánchez et al. [6]. Note that the total
(a)). In all cases the condensate amount represents one percent of the
capital investment for each alternative is annualized taking 10 years as
batch size. For each design, which is characterized by the operation
lifetime with 10% interest rate. Among the operating costs, the mem­
temperatures of the condensers, both the fractional recovery of the
brane replacement calculated from lifetime given by membrane sup­
components and condensate in each condenser are reported. In order to
pliers is also taken into account as an operating cost item. Refrigeration
assess the performance of a given design, values for the A/E ratio and the
utility price is estimated for each condensing temperature using the
overall fractional recovery of aromas are given for both the case of
correlations provided by Ulrich and Vasudevan [30]. Costs related to
mixing condensates obtained from both condensers and that of an
annual utilities consumption are calculated by considering the operating
optimal combination to approximate the A/E ratio of the lager beer. In
cycle time of each process alternative in a time-horizon of 240 days/
the latter case, the condensate that is not added to the dealcoholized
year.
beer can be further processed in the permeate valorization task via
distillation. Design and operation variables together with the cost per
4. Results
liter are also reported.
The membrane area requirement for all designs of Table 5 is ach­
Tables 3 and 4 show the results of the experiments corresponding to
ieved through a spiral wound module of ten-cm diameter and one
the aroma recovery task carried out at lab scale. In all cases, the feed
hundred-cm length with 13 m2 of membrane area [24]. In order to
vessel is filled with 500 g of a lager beer and pervaporation with Per­
minimize polarization of temperature and concentration, the retentate
vatech PDMS membrane is performed until collecting ca. 5 g of
flow rate is set to 8 m3/h with a pressure drop of 0.5 bar/m [24]. It is
permeate, which represents 1% of the feed. The membrane area is 28
observed that the lower the operation temperature of the first condenser
cm2 and the feed flow rate, which corresponds to a tangential velocity of
(and hence, that of the second condenser), the greater the fractional
0.45 m/s and a Reynolds number of around 1500, allows considering the
recovery of the condensate in the first condenser and the lower the
polarization of both concentration and temperature as negligible.
marginal cost of the aroma recovery task. The higher operating costs of
Table 3 shows the operating conditions for each experiment,
the compressors for alternative B5 are offset by the lower investment
permeate and CO2 fluxes obtained, as well as the mass composition of
costs in the condensers.
ethanol in permeate. The values reported for the permeate fluxes
Focusing on design B5, a high fractional recovery of aromas is
represent average values for the complete aroma recovery task. Results
observed with the exception of ethyl acetate which is recovered at 47 %
for the model of the overall permeate flux as a function of both tem­
(2 % in the first condenser and 45 % in the second condenser). The total
perature and downstream pressure are also shown [26]. Regarding the
aroma recovery represents 21.3 % of the aroma content in the original
CO2 fluxes, they were estimated from measurements of CO2 concentra­
beer with a ratio of alcohols to esters of 6.71. The above performance
tion in retentate. While the CO2 concentration measured in the initial
measures change to 12.2% and 6.22, respectively, if the condensates are
beer was 2.9 g/kg, the CO2 concentration in the final retentate after
mixed at the ratio 0.3/1 and added together with an amount of fresh
obtaining a permeate equivalent to ca. 0.5% of the content in the feed
beer equivalent to 1 % of the batch size to the dealcoholized beer.
tank (i.e., 2.5 g of permeate) showed values below the lower limit of the
Table 6 shows the total and component mass balances for this
equipment (<1.5 g/kg). Thus, it was assumed that all the CO2 content in
alternative.
the fresh beer was permeated during this period. Values reported in
A detailed analysis of the results shown in Table 5 allows inferring
Table 3 for the average fluxes reflect the above assumption.
that the low performance of the condensing system in the recovery of
Fig. 4 depicts the agreement between experimental values and
ethyl acetate is closely linked to the joint influence of the inerts (espe­
calculated ones for permeate flux as a function of main operating vari­
cially carbon dioxide) and the freezing point curve of the water-ethanol
ables. Model parameters (PWO /l) and EapW shown in Table 3 were both

Table 4
Aroma concentrations, % recovery and concentration ratio between higher alcohols and esters in the permeate corresponding to aroma recovery step with Pervatech
PDMS membrane.
Run # Ethyl acetate Propanol Isobutanol Isoamyl alcohol

C in ppm %recovery C in ppm %recovery C in ppm %recovery C in ppm %recovery

A 232 21.0 55 6.9 112 12.3 885 15.2


B 356 32.2 61 7.6 157 17.3 1230 21.2
C 290 26.2 61 7.6 128 14.1 1081 18.6
D 377 34.1 75 9.4 142 15.6 1150 19.8
Isoamyl acetate Acetaldehyde
Run # C in ppm %recovery C in ppm %recovery Higher alcohols/esters
A 29 22.1 53 15.5 4.0
B 48 36.6 71 20.8 3.6
C 38 29.0 79 23.2 3.9
D 52 39.7 120 35.2 3.2

7
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

Fig. 4. Experimental and calculated values for permeate flux as a function of temperature (K) and downstream pressure (kPa). R2 = 0.995.

Table 5
Simulation results of different feasible designs of the condensation task at industrial scale for the operating conditions corresponding to run # B. In all cases the
condensate amount represents one percent of the batch size. Alternative corresponding to two partial condensers in series.
Temperature (◦ C) B1 B2 B3 B4 B5

Cond. 1 Cond. 2 Cond. 1 Cond. 2 Cond. 1 Cond. 2 Cond. 1 Cond. 2 Cond. 1 Cond. 2
− 1 − 4 − 2 − 42 − 3 − 43 − 4 − 45 − 5 − 46

Fractional recovery of component


Ethanol 0.22 0.74 0.24 0.72 0.27 0.69 0.30 0.67 0.33 0.64
Water 0.81 0.18 0.83 0.16 0.85 0.15 0.86 0.13 0.88 0.12
Ethyl acetate 0.01 0.38 0.01 0.39 0.02 0.41 0.02 0.44 0.02 0.45
Isobutanol 0.32 0.68 0.37 0.63 0.41 0.58 0.46 0.53 0.51 0.48
Acetaldehyde 0.02 0.97 0.03 0.97 0.04 0.96 0.04 0.95 0.06 0.94
Isoamyl alcohol 0.10 0.88 0.13 0.86 0.16 0.83 0.20 0.80 0.24 0.76
Isoamyl acetate 0.54 0.46 0.60 0.40 0.66 0.34 0.71 0.29 0.75 0.25
Propanol 0.27 0.69 0.30 0.66 0.34 0.63 0.37 0.60 0.41 0.57
Condensate fractional recovery 0.66 0.33 0.68 0.39 0.70 0.29 0.72 0.27 0.74 0.25

Design variables
Membrane area (m2) 13 13 13 13 13
Condensers area (m2) 1.75 0.81 1.74 0.75 1.73 0.70 1.71 0.64 1.69 0.59
Heat exchanger (m2)** 0.3 0.3 0.3 0.3 0.3
Aroma concentrate tank (m3) 0.04 0.04 0.04 0.04 0.04

Operation variables
Operation time (h) 3.04 3.04 3.04 3.04 3.04
Vacuum pump electricity (kWh/batch) 13.3 13.3 13.2 13.0 13.0
Circulating pump electricity (kWh/batch) 1.04 1.04 1.04 1.04 1.04
Steam consumption (kg/batch) 29 29 29 29 29
Condensation duties (kWh/batch) 13.4 4.5 13.8 4.2 14.1 3.9 14.4 3.6 14.7 3.3
Compressors electricity (kWh/batch) 4.1 2.9 4.3 2.8 4.5 2.6 4.7 2.5 4.9 2.3

Overall Performance
A/E ratio 7.70 7.50 7.31 6.86 6.71
Aroma overall fractional recovery 0.209 0.210 0.211 0.212 0.213
Flow rates optimal combination 0 1 0 1 0.07 1 0.23 1 0.3 1
A/E ratio* 6.90 6.40 6.21 6.23 6.22
Aroma overall fractional recovery* 0.119 0.109 0.107 0.119 0.122
Cost per liter ($/l) 0.0944 0.0942 0.0939 0.0936 0.0933

* Final values obtained by adding to the alcohol-free beer the condensed permeate in the optimal combination of flow rates together with 30 kg of the lager beer. **The
same heat exchanger used for dealcoholization is used for compensating the temperature drop in the raw beer due to vaporization taking place in in pervaporation
module.

8
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

Table 6 Table 7
Mass balance corresponding to design B5. Flow rates are given in [Kg/h]. Simulation results of the best feasible designs of the condensation task at in­
Alternative corresponding to two partial condensers in series. dustrial scale for the operating conditions corresponding to runs # A-D. In all
Permeate Condens1 Vap1 Condens2 Vap2
cases the condensate amount represents one percent of the batch size. Alterna­
tive corresponding to two partial condensers in series.
Overall 17.480 7.292 10.188 2.464 7.725
CO2 7.504 0.000 7.504 0.000 7.504 A4 B5 D6
Air 0.123 0.000 0.123 0.000 0.123 Pervaporation 1.60 2.13 2.13
Ethanol 2.552 0.849 1.703 1.635 0.068 pressure (kPa)
Water 7.302 6.411 0.891 0.879 0.012 Pervaporation 40 40 30
Ethyl acetate 3.51E-03 7.27E-05 3.44E-03 1.58E-03 1.85E-03 temperature (◦ C)
Isobutanol 1.55E-03 7.97E-04 7.50E-04 7.49E-04 1.16E-06 Cond. Cond. Cond. Cond. Cond. Cond.
Acetaldehyde 7.00E-04 3.92E-05 6.60E-04 6.59E-04 1.31E-06 1 2 1 2 1 2
Isoamyl acetate 4.73E-04 1.13E-04 3.60E-04 3.58E-04 2.26E-06
Isoamyl 1.21E-02 9.14E-03 2.98E-03 2.98E-03 5.15E-07 Temperature (◦ C) − 2 − 38 − 5 − 46 − 5 − 46
alcohol Fractional recovery of component
Propanol 6.01E-04 2.45E-04 3.56E-04 3.43E-04 1.35E-05 Ethanol 0.153 0.772 0.333 0.641 0.328 0.645
Water 0.754 0.241 0.878 0.120 0.876 0.122
Ethyl acetate 0.007 0.284 0.021 0.451 0.020 0.448
system on the feasible condensers temperatures. Thus, the temperature Isobutanol 0.223 0.772 0.515 0.484 0.508 0.491
of the second condenser for alternative B5 (-46 ◦ C) is well above that Acetaldehyde 0.013 0.973 0.056 0.942 0.054 0.944
adopted when operating at temperatures below the freezing point curve Isoamyl alcohol 0.055 0.926 0.238 0.757 0.232 0.763
Isoamyl acetate 0.396 0.604 0.754 0.246 0.749 0.251
(typically below-75 ◦ C, [4]). However, even at the cost of lower overall
Propanol 0.192 0.745 0.407 0.571 0.402 0.575
aroma recovery, with the multiple condensers alternative it would be Condensate 0.60 0.38 0.74 0.25 0.73 0.26
possible to approximate the ratio of alcohols to esters of the original fractional
beer. recovery
Table 7 shows the simulations of the most economical designs cor­
responding to runs A-D. Alternative D6, which operates at relatively low Design variables
temperatures, require 26 m2 of membrane area (two modules in paral­ Membrane area 13 13 26
(m2)
lel) and a circulation pump of 16 m3/h, which affects the economic
Condensers area 1.97 1.14 1.69 0.59 1.81 0.64
performance of this alternative compared to that of designs A4 and B5. (m2)
Another relevant cost aspect is linked to the vacuum level. The variant Heat exchanger 0.3 0.3 0.3
A4, which operates at 1.6 kPa, shows higher electrical energy con­ (m2)**
sumption cost than the designs operating at 2.1 kPa. Ultimately, from an Aroma concentrate 0.04 0.04 0.04
tank (m3)
economic perspective, the ranking is B5, A4 and D6. From an aroma Operation variables
recovery perspective the ranking is D6, B5 and A4 since operation at Operation time (h) 2.80 3.04 2.82
moderate vacuum levels minimises the carry-over effect of the carbon Vacuum pump 16.8 13.0 13.1
dioxide. electricity (kWh/
batch)
The variants corresponding to runs A and C deserve some relevant
Circulating pump 0.96 1.04 0.96
considerations. While for run C no feasible operating conditions could be electricity (kWh/
found, alternative A4 showed not only a very poor aroma recovery batch)
compared to variants B5 and D6 but also a ratio of alcohols to esters far Steam consumption 29 29 29
away from that of the original beer. For these reasons, for runs A and C it (kg/batch)
Condensation 12.5 5.5 14.7 3.3 14.5 3.4
is advisable to adopt a condensation scheme operating at temperatures duties (kWh/
below the freezing point curve of the water-ethanol system. batch)
Note that in order to obtain a final A/E ratio of 6.22, design D6 re­ Compressors 3.9 3.4 4.9 2.3 4.8 2.4
sorts to an optimal flow combination of 0.43/1 against a combination of electricity (kWh/
batch)
0.3/1 of design B5. This fact triggers the better total aroma recovery of
variant D6 and makes it very attractive despite being a little more
expensive than B5. Table 8 shows the mass balance corresponding to Overall Performance
A/E ratio 10.87 6.71 5.98
design D6, which in turn is the variant selected as the optimal one with a Aroma overall 0.147 0.213 0.210
cost of 0.1070 USD/L. fractional
Finally, Table 9 shows the simulations of the designs corresponding recovery
to runs A-D corresponding to the alternative of two total condensers in Flow rates optimal 0 1 0.30 1 0.43 1
combination
parallel working out of phase. As mentioned above, total condensation is
A/E ratio* 11.15 6.22 6.22
the most convenient alternative for the aroma recovery task under the Aroma overall 0.099 0.122 0.141
operating conditions corresponding to the experimental runs A and C. fractional
When comparing designs B5 and D6 with those corresponding to the recovery*
total condensation alternative the following trends apply for overall Cost per liter ($/l) 0.1027 0.0933 0.1070

fractional recovery, A/E ratio and cost per liter. While it is expected a * Final values obtained by adding to the alcohol-free beer the condensed
better performance of the total condensation variant in terms of overall permeate in the optimal combination of flow rates together with 30 kg of the
fractional recovery, the partial condensation alternative would be better lager beer. **The same heat exchanger used for dealcoholization is used for
in both economic and final quality terms as the latter operates in milder compensating the temperature drop in the raw beer due to vaporization taking
conditions than the total condensation variant and enables an adjust­ place in in pervaporation module.
ment of the A/E ratio. On the other hand, for operation at high vacuum
levels, total condensation becomes the preferred alternative. We select
variant A of Table 9 as the more attractive alternative in terms of eco­
nomics (0.181 USD/L), overall fractional recovery (14%) and higher
alcohols to esters ratio (6.1).

9
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

Table 8 5. Conclusions
Mass balance corresponding to design D6. Flow rates are given in [Kg/h].
Alternative corresponding to two partial condensers in series. As pointed out in the review paper by Skiborowski et al. [31], hybrid
Permeate Condens1 Vap1 Condens2 Vap2 processes based on the combination of pervaporation and distillation (in
Overall 18.991 7.814 11.177 2.740 8.437
our case Partial or Total Condensation) requires a conceptual model for
CO2 8.172 0.000 8.172 0.000 8.172 the membrane unit relying on experiments at lab scale in order to ach­
Air 0.177 0.000 0.177 0.000 0.177 ieve semi-empirical description of the physics governing the membrane
Ethanol 2.754 0.903 1.851 1.777 0.074 flux behavior. In this sense, in this work a model was obtained that
Water 7.888 6.911 0.977 0.963 0.014
adequately characterized the ethanol and overall fluxes through the
Ethyl acetate 4.01E-03 8.12E-05 3.93E-03 1.80E-03 2.13E-03
Isobutanol 1.51E-03 7.68E-04 7.43E-04 7.42E-04 1.16E-06 Pervatech PDMS membrane as a function of temperature and permeate
Acetaldehyde 1.28E-03 6.91E-05 1.20E-03 1.21E-03 2.45E-06 pressure. This model allowed estimating the membrane area re­
Isoamyl acetate 5.53E-04 1.28E-04 4.25E-04 4.22E-04 2.70E-06 quirements for different operating conditions.
Isoamyl 1.22E-02 9.16E-03 3.08E-03 3.08E-03 5.38E-07 Partial or total condensation of the permeate stream, on the other
alcohol
Propanol 7.98E-04 3.21E-04 4.77E-04 4.59E-04 1.82E-05
hand, can be properly modeled as either a flash unit or a total condenser
under vacuum conditions and presence of CO2 and air, respectively. In
both cases, the freezing point curve corresponding to the mixture
ethanol-water must be considered as a constraint.
Table 9
This manuscript explored, from simulations, the possibility of using
Simulation results of designs of the condensation task at industrial scale for the
two partial condensers to approximate the A/E ratio of a lager beer.
operating conditions corresponding to runs # A-D. In all cases the condensate
amount represents one percent of the batch size. Alternative corresponding to From an economic point of view, it was concluded that operating tem­
two total condensers in parallel working out of phase. peratures in the range of [30 ◦ C–40 ◦ C] give rise to designs that would be
feasible to implement on an industrial scale by minimising the mem­
A B C D
brane area requirement. On the other hand, it was found that the vac­
Design variables uum level to be adopted plays an essential role in the recovery of aroma
Membrane area (m2) 13 13 26 26
components in the presence of carbon dioxide. Compared to designs at
Condenser area (m2)** 1.27 1.14 1.52 1.23
Heat exchanger (m2)*** 0.3 0.3 0.3 0.3 moderate vacuum levels, alternatives at relativly low permeate pres­
Aroma concentrate tank (m3) 0.04 0.04 0.04 0.04 sures showed a reduction in the operating window of the process in
terms of the feasible operating temperatures of the condensers, trig­
Operation variables gering a decrease in fractional recovery of aroma compounds and
Operation time (h) 2.8 3.1 2.4 2.8 compromising the feasibility of the process.
Vacuum pump electricity (kWh/batch) 14.1 11.8 18.4 11.7 Results achieved for the case of total condensation with a duplicate
Circulating pump electricity (kWh/batch) 1.0 1.1 1.3 1.6
total condenser allowing continuous operation [4] indicate that, if
Steam consumption (kg/batch) 29 29 29 29
Condensation duties (kWh/batch) 20.2 20.4 20.1 19.9 properly designed, the A/E ratio obtained at lab scale level would
coincide with that achieved after total condensation and thus, the
Overall Performance
adjustability of the ratio of alcohols to esters would decrease with
A/E ratio* 6.10 5.01 5.57 4.56 respect to that achieved when partial condensers are used. Although the
Aroma fractional recovery* 0.140 0.198 0.174 0.198 use of a total condenser results in more expensive designs, this alter­
Cost per liter ($/l) 0.181 0.175 0.204 0.188 native is very advantageous in terms of overall aroma recovery and
* Final values obtained by adding to the alcohol-free beer the condensed process feasibility.
permeate together with 30 kg of the lager beer. ** Shell and tube design is From the results obtained, the importance of developing conceptual
selected instead of double pipe heat exchanger. ***The same heat exchanger used models that make use of both experimental runs of the aroma recovery
for dealcoholization is used for compensating the temperature drop in the raw process at laboratory scale and simulations of the condensation process
beer due to vaporization taking place in in pervaporation modul. from parameters corresponding to the vapour-liquid equilibrium of the
multi-component system can be deduced. The main weakness of the
By the way of summary of results achieved in this and the previous approach proposed in this manuscript, which resulted in very conser­
manuscript [6], Fig. 5 shows the cost breakdown of the overall processes vative designs, was the impossibility to perform on-line measurements
depicted in Fig. 2 analyzed at their respective optimal points. Besides the of the CO2 flow rate. This restriction can be overcome in pilot plant tests
costs corresponding to the aroma recovery task through pervaporation where it is also possible to include the condensers. Finally, it is in that
followed by: (a) two partial condensers in series; (b) two total con­ instance where not only the laboratory-scale results could be tested, but
densers in parallel, costs of the hybrid process composed of a membrane also a refinement of the optimum could be obtained using the models
unit and a distillation unit are also included in Fig. 5. developed at laboratory scale resorting to the so called modeling for
The cost of the hybrid process of dealcoholization of lager beer optimization approach [32]). Experimental results at pilot scale would
through a nanofiltration stage with valorization of the permeate via serve not only to update the model parameters but also to find the actual
distillation represents an annual cost of 237,300 USD/year with a ratio optimum by minimising the number of additional runs.
of investment costs to operating costs of 51.5/48.5. The credit for
permeate valorization reduces the cost of the process to 206,000 USD/ Funding
year.
While the aroma recovery stage with partial condensation, with a This work was made in the framework of project PIP 779 “Concep­
cost ratio of 48.6/51.4, represents 24.5% of the cost of the hybrid pro­ tual design of membrane-based hybrid processes” with the financial
cess, the cost of the variant with full condensation (55.1/48.5) amounts support from CONICET (Argentina).
to 35.4% of the cost of the hybrid formed by a nanofiltration unit fol­
lowed by a distillation unit. Figures above, corresponding to the percent CRediT authorship contribution statement
participation of the aroma recovery variants (a) and (b) change to 27.2%
and 38.7%, respectively, when the credit due to permeate valorization Danilo Alexander Figueroa Paredes: Investigation, Formal anal­
through distillation is taken into account. ysis, Methodology. Ramiro Julián Sánchez: Investigation, Formal
analysis. Micaela Magariño: Investigation. José Espinosa:

10
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

Fig. 5. Cost comparison corresponding to the whole dealcoholization process alternatives depicted in Fig. 2. (a) two partial condensers in series; (b) two total
condensers in parallel.

Conceptualization, Methodology, Investigation, Supervision, Funding interests or personal relationships that could have appeared to influence
acquisition. the work reported in this paper.

Declaration of Competing Interest

The authors declare that they have no known competing financial

Appendix

A1. Property methods

A1.1. Saturated vapor pressure


Antoine equation and Coefficients [27].
B
ln(VP) = A + + D.ln(T) + E.T F
T +C
Where:
VP: Vapour pressure (kPa)
T: Temperature (K). Antoine coefficients are shown in Table A1.

Table A1
Antoine coefficients.
Component Antoine coefficients

A B C D E F

Ethanol 86.486 − 7931.1 0 − 10.2498 6.39E-6 2


Water 65.928 − 7227.5 0 − 7.17695 4.03E-6 2
Isoamyl alcohol 14.697 − 3026.4 − 104.10 0 0 2
Acetaldehyde 37.270 − 4073.0 0 − 3.3618 3.81E-6 2
Ethyl acetate 88.370 − 7147.9 0 − 10.9917 8.55E-6 2
Isoamyl acetate 75.298 − 7535.7 0 − 8.87813 5.69E-6 2
Isobutanol 14.856 − 2874.7 − 100.30 0 0 2
Propanol 86.486 − 7931.1 0 − 10.2498 6.39E-6 2

11
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

A1.2. Activity coefficients


Wilson equation, binary interaction parameters and Wilson molar volumes [27].
[ ]
∑n ∑n
xA
lnγi = 1.0 − ln xj Aij − ∑n k ki
j=1 k=1 j=1 xj Akj

Where:
γ i = activity coefficient of component i.
[ ]
a
Vj RTij
Aij = e
Vi
xi = mole fraction of component i.
T = temperature (K).
aij = non-temperature dependent energy parameter between components i and j (cal/gmol.K).
Vi = molar volume of pure liquid component i (m3/kmol). Interaction parameters and Wilson molar volumes are shown in Table A2.

Table A2
Interaction parameters and Wilson molar volumes.
Subsystem Ethanol-Water

aij Ethanol Water Vi

Ethanol 0 276.75570 0.058492


Water 956.2974 0 0.018069
Subsystem Isoamyl alcohol-Ethanol-Water
aij Isoamyl alcohol Ethanol Water Vi
Isoamyl alcohol 0 39.5288 1955.406 0.109300
Ethanol 163.9351 0 276.7557 0.058492
Water 1299.864 975.4859 0 0.018069
Subsystem Acetaldehyde -Ethanol-Water
aij Acetaldehyde Ethanol Water Vi
Acetaldehyde 0 1337.281 285.5863 0.056600
Ethanol − 1161.145 0 276.7557 0.058492
Water 1045.566 975.486 0 0.018069
Subsystem Ethyl acetate -Ethanol-Water
aij Ethyl acetate Ethanol Water Vi
Ethyl acetate 0 297.58 1147.4270 0.098948
Ethanol 448.749 0 276.7557 0.058492
Water 2578.240 956.297 0 0.018069
Subsystem Isoamyl acetate -Ethanol-Water
aij Isoamyl acetate Ethanol Water Vi
Isoamyl acetate 0 0 3410.8110 0.150200
Ethanol 0 0 276.7557 0.058492
Water 2622.590 975.486 0 0.018069
Subsystem Isobutanol -Ethanol-Water
aij Isobutanol Ethanol Water Vi
Isobutanol 0 165.384 1272.149 0.075100
Ethanol 76.8889 0 276.7557 0.058492
Water 1777.244 975.486 0 0.018069
Subsystem Propanol -Ethanol-Water
aij Propanol Ethanol Water Vi
Propanol 0 − 40.8745 1128.1870 0.0769
Ethanol 102.6706 0 276.7557 0.058492
Water 1338.76 975.486 0 0.018069

A2. Heat exchanger dimensioning

For heat exchangers, the following dimensioning equation is used:


Q(K)
A(m2 ) = ( )
U mW2 K ΔTlm (K)

with typical values for the overall heat-transfer coefficients U adopted from Table A.7-1 in Douglas [33].

12
D.A. Figueroa Paredes et al. Separation and Purification Technology 290 (2022) 120821

References [16] M.A.J. Huijbregts, Z.J.N. Steinmann, P.M.F. Elshout, G. Stam, F. Verones, M.D.
M. Vieira, A. Hollander, M. Zijp, R. van Zelm, ReCiPe 2016: A harmonized life cycle
impact assessment method at midpoint and endpoint level, Int. J. Life Cycle Assess.
[1] A.I. Paz, C.A. Blanco, C. Andrés-Iglesias, L. Palacio, P. Prádanos, A. Hernández,
22 (2017) 138–147, https://doi.org/10.1007/s11367-016-1246-y.
Aroma recovery of beer flavors by pervaporation through polydimethylsiloxane
[17] S. Buiatti, Beer Composition: An Overview, in: V.R. Preedy (Ed.), Beer in Health
membranes, J. Food Process Eng. 40 (6) (2017) e12556, https://doi.org/10.1111/
and Disease Prevention, Academic Press Pubs, San Diego, 2009, pp. 211–225.
jfpe.2017.40.issue-610.1111/jfpe.12556.
[18] M. Catarino, A. Mendes, Dealcoholizing wine by membrane separation processes,
[2] M. Catarino, A. Ferreira, A. Mendes, Study and optimization of aroma recovery
Innovative Food Sci. Emerg. Technol. 12 (2011) 330–337, https://doi.org/
from beer by pervaporation, J. Membr. Sci. 341 (1-2) (2009) 51–59.
10.1016/j.ifset.2011.03.006.
[3] A. del Olmo, C.A. Blanco, L. Palacio, P. Prádanos, A. Hernández, Pervaporation
[19] G.J. Pickering, Low- and reduced-alcohol wine: A review, J. Wine Res. 11 (2)
methodology for improving alcohol-free beer quality through aroma recovery,
(2000) 129–144, https://doi.org/10.1080/09571260020001575.
J. Food Eng. 133 (2014) 1–8, https://doi.org/10.1016/j.jfoodeng.2014.02.014.
[20] M. Margallo, R. Aldaco, A. Barceló, N. Diban, I. Ortiz, A. Irabien, Life cycle
[4] M. Catarino, A. Mendes, Non-alcoholic beer—A new industrial process, Sep. Purif.
assessment of technologies for partial dealcoholisation of wines, Sustain. Prod.
Technol. 79 (3) (2011) 342–351, https://doi.org/10.1016/j.seppur.2011.03.020.
Consump. 2 (2015) 29–39, https://doi.org/10.1016/j.spc.2015.07.007.
[5] T. Brányik, D.P. Silva, M. Baszczyňski, R. Lehnert, J.B. Almeida e Silva, A review of
[21] C. Brazinha, J.G. Crespo, Aroma recovery from hydro alcoholic solutions by
methods of low alcohol and alcohol-free beer production, J. Food Eng. 108 (4)
organophilic pervaporation: Modelling of fractionation by condensation, J. Membr.
(2012) 493–506, https://doi.org/10.1016/j.jfoodeng.2011.09.020.
Sci. 341 (2009) 109–121, https://doi.org/10.1016/j.memsci.2009.05.045.
[6] R.J. Sánchez, D.A. Figueroa Paredes, D.S. Laoretani, Y. Villada, M. Fuentes,
[22] D.M. Aguilar-Valencia, M.A. Gómez-García, J. Fontalvo, Effect of pH, CO2, and
J. Espinosa, On the Conceptual Design of the Hybrid Nanofiltration/Distillation
High Glucose Concentrations on Polydimethylsiloxane Pervaporation Membranes
Process in the production of alcohol-free Beers, Sep. Purif. Technol. 267 (2021),
for Ethanol Removal, Ind. Eng. Chem. Res. 51 (2012) 9328–9334, https://doi.org/
https://doi.org/10.1016/j.seppur.2021.118625.
10.1021/ie3002765.
[7] S. Avlonitis, W.T. Hanbury, M.B. Boudinar, Spiral wound modules performance an
[23] A. Baudot, M. Marin, Improved Recovery of an Ester Flavor Compound by
analytical solution: Part II, Desalination 89 (3) (1993) 227–246, https://doi.org/
Pervaporation Coupled with a Flash Condensation, Ind. Eng. Chem. Res. 38 (1999)
10.1016/0011-9164(93)80139-E.
4458–4469.
[8] G. Schock, A. Miquel, Mass transfer and pressure loss in spiral wound modules,
[24] X. Sun, G. Dang, X. Ding, C. Shen, G. Liu, C. Zuo, X. Chen, W. Xing, W. Jin,
Desalination 64 (1987) 339–352, https://doi.org/10.1016/0011-9164(87)90107-
Production of alcohol-free wine and grape spirit by pervaporation membrane
X.
technology, Food Bioprod. Process. 123 (September 2020) 262–273, https://doi.
[9] J. Espinosa, E. Salomone, Minimum reflux for batch distillations of ideal and
org/10.1016/j.fbp.2020.07.006.
nonideal mixtures at constant reflux, Ind. Eng. Chem. Res. 38 (7) (1999)
[25] S.V. Sumbhate, S. Nayak, D. Goupale, A. Tiwari, R.S. Jadon, Colorimetric Method
2732–2746, https://doi.org/10.1021/ie990022y.
for the Estimation of Ethanol in Alcoholic-Drinks, J. Anal. Tech. 1 (2012) 1.
[10] M. Skiborowski, S. Recker, W. Marquardt, Shortcut-based optimization of
[26] A. Kujawska, K. Knozowska, J. Kujawa, W. Kujawski, Influence of downstream
distillation-based processes by a novel reformulation of the feed angle method,
pressure on pervaporation properties of PDMS and POMS based membranes, Sep.
Chem. Eng. Res. Des. 132 (2018) 135–148, https://doi.org/10.1016/j.
Purif. Technol. 159 (2016) 68–80, https://doi.org/10.1016/j.seppur.2015.12.057.
cherd.2018.01.019.
[27] Hysys User Manual; Hyprotech Ltd.: Calgary, Canada, 1999.
[11] E. Salomone, J. Espinosa, Prediction of homogeneous azeotropes with interval
[28] W. Ernest, in: Flick: Industrial Solvents Handbook, fifth ed., Noyes Data
analysis techniques exploiting topological considerations, Ind. Eng. Chem. Res. 40
Corporation (ndc), Westwood, NJ/USA, 1998, p. S. 252. ISBN 0-8155-1413-1.
(6) (2001) 1580–1588, https://doi.org/10.1021/ie000608g.
[29] A. Fahmy, D. Mewes, K. Ebert, Design methodology for the optimization of
[12] M. Skiborowski, J. Bausa, W. Marquardt, A unifying approach for the calculation of
membrane separation properties for hybrid vapor permeation-distillation
azeotropes and pinch points in homogeneous and heterogeneous mixtures, Ind.
processes, Sep. Sci. Technol. 36 (15) (2001) 3287–3304, https://doi.org/10.1081/
Eng. Chem. Res. 55 (24) (2016) 6815–6834, https://doi.org/10.1021/acs.
SS-100107903.
iecr.6b01303.
[30] G.D. Ulrich, P.T. Vasudevan, How to estimate utility costs, Chem. Eng. 113 (4)
[13] M. Diehl, A. Schaefer, H.G. Bock, J.P. Schloeder, Optimization of Multiple-Fraction
(2006) 66–69.
Batch Distillation with Recycled Waste Cuts, AIChE J. 48 (12) (2002) 2869–2874,
[31] M. Skiborowski, A. Harwardt, W. Marquardt, Conceptual design of distillation
https://doi.org/10.1002/aic.690481214.
based hybrid separation processes, Annual Reviews Chemical, Biomol. Eng. 4
[14] M.C. Mussati, P. Aguirre, J. Espinosa, O. Iribarren, Optimal Design of Azeotropic
(2013) 45–68, https://doi.org/10.1146/annurev-chembioeng-061010-114129.
Batch Distillation, AIChE J. 52 (3) (2006) 968–985, https://doi.org/10.1002/
[32] E.C. Martínez, M.D. Cristaldi, R.J. Grau, Design of Dynamic Experiments in
aic.10696.
Modeling for Optimization of Batch Processes, Ind. Eng. Chem. Res. 48 (2009)
[15] A.W. Sleeswijk, L.F. van Oers, J.B. Guinée, J. Struijs, M.A. Huijbregts,
3453–3465, https://doi.org/10.1021/ie8000953.
Normalisation in product life cycle assessment: An LCA of the global and European
[33] J.M. Douglas. Conceptual Design of Chemical Processes, McGraw-Hill, New York,
economic systems in the year 2000, Sci. Total Environ. 390 (1) (2008) 227–240,
1988.
https://doi.org/10.1016/j.scitotenv.2007.09.040.

13

You might also like