Separation and Purification Technology 335 (2024) 126076

Contents lists available at ScienceDirect

Separation and Purification Technology

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

Partial dealcoholization of a Malbec wine through pervaporation with a

PDMS membrane: Effect of operation temperature on process economics,
volatile aroma composition and sensory properties
Danilo Alexander Figueroa Paredes a, 1, Mariela Assof b, 1, Ramiro Julián Sánchez a, c,
Santiago Sari d, Aníbal Catania d, Martín Fanzone b, José Espinosa a, c, *
a
INGAR (CONICET-UTN), Avellaneda 3657, S3002GJC Santa Fe, Argentina
b
Laboratorio de Aromas y Sustancias Naturales, Estación Experimental Agropecuaria Mendoza, Instituto Nacional de Tecnología Agropecuaria (INTA), San Martín
3853, 5507 Luján de Cuyo, Mendoza, Argentina
c
Universidad Nacional del Litoral, Facultad de Ingeniería Química, Santiago del Estero 2829, 3000 Santa Fe, Argentina
d
Centro de Estudios de Enología, Estación Experimental Agropecuaria Mendoza, Instituto Nacional de Tecnología Agropecuaria (INTA), San Martín 3853, 5507 Luján
de Cuyo, Mendoza, Argentina

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

Editor: B. Van der Bruggen In the present work, partial dealcoholization of a Malbec red wine with 14.8% v/v alcohol by pervaporation
through a PDMS membrane (polydimethylsiloxane) was addressed. The influence of the process temperature on
Keywords: the economy of the process, the composition of volatile aroma compounds and wine sensory characteristics was
Wine dealcoholization deeply analyzed.
Pervaporation
To do so, partial dealcoholization experiments at lab scale were run with the pervaporation setup operated at a
Conceptual modelling
permeate pressure of 0.013 bar and at three different temperatures; i.e. 35, 40 and 45 ◦ C. Two types of runs were
Scaling-up
Economics carried out: (i) simple runs to build a conceptual model of the process and (ii) repetitions to accumulate an
Volatile aroma profile adequate volume of partially dealcoholized wine (10.5% v/v) to allow the sensory analysis of the following
Sensory analysis Malbec wines: the initial wine, the partially dealcoholized wines and the blends in a 70/30 ratio (13.5% v/v)).
From the conceptual modeling stage the required value of the fractional pervaporation advance ηreq of 0.09
was obtained, as well as the average values of permeate overall fluxes at 35, 40 and 45 ◦ C; i.e., 1.02, 1.38 and
1.82 kg m− 2 h− 1, respectively. For all variants, a permeate product with an alcohol content of 54% v/v was
estimated. Experimental values agreed reasonably well with those obtained from the model simulation.
The conceptual modeling of the process enabled the scaling-up of the process and the subsequent economic
analysis. The best figures corresponded to the design with heat integration at 45 ◦ C when earnings estimated
from the sale of the ethanol-rich permeate to the local market was taken into account.
Fresh Malbec, partially dealcoholized wines at the selected temperatures, and their corresponding blends were
chemical and sensory characterized. Dealcoholized wines showed differences concerning the operating tem­
perature on the volatile and total acidity, fructose, anthocyanins, tannins, total polymers, and CIELAB param­
eters, except lightness. The total and volatile acidity were affected by the temperature and showed an increase in
dealcoholized and blended wines with the respective decrease of the pH. Dealcoholization favored the presence
of large polymeric pigments (0.6 in initial wines and an average of 0.96 in blends), which are associated posi­
tively with color stability.
Volatiles were also affected by the operating temperature, decreasing the concentration of all chemical groups
in these wines. A total of thirty-two free volatile organic compounds (VOCs) including higher alcohols (HA),
medium chain fatty acid ethyl esters (MCFAEE), higher alcohol acetates (HAA), medium chain fatty acids
(MCFA), norisoprenoids (NI) and terpenoids (TP) were identified and quantified.
Finally, sensory evaluation demonstrated that partially dealcoholized wines could not be differentiated by the
operation temperature. The fullness, astringency, pungency, floral notes, bitterness, color intensity, and sweet­
ness were significant variables that allowed differentiate between initial, partially dealcoholized, and blend

* Corresponding author at: INGAR (CONICET-UTN), Avellaneda 3657, S3002GJC Santa Fe, Argentina.
E-mail address: [email protected] (J. Espinosa).
1
Danilo A. Figueroa Paredes and Mariela Assof contributed equally to this work.

https://doi.org/10.1016/j.seppur.2023.126076
Received 5 October 2023; Received in revised form 13 December 2023; Accepted 14 December 2023
Available online 28 December 2023
1383-5866/© 2023 Elsevier B.V. All rights reserved.
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

wines. Despite the loss of volatile compounds, blended wines showed higher intensity of floral attributes, and in
mouth were lesser astringent, bitter, and pungent than the fresh wine.
From an overall perspective, pervaporation of fresh wine with a PDMS membrane at 45 ◦ C and the adoption of
a dilution ratio of 70/30 for initial wine to partially dealcoholized wine result in a promising strategy to reduce
the alcohol content without affecting the sensory quality of the final red wine.

Nomenclature yEt ethanol molar fraction in permeate

yW water molar fraction in permeate
Abbreviations for Section 3.5.2 local flux model
A membrane Area (m2) Abbreviations for Section 3.5.3 batch pervaporation model from
jEt molar flux of ethanol (mol m− 2 h− 1) experiments at lab scale
jW molar flux of water (mol m− 2 h− 1) M mass in feed vessel
J overall mass flux (g m− 2 h− 1) M0 , Mt initial and integrated mass in feed vessel
MWEt molar mass of ethanol (g/mol) J instantaneous overall flux (g m− 2 h− 1)
MWW molar mass of water (g/mol) JCum cumulative overall flux (g m− 2 h− 1)
MWS molar mass of non-volatile solids (g/mol) XEt instantaneous ethanol mass fraction in feed tank
XEt Ethanol mass fraction in feed
XEt0, XEtt initial and integrated ethanol mass fraction in feed tank
XW water mass fraction in feed
XS instantaneous solid mass fraction in feed tank
XS non-volatile solids mass fraction in feed
XS0 , XSt initial and integrated solid mass fraction in feed tank
xEt ethanol molar fraction in feed
YEt instantaneous ethanol mass fraction in permeate
xW water molar fraction in feed
YEtCum cumulative ethanol mass fraction in permeate
xS non-volatile solids molar fraction in feed
PerCum cumulative permeate
YEt ethanol mass fraction in permeate
η pervaporation fractional advance
YW water mass fraction in permeate

of 13.11% v/v. For the scaling-up of the process, the authors propose a
conceptual model that assumes that fluxes obtained for the total
1. Introduction extraction of ethanol (i.e., final alcohol content xRE < 0.5% v/v) can be
applied to partial dealcoholization once the operation temperature is
Without claiming to be an exhaustive analysis in the field of wine selected. While this assumption is adequate for final alcohol composi­
dealcoholization, this introduction reviews the most significant contri­ tions in the vicinity of 0.5% v/v, it can lead to errors in the estimation of
butions in the area, which are also relevant to highlight the scope of the the membrane area requirement. For such cases it would be more
present work and to adopt values for the operating conditions of the convenient to develop a model with parameters that depend on the
proposed process. desired degree of concentration.
Brueschke et al. [3] present several examples of partial deal­
1.1. Dealcoholization using pervaporation membranes coholization of wines and beers using pervaporation technology. Ex­
amples 1–6 in their invention refer to partial dealcoholization of wines
In their work, Sun et al. [1] address the production of alcohol-free with the process operating at retentate temperatures selected in the
wine and grape spirit from a Cabernet Sauvignon red wine resorting to range between 30 ◦ C and 45 ◦ C and permeate pressures between 15
two pervaporation stages with a PDMS (polydimethylsiloxane) mbar and 20 mbar. For the permeate condensation stage they propose a
membrane. multiple condenser system with the last one placed after the vacuum
Continuous operation of the dealcoholization task is realized through pump.
the use of three feed tanks of 3 m3 each, a circulating pump and twelve
membrane modules with an overall area of 156 m2. The operating 1.2. Dealcoholization using either nanofiltration or reverse osmosis
conditions selected for the first stage were 45 ◦ C for feed temperature, 5 membranes
kPa for permeate pressure and 48 m3/h for the feed flow rate through an
arrangement of six modules operating in parallel with two modules in Catarino and Mendes [4] studied the partial dealcoholization of a red
series each. Complete dealcoholization of the red wine was achieved in wine with 12% v/v alcohol using different nanofiltration and reverse
10 h with an average permeate flux of 745 gm-2h− 1. The permeate, osmosis membranes for dealcoholization up to a content of approxi­
whose average alcohol composition reached 28.2% v/v, was subjected mately 7–8% v/v ethanol in a single step process by operating the
to a second pervaporation stage in order to achieve a product with an experimental equipment in diafiltration mode; i.e., with water addition.
alcohol composition of around 50% v/v after 10 h of operation. The Selected operating temperature and pressure were 30 ◦ C and 16 bar,
average permeate flux, significantly higher than that of the first stage, respectively. The performance of each membrane was first assessed by
was 1328 g m− 2 h− 1. For this stage, the authors considered the use of estimating both the average flux required to perform the separation and
two feed tanks of 1 m3 each and four pervaporation modules (2 × 2 the rejection coefficient for ethanol. Subsequently, each membrane was
arrangement). Among their contributions, the authors also performed a characterized according to the physicochemical and organoleptic
comparison between fluxes and compositions achieved from the red properties of the corresponding dealcoholized wine. High flux mem­
wine and from an ethanol–water mixture of similar composition. branes were also used for producing reconstituted wine samples based
Resorting to laboratory scale experiments with a PDMS membrane at on a deep dealcoholization process until achieving an intermediate
temperatures in the interval [40-70 ◦ C], Takács et al. [2] developed an product with ca. 5% v/v of alcohol for subsequent blending with fresh
Arrhenius-type model for both the overall and ethanol fluxes describing wine to obtain a final product with 7% v/v of ethanol. Finally, the
the total dealcoholization of a Hungarian wine with an alcohol content addition of an aroma-rich concentrate to the partially dealcoholized

2
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

wines, obtained the former through pervaporation of the fresh wine at membrane units and a fermenter to produce low alcohol white wines.
the very beginning of the process, was also taken into account in the First, the authors tested a two-step NF process for the reduction of must
assessment process. Excellent results for aroma-enriched reconstituted sugars. Next, they used PV with a PDMS membrane for the recovery of
wines were obtained. However, as producing reconstituted wines is aroma precursors from grape must, followed by two-stage NF and
more time consuming than single step processes, non-desired aroma restitution of the flavor precursors to the filtered must. Duplicate fer­
sensations can appear. mentations were performed from control must and filtered must blends.
Fig. 3 of the manuscript appropriately explains the variants considered.
1.3. Evaporative pertraction in partial dealcoholization of wines Sensory analysis of the wines produced showed that consumers would
find no major differences between the wines from the PV-NF hybrid and
In Diban et al. [5], experiments of partial dealcoholization of three the control wines. Furthermore, analysis of the aroma compounds
wine varieties through evaporative pertraction (EP) were performed showed that in comparison to the control wine, wines coming from
using an industrial scale hollow fiber contactor, provided with a mem­ pervaporation experiments had a more similar aromatic profile than the
brane area of 20 m2. It was observed experimentally that the modifi­ ones coming only from the nanofiltration tests.
cation of the working variables feed and stripping flow rates, pH of
stripping phase (water) and feed to stripping volume ratio have a deep 1.5. Dealcoholization via sugar control in musts
influence on both ethanol reduction and aroma compounds losses. From
the pilot-scale experiments, the authors built a model which was able to García-Martín et al. [9] studied the reduction of sugar in the grape
properly describe the mass fluxes through the membrane of ethanol and must to obtain wines with two percent points reduction in their alcoholic
the most abundant aroma compounds in the studied wine varieties. For degree. The sugar reduction of sugar was accomplished by performing
the scaling-up of a process aiming to reduce the alcohol content of a two successive nanofiltration steps. To this end, the authors worked with
Garnacha wine variety in two percent points, the authors proposed two types of musts: one from the Verdejo variety of white grapes and the
quasi-optimal values for the main working variables that would allow other from red grapes of the Tinta de Toro variety. Each must was fer­
minimizing the losses of the aroma compounds. mented both after treatment and, to be used as control, without any
filtration in order to check the effectiveness of the process. Once
1.4. Partial dealcoholization of wines through hybrid technologies fermentation is completed, wide-ranging analysis was used to study all
possible changes in the characteristics of the wine from a chemical point
The work of Duc-Truc et al. [6], applied to five Cabernet Sauvignon of view. The alcohol reduction reached by the wines obtained after
wines, delves into the influence of a hybrid reverse osmosis-evaporative nanofiltration and mixing of both white and red musts was satisfactory.
pertraction partial dealcoholization process on both composition
(Table 3) and sensory properties (Table 1). Since the alcohol reduction 1.6. LCA in partial dealcoholization of wines
was small (between 1.3 and 2.5% v/v), the authors found no significant
changes in the intensity of aromas and flavors. However, significant According to Margallo et al. [10], LCA has been proven to be an
differences were observed in the sensory perception of dealcoholized effective mechanism for reporting environmental performance in the
wine, which, for the authors, strongly depended on the alcohol con­ food and beverage sector in general, and in the viticulture and vinifi­
centration. Fig. 3(c) of the aforementioned manuscript shows an opti­ cation sector specifically. In order to cover a gap in the environmental
mum composition for wine A at an alcohol concentration of 16.6% v/v. assessment on wine dealcoholization, the authors evaluate different
Different blends of wine A (17% v/v) with dealcoholized wine (14.5% dealcoholization processes; namely, evaporative pertraction (EP), spin­
v/v) including the 50/50 blend were analyzed. ning cone column (SCC) and reverse osmosis (RO). The authors make
The research work of Longo et al. [7] provides a clear picture of the focus on the partial dealcoholization process consisting in a small
changes in both volatile component composition and sensory attributes adjustment in the alcohol content between 1% and 2% as this task is
of Verdelho and Petit Verdot wines produced from early and late harvest currently one of the most important objectives for the wine industry in
grapes and a dealcoholization process. Using a hybrid reverse osmosis- accordance with regulations from the European Commission (EC, 2009).
evaporative pertraction process, a significant loss of volatile com­ In the manuscript, energy and material valorization were evaluated
pounds, including esters and higher alcohols, was evidenced. The for the wastewater from the SCC alternative, which in turn is formed by
aforementioned authors found significant differences between the sen­ a two-stage distillation process in order to obtain a dealcoholized wine
sory attributes of early and late harvest wines. However, the deal­ with recovered aroma compounds. In the case of material valorization,
coholized wines were clearly different from the early and late harvest the wastewater from the SCC process is sent from the winery to an
wines, even though these wines had a similar alcohol level to the early alcohol industry where it is concentrated by means of distillation to
harvest wines. Finally, they highlighted the importance of the non- obtain a stream with greater ethanol content (96% v/v). In the energy
volatile matrix of the wine and of the dealcoholization level on the valorization scenario the effluent is concentrated off-site via distillation
membrane behavior in the evaporative pertraction unit, especially for to be used to produce energy. As a debatable aspect of the work, at least
the case of wines with high alcohol content. in our opinion, we can highlight the non-incorporation of a distillation
Salgado et al. [8] resorted to a hybrid process composed by column to the EP process both to minimize the water consumption of the
process and, perhaps more important, to obtain a process effluent rich in
Table 1 alcohol.
Design and operation variables for the process alternatives studied. In a recent manuscript, Figueroa Paredes et al. [11] analyzed the
Feed vessel volume (m3)
environmental performance of different dealcoholization technologies
Permeate vessel volume (m3) in the production of partially dealcoholized wines; namely, evaporative
Membrane area (m2) pertraction (EP), nanofiltration (NF), reverse osmosis (RO), pervapora­
Retentate heat exchanger area (m2) tion (PV) and the spinning cone column (SCC). In all cases, a reduction
Condenser area (m2)
in the alcohol content of two percent points in volume was considered to
Refrigerant condenser area (m2)
Recirculating pump electricity consumption (kWh/batch) compensate the diminution in the quality of wines with an excessive
Vacuum pump electricity consumption (kWh/batch) concentration of ethanol while satisfying the consumer demand and
Steam consumption(kg/batch) produce fresh and balanced wines with lower alcoholic strength. For the
Condensation duty (kWh/batch) cases of nanofiltration, reverse osmosis and pervaporation, the decrease
Cooling water (m3/batch)
of alcohol content by 2 percentage points was achieved by mixing fresh

3
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

wine with partially dealcoholized wine in a 50/50 ratio.

Since the first three technologies require water for their operation
(either as a stripping agent or in the diafiltration stage) and in order to
minimize their consumption, distillation was incorporated to treat the
respective effluents [12]. In this way, a hybrid process consisting in a
membrane unit coupled to a distillation unit was considered for these
technologies [13]. The alcohol-rich overhead products were subjected to
off-site material valorization. Details on the models from which the
environmental performance of each technology was determined can be
found in the referenced manuscript [11]. For the case of pervaporation
PV, the authors resorted to lab scale experiments at 35 ◦ C and 0.013 bar.
For this technology, two scenarios were taken into account; namely, a
conventional process PV and a process, PVHI, considering the heat
integration between retentate and permeate streams through a vapor
compression refrigeration system. Variants PV and PVHI produced high
ethanol content permeate streams. Thus, these streams were able to be
sent directly to off-site material valorization in a distillery located in the
neighborhood of the wine facility. Something similar occurred for the
SCC variant.
As results, in all cases both a partially dealcoholized wine and an
ethanol-rich effluent (50% v/v or more) were obtained with the last able
to be sent to a distillery for material valorization. By applying the ReCiPe
methodology [14,15] it was concluded that regardless of the end point
chosen, the PVHI variant showed the best environmental performance
and thus was the most environmentally friendly alternative followed by
variants SCC, PV, RO, NF and EP.

2. Problem statement

Argentina is an important wine-producing country in the southern

hemisphere, with 211,099 ha of vineyards representing ~3% of the
global wine grape cultivation area. Mendoza province has 70.6% of all
vineyards, with the wine industry being one of its main economic
development sources. The most cultivated variety in Mendoza is Malbec
(26.3%), which is considered an emblematic cultivar for winemaking. It
is followed in importance by Bonarda (9.9%), Cereza (9.1%), Criolla
Grande (8.3%) and Cabernet Sauvignon (7%). Malbec is the variety that
has increased the most in recent years. In Mendoza, it went from 14,338
ha in 2000 to the current 39,248 ha (+24,910 ha) [16].
Malbec wines are characterized by a high color intensity, with red or
markable violet/blue hues, soft tannins, astringent and slightly sweet
sensation in the mouth, and fresh or cocked red fruits and spicy in the
nose [17,18].
In the present work and applied to a Malbec red wine with 14.8% v/v
alcohol from the Luján de Cuyo region (Mendoza, Argentina), the in­
fluence of the operating temperature in the partial dealcoholization by
pervaporation through a commercial PDMS membrane (Pervatech BV,
The Netherlands) on the economy of the process, the composition of
volatile aroma compounds and the sensory properties will be analyzed.
Fig. 1 presents in detail the methodological framework adopted for this
work.
In order to assess the process economy (Step 4 of Fig. 1), conceptual
designs of a process consisting of a feed tank-pervaporation unit oper­
ating at temperatures of 35, 40 and 45 ◦ C [1–3] and at a permeate
pressure of 0.013 bar [3] will be carried out. The process scale-up will
consider the serial operation of two daily batches of 5000 L each for 12
h. Permeate streams with high ethanol content will be sold to a wine
distillery for their material valorization [11]. Fig. 2 depicts the sche­
matic diagram of a pervaporation process with heat integration between
retentate and permeate streams through a vapor compression refriger­
ation system. This variant was selected due to its outstanding environ­
Fig. 1. Methodological framework followed in this work.
mental performance when compared with other membrane alternatives
[11]. The main objective of reducing the alcohol content in 1.3 percent
points will be achieved by mixing fresh wine with partially deal­
coholized wine (ca. 10.5% v/v) in a 70:30 ratio. The choice of the value
for this parameter was decided on the basis of the promising results

4
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

Fig. 2. Schematic of the pervaporation process with heat integration (PVHI) for the variant at 45 ◦ C and 0.013 bar.

obtained from the sensory quality perspective for blends with alcohol Agropecuaria (INTA). Malbec grapes were crushed and destemmed
content reductions well below 2 percentage points [6]. (Metal Liniers model MTL 12, Mendoza, Argentina), and the grape juice
To carry out Step 4, it is required to obtain a conceptual model of the was supplemented with potassium metabisulfite (100 mg K2S2O5/kg),
process from experiments at lab scale with ethanol–water mixtures and yeast nutrient (Fermaid K®, Lallemand Inc., Canada), and inoculated
wine (Stages 1 and 2 of Fig. 1). The development of a local flux model with commercial Saccharomyces cerevisiae yeast strains. Daily controls
from experiments with binary ethanol–water mixtures will allow were made to establish the end of the alcoholic fermentation. Malbec
obtaining parameters that will then be used to build a model of the wine had completed the malolactic fermentation spontaneously and
dealcoholization process considering the wine as a pseudo-ternary finished wines were filtered, stabilized, bottled with screw caps, and
mixture composed of ethanol, water and non-volatile solids. From stored at 18 ± 2 ◦ C.
these tasks, overall flux, ethanol composition in permeate and the Malbec blended wines: The initial wine MB (14.8% v/v of alcohol)
required pervaporation advance will be obtained at each of the selected and wines partially dealcoholized (~10.4% v/v of alcohol at the
temperatures. selected temperatures (MB35, MB40, MB45) were used to obtain
Subsequently, in order to perform steps 5, 6 and 7 of Fig. 1; namely blended wines (MB blend 35, MB blend 40, MB blend 45) with 13.5% v/
the assessment of the effect of process temperature on chemical pa­ v of alcohol. The blends were produced in duplicated for each temper­
rameters, aroma profile and sensory characteristics of wines, additional ature and stored at 4 ◦ C for one week to equilibrate their components
runs will be done until accumulating three liters of partially deal­ before the chemical and sensory analysis.
coholized wine for each of the selected temperatures (Step 3 of Fig. 1).
For each alternative, the aroma profile of the initial wine (14.8% v/v), 3.1.2. Binary ethanol–water mixtures
the partially dealcoholized wines (10.5% v/v) and the blends (13.5% v/ Analytical grade ethanol (99.5%, Cicarelli) and distillated water
v) will be determined. All these wines will be subjected to sensory were used to prepare binary ethanol–water mixtures.
analysis by a tasting panel. The approach to be followed, in this case, is
similar to that carried out by the group in a work related to a Torrontés 3.1.3. Pervaporation commercial membrane
wine [19]. The flat sheet PDMS (polydimethyl siloxane) membrane from Per­
Finally, the optimal operation temperature will be chosen in Step 8 vatech BV (The Netherlands) with organophilic characteristic was uti­
by unifying the different quality criterions considered throughout the lized in the experiments.
work.

3. Materials and methods 3.2. General analytical parameters of wine

3.1. Materials Prior the analysis, the samples were centrifuged at 11,000×g to
eliminate impurities. Wine samples were scanned using and Alpha FT-IR
3.1.1. Initial Malbec wine Wine Analyzer (Bruker Optics, Ettilengen, Germany) to evaluate the
Malbec wines were produced using the standard protocol of the titratable acidity (tartaric acid, g/L), volatile acidity (acetic acid, g/L),
Experimental Cellar at the Instituto Nacional de Tecnologia malic acid (g/L), lactic acid (g/L), pH, and alcohol content (% v/v).
Absorbance measurements were made with a Lambda-25 Perkin-Elmer

5
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

UV–Visible Spectrophotometer (PerkinElmer, Hartford, CT, USA). Total 3.5. Conceptual modeling of the dealcoholization task through
tannins, total anthocyanins, small polymeric pigments (SPP), large pervaporation
polymeric pigments (LPP), total polymeric pigments (TP) and CieLab
parameters were measured according to Fanzone et al. [20]. Total 3.5.1. Brief description of the proposed models
phenols were determined following the methodology described by Jofré On the one hand, the pervaporation of the binary ethanol–water
et al. [21]. Chemical analysis was performed in duplicate. mixture is experimentally tested in duplicate at different temperatures
(30, 35, 40 and 45 ◦ C) and feed compositions (0.084 g/g and 0.117 g/g)
3.3. Pervaporation equipment with the aim of obtaining the parameters of a local transmembrane flux
model. In all experiments, the permeate pressure is set to 0.013 bar. The
The pervaporation experiments were accomplished in a CELFA model parameters are fine-tuned through the minimization of squared
Laboratory P-28 (CM-Celfa Membrantrenntechnik AG, Switzerland) unit errors with the collected experimental data.
provided with a pressure vessel (0.5 L) with double wall cylinder and a On the other hand, batch pervaporation of Malbec wine is experi­
membrane cell allowing the use of flat membrane discs of 75 mm mentally carried out at three selected temperatures (35, 40 and 45 ◦ C)
diameter with an effective area of 28 cm2 (see Fig. 3 in Figueroa Paredes and 0.013 bar by running the process for values of the fractional per­
et al. [11]). Recirculation flow rate is provided by a gear pump vaporation advance η from 0.05 to 0.1. Given the reasonable assumption
(Scherzinger, Germany) allowing flow velocities from 0.1 to 6 m s− 1. that the permeate primarily consists of ethanol and water, the flux and
Temperatures in the desired range of [30–45 ◦ C] can be set through an permeate composition models requires the estimation of molar fractions
external thermostat. To allow a continuous permeate sample collection, in the feed for driving force calculations. To simplify the wine compo­
two traps in parallel immersed in a cold bath are coupled to the sition conversion in molar basis, the non-volatile solid compounds are
permeate exit. A three-stage chemistry-design diaphragm pump (MD 4C combined into a single third component consisting of 40% fructose and
NT, Vacuubrand, Germany) is used to maintain the desired permeate 60% tartaric acids with an average molar mass of 162. Experimental
pressure at the desired value of 0.013 bar. data of wine dealcoholization is compared with results from a model of
the process at lab-scale using the adjusted parameters at local flux for
3.4. Pervaporation experiments binary mixture.
Finally, scaling-up of batch pervaporation unit is assessed for the
Pervaporation experiments of binary ethanol–water mixtures with a three selected operation temperatures to estimate the cost of processing
PDMS membrane were accomplished by duplicate. In each experiment, two daily batches of five thousand liters of wine operating during 12 h
an initial solution of 550 g with ethanol mass fractions of 0.084 g/g and each.
0.117 g/g was employed. Under a permeate pressure of 0.013 bar, three
permeate samples were collected at intervals of 25, 35, 40, and 60 min, 3.5.2. Local flux model
corresponding to operating temperatures of 30, 35, 40 and 45 ◦ C, The molar permeate flux for each component, which incorporate
respectively. The permeate collection associated with the chosen in­ component permeance and driving force in terms of temperature and
tervals constitutes a fraction below to 1.3% of the initial feed. To ensure permeate pressure is calculated using Eqs. (1) and (2) (see the corre­
the condensation of the permeate, the cold bath temperature was set at sponding abbreviations in nomenclature section):
− 20 ◦ C. Recirculation flow rate was selected to achieve sufficiently high ( )
PEt ( )
values of the Reynolds number (around 2300) to avoid polarization of jEt = xEt γ Et pfEt − yEt p (1)
l
both concentration and temperature. While the overall flux was ob­
tained by weighing permeate samples, its ethanol content was estimated ( )
PW ( )
by using the colorimetric method reported by Sumbhate et al. [22] as jW = xW γ W pfW − yw p (2)
l
done in a previous work [13].
Regarding the partial dealcoholization experiments, the pervapora­ ( ) ( )
In Eqs. (1) and (2), PlEt and PlW are the permeances (mol m-2h− 1 kPa− 1)
tion setup was operated at a permeate pressure of 0.013 bar and at three
for ethanol and water, respectively. While xEt and xW are the molar
different temperatures; i.e. 35, 40 and 45 ◦ C. In order to prevent wine
fractions of ethanol and water in feed, yEt and yw are the molar fractions
oxidation, a nitrogen blanket is maintained during experiments as is
of ethanol and water in permeate. γEt and γ W are the activity coefficients
done in industrial practice [1]. In all cases, the feed to the process f f
consisted of 550 g of a Malbec wine with an alcohol content of 14.8% v/ for ethanol and water and finally, pEt and pW are the saturated vapor
v. Two types of runs were carried out: (i) simple runs to build a model of pressures (kPa) for ethanol and water in feed and p is permeate pressure
the process and (ii) repetitions to accumulate an adequate volume of (kPa). While activity coefficients are estimated using the Wilson method
partially dealcoholized wine (10.5% v/v) to allow the sensory analysis in terms of temperature and molar liquid compositions, vapor pressures
of the following Malbec wines: the initial wine, the partially deal­ are calculated resorting to the extended Antoine equation. Parameters
coholized wines and the blends in a 70/30 ratio (13.5% v/v)). For the for pure components and for the binary ethanol–water mixture were
first objective, the evolution of the batch process was monitored taken from the Hysys database [23].
extracting permeate samples for values of the pervaporation fractional Temperature dependence of the ethanol and water permeances can
advance η between 0.05 and 0.1. At this stage, the required value for the be calculated from an Arrhenius type equation:
pervaporation fractional advance ηreq to achieve the desired degree of ( ) (
PEt PEtO
) (
− EapEt
(
1 1
))
dealcoholization is obtained from the process model. For the second = exp − (3)
l l R 303.15 T
goal, six repetitions at each temperature were run until reaching the
required value of the fractional pervaporation advance ηreq . (
PW
) (
PWO
) (
− EapW
(
1 1
))
The overall determination of non-volatile solid compounds in wine = exp − (4)
l l R 303.15 T
was accomplished by carefully drying the wine sample in a forced
convection oven at 120 ◦ C for 24 h and weighing the remaining solids on The reference permeances (PEtO /l) and (PWO /l) (mol m-2h− 1 kPa− 1) and
a RADWAG AS/X analytical balance. the ethanol and water activation energy of permeation EapEt and EapW (kJ
mol− 1) are the model parameters. R is gas constant (J mol− 1 K− 1) and T
is temperature (K).
Assuming that ethanol and water are the only components present in

6
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

the permeate, the molar flux and composition in the permeate can be XEt (t = t) = XEtt (14)
determined through iterative solution of Eqs. (5) and (6), in conjunction
with Eqs. (1)-(4). The iterative process is performed for given values of XS (t = t) = XSt (15)
the operating temperature, the permeate pressure and the feed
composition. Cumulative values of the permeate amount, the overall flux JCum and the
permeate composition YEtCum in the permeate collector can be expressed
yEt = jEt /(jEt + jW ) (5)
in terms of pervaporation fractional advance η.
yW = 1 − yEt (6) Mt
η = 1− (16)
M0
The overall flux J (g m− 2 h− 1) and mass fraction in permeate are sub­
sequently calculated by considering the molecular weight of each PerCum = M0 η (17)
component.
PerCum
JCum = (18)
3.5.3. Batch pervaporation model from experiments at lab scale At
The scheme of pervaporation process, which is carried out in batch-
wise mode, is depicted in Fig. 3. The amount and composition in the feed XEt0 − XEtt (1 − η)
YEtCum = (19)
tank is characterized by the differential balances denoted in Eqs. (7)–(9) η
(see the corresponding abbreviations in nomenclature section). Here
The system of ordinary differential equations, represented by Eqs. (7)-
several considerations are taken into account: (i) both the feed tank and
(9), is numerically integrated using the Runge-Kutta method.
the permeate collector exhibit perfect mixing; (ii) the membrane swiftly
adjusts to the new concentrations in the feed; and (iii) the time required
for the recirculation is negligible compared to that of processing time. 3.6. Scaling-up and economic analysis
For the model estimations at lab-scale, it is assumed that the tem­
perature and composition remain homogeneous along the membrane With regards to the scaling-up process, it’s essential to bear in mind
cell with membrane area A of 0.0028 m2. that, as part of a screening procedure, it represents a preliminary design
rather than a rigorous one. The scaled-up process involves the operation
dM
= − JA (7) of two batches of five thousand liters of wine per day, each running for
dt up to 12 h. The design of the pervaporation unit (Fig. 2) with a fixed
membrane area at a given temperature requires the estimation of the
dXEt JA
= − (YEt − XEt ) (8) operating time for a given reduction of the alcohol content, which in this
dt M
case is about four percentage points. Here an iterative procedure in­
dXS XS JA volves the error convergence between the numerical integration of Eqs.
= (9) (7)–(9), with integer numbers of membrane modules (13 m2 each, [1])
dt M
in parallel arrangement, and the target ethanol composition in the feed
when Eqs. (7)-(9) are integrated, the following initial conditions are tank at the end of the process. To account for a margin of error, the
used: calculated operating time is overestimated by 20%. The recirculating
pump operates at a rate of 8 m3h− 1 for each element in parallel [1].
M(t = 0) = M0 (10)
Whereas the steam requirement necessary to maintain the temperature
XEt (t = 0) = XEt0 (11) of the retentate stream is determined by calculating the energy needed
to evaporate the permeate, which is sourced from the internal energy of
XS (t = 0) = XS0 (12) the retentate, the steam consumption during the start-up is estimated to
warm the feed content from ambient temperature. The sizing of the heat
The integration ends when the operating time t is reached. exchanger is estimated using typical overall heat transfer coefficients.
Considering a permeate pressure of 0.013 bar and a condensing tem­
M(t = t) = Mt (13)
perature of − 10 ◦ C, the permeate condensing is achieved via a vapor
compression refrigeration system. This subsystem followed by a vacuum

Fig. 3. Batch pervaporation unit.

7
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

pump must be sized taking into account the leakage of air into the Table 3
equipment under vacuum. For the cases of considering heat integration, Design and operating variables as a function of process temperature. Pervapo­
the temperature of the retentate stream is maintained by using the ration with (PVHI) and without heat integration (PV) are considered.
condensation heat from the cooling service fluid within the refrigeration PV PVHI
system, thus avoiding steam consumption during normal operation (see Operating conditions
Fig. 2). Furthermore, the incorporation of heat integration also leads to a Operating temperature (◦ C) 35 40 45 35 40 45
decreased consumption of cooling water for refrigerant condensation. Condensing temperature (◦ C) − 10 − 10 − 10 − 10 − 10 − 10
The economic analysis for the alternatives studied is performed ac­ Permeate pressure (kPa) 1.33 1.33 1.33 1.33 1.33 1.33
Design variables
cording to the methodology explained in Seider et al. [24] by calculating
Feed vessel volume (m3) 6.25 6.25 6.25 6.25 6.25 6.25
the corresponding annualized investment and the operating costs. Permeate vessel volume (m3) 0.61 0.61 0.61 0.61 0.61 0.61
Table 1 shows the design and operation variables considered at the Membrane area (m2) 52 39 26 52 39 26
conceptual design level, which in turn are the input data of the economic Retentate heat exchanger area 0.5 0.8 1.1 0.5 0.8 1.1
model. (m2)
Condenser area (m2) 16.8 17.0 14.9 16.8 17.0 14.9
Among the operating costs, the membrane replacement calculated Refrigerant condenser area (m2) 3.2 3.3 2.9 0.7 0.6 0.5
from lifetime given by membrane suppliers is also taken into account as Operating variables
an operating cost item. Refrigeration is considered as a utility [25], with Operation time (h) 10.2 10.1 11.4 10.2 10.1 11.4
a focus on function rather than the sizing and costing of a vapor Recirculating pump electricity 19.1 15.1 12.8 19.1 15.1 12.8
consumption (kWh/batch)
compression refrigeration system. The consumption of cooling water for
Vacuum pump electricity 7.7 7.0 7.3 7.7 7.0 7.3
the refrigerant condensing is accounting as an item apart among utili­ consumption (kWh/batch)
ties. Prices related to utilities are shown in Table 2 using the correlations Steam consumption(kg/batch) 434 475 517 85 128 172
provided by Ulrich and Vasudevan [26]. The price of the alcohol-rich Condensation duty (kWh/batch) 222 222 221 222 222 221
permeate, destined for a wine distillery located ten kilometers away Cooling water (m3/batch) 0.66 0.67 0.59 0.17 0.19 0.18

from the wine facility to produce high-quality grape brandy, is also

provided. Here, the credit earned through the sale of the valorized consensus the outstanding attributes (color, aroma, mouthfeel).
permeate is accounted for. Costs related to annual utilities consumption Normalized-tasting glass wine was used [29] and labeled with three-
are calculated by considering the operating cycle time of each process digit random numbers. The three remained tasting sessions were
alternative in a time-horizon of 360 days/year. The operating cost employed to evaluate by descriptive sensory analysis. All wines were
associated with the reduction in wine volume during the deal­ punctuated on an unstructured scale using Soldesa Software (ISETA,
coholization process is taken into account. Note that the total capital Buenos Aires, Argentina). Additionally to the statistical analysis of
investment for each alternative is annualized taking ten years as lifetime sensory evaluation, the correlation of each panelist with the panel
with 10% interest rate. Finally, mixing of the partially dealcoholized means and their contribution to the panelist × wine interaction for each
wine with fresh wine to obtain a blended wine with a reduction in the attribute were evaluated to determine the panel performance. All as­
alcohol content of 1.3 percent points is also considered when estimating sessments were conducted in individual tasting cabins (22–24 ◦ C,
economic figures for the process variants at 35, 40 and 45 ◦ C. 45–55% RH) [20,30].

3.7. Chromatographic analyses of free volatile compounds

3.9. Statistical analyses

Free volatile compounds were studied according to Massera et al.

Statistical analysis of chemical analysis was performed with Stat­
[27], with some modifications. Five milliliters of wine were employed to
graphics Centurion XVI software (Statistical Graphics Corp., Warrenton,
extract and concentrate the volatiles using PDMS/DVB SPME extraction
VA, USA, 2009) and R (R Core, 2018) to evaluate the sensory results. The
fiber (Stableflex, Sigma-Aldrich, Gillingham, UK). The separation,
homogeneity of variance was evaluated using Levene’s test and
identification, and quantification were carried out with a Varian CP-
analyzed by one-way analysis of variance with Tukey’s honestly sig­
3800 Gas Chromatograph and a Saturn 2200 (Varian, USA) Mass
nificant difference (HSD) test (α = 0.05). p < 0.05 was considered to be
Spectrometer. Identification was made by comparing the Mass Spectra
statistically significant. Flash profile data analyses were performed by
of each compound with the mass spectra of pure standards and the Mass
multiple factor analysis (MFA) and confidence ellipses indicating 95%
Spectra of the NIST 2.0 Library, considering RMatch and FMatch pa­
confidence intervals were formed using the SensoMineR R package
rameters higher than 800. To confirm the identification, the experi­
[20,31].
mental Linear Retention Index (LRI) were compared with those
compiled by the NIST WebBook database [28] and with those obtained
4. Results
with pure reference standards.
4.1. Model parameters and pervaporation performance from experiments
3.8. Sensory evaluation of wines at lab scale (steps 1, 2 and 3 of Fig. 1)

Sensory analysis was performed by Descriptive Analysis. Four sepa­ Starting with experiments on binary ethanol–water mixtures, Fig. 4
rate sessions were made by ten trained judges (six females and four (a) shows the experimental permeate fluxes obtained at different tem­
males) aged 26–50 years, who are members of trained panel of INTA. In peratures and feed compositions as explained in Section 3.5.1. In a
the first session, all wines were evaluated to discuss and select by similar way, Fig. 4(b) illustrates the mean value of experimental
permeate composition for the two selected feed compositions. Since the
Table 2 permeate composition did not show significant variations with respect
Costs for utilities and the alcohol-rich permeate. to temperature at each feed composition level, mean values were
Alcohol-rich permeate (US$/L) 1.0
adopted for the parameter’s regression of the local flux model described
Steam (1 barg) (US$/kg) 0.034 in Section 3.5.2. The adjusted parameters obtained by minimizing the
Cooling water (US$/m3) 0.125 least square errors through nonlinear regression were the following:
Refrigeration (US$/GJ) 16.6 EapEt and EapW , with values of − 5.04 and 3.19 (kJ mol− 1) respectively;
Electricity (US$/MWh) 147.4
and, (PEtO /l) and (PWO /l), with values of 3.75 and 6.99

8
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

Fig. 4. Binary ethanol–water experiments: (a) Experimental and calculated

overall flux vs. operation temperature (oC) at feed ethanol compositions of Fig. 5. Partial dealcoholization experiments: (a) Experimental and calculated
0.084 (g/g) and 0.117(g/g) and, (b) mean experimental and calculated ethanol cumulative overall flux vs. pervaporation fractional advance at 35, 40 and 45 ◦ C
mass fraction in permeate vs. ethanol mass fraction in feed. and, (b) experimental and calculated cumulative ethanol composition in
permeate vs. pervaporation fractional advance.
(mol m− 2 h− 1 kPa− 1), respectively. The permeance values obtained in
this study were found to be very similar to those reported by Kujawska 45 ◦ C) and at a pervaporation advance of 0.045. For the remaining data
et al. [32] in their VPV experiments for an ethanol–water mixture at points, the error remains consistently below 2.5%. Fig. 5(b) shows the
30 ◦ C and at an ethanol feed composition of 4 wt%. Similarly, water experimental and simulated cumulative ethanol composition in
reference permeance is very close to that obtained by Figueroa Paredes permeate versus the fractional pervaporation advance. The highest de­
et al. [33] for beer pervaporation at temperatures from 20 to 40 ◦ C. viation for the predictions in cumulative ethanol compositions is 2.7%.
Fig. 4(a) and (b) also depict the calculated values of permeate flux The analysis of Fig. 5 shows the strong dependence of the overall flux
and permeate composition, respectively. A maximum error of 3.5% is with the operating temperature, unlike what occurs with the cumulative
observed between the experimental and predicted permeate flux values. ethanol composition in the permeate. Both experimental and simulated
Furthermore, up to 1.4% of error is reached among the experimental and results are presented as a function of the pervaporation fractional
predicted compositions. advance η instead of time as the former variable allows to describe the
Regarding laboratory-scale experiments with the red wine, three dealcoholization process by means of a recipe and also easy the scaling-
batches were carried out from a feed amount of 550 g as explained in up step.
Section 3.5.1. The measured initial mass fraction of non-volatile solids According to the simulations, if it is desired to decrease the alcohol
counted as a single compound was 0.028. In practical terms, the molar content of the wine by four percentage points, it is required to extend the
fraction of non-volatile solids may be negligible; nevertheless, it is dealcoholization operation to reach an η value of 0.085 regardless of the
crucial to consider that it can influence the calculation of the molar operating temperature. However, in order to guarantee the deal­
fractions of ethanol and water. coholization objective, a pervaporation advance of 0.09 was adopted.
Fig. 5(a) illustrates both experimental and calculated overall flux For the aforementioned pervaporation advance, compositions of ca.
against the pervaporation fractional advance η at the three selected 10.4% v/v (0.083 g/g) were experimentally achieved. These values
temperatures. The mean permeate flux prediction exhibits the highest agreed reasonably well with those obtained from the model simulation
error, approximately 7%, at the highest temperature analyzed (i.e., (10.6% v/v (0.084 g/g)).

9
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

Both the cumulative permeate composition that, at the required highest cost in terms of investment is linked with the membrane area
value of the pervaporation fractional advance, reaches a value of 0.465 for the design at a temperature of 35 ◦ C. The cost related to the in­
g/g (54% v/v) and its aroma content (see Section 4.4) allows one to vestment of the permeate condenser is the second most significant item
think the grape juice produced as a raw liquor for off-site producing for the designs at temperatures of 40 and 45 ◦ C. Taking into account the
quality grape brandy [1]. annualized capital investment, the design corresponding to the highest
As shown above, the model was built up from simple runs for the analyzed temperature of 45 ◦ C proves to be the most economical, with
selected temperatures at laboratory scale. As in order to perform the savings approximately of 10% and 17% compared to the designs at
sensory analysis a minimum of three liters of dealcoholized wine at each 40 and 35 ◦ C, respectively. Furthermore, for each temperature, the
operating temperature is needed, six additional runs were made for each capital investment for the designs with heat integration consideration is
operating temperature until reaching the required value of the perva­ notably 3% lower.
poration advance ηreq of 0.09. It should be noted that the operating times Among the utility costs, the most expensive component is steam
of the additional runs differed by average values of 8.4, 16.1 and 12.0% consumption for the design alternatives without heat integration, while
with respect to the values obtained through model simulations at tem­ refrigeration costs become the most relevant utility cost when heat
peratures of 35, 40 and 45 ◦ C; that is, 16.6, 11.9 and 9.6 h, respectively. integration is incorporated. The cost of membrane replacement, relative
to its initial investment, is also at its lowest for the design at 45 ◦ C.
Consistent with the trend observed in the investment cost, the total
4.2. Effect of dealcoholization temperature on process economics (step 4
operating cost also decreases as the temperature increases. Heat inte­
of Fig. 1)
gration significantly impacts the reduction of operational costs,
achieving a remarkable 17% reduction. Finally, considering both in­
The design and operating variables for each studied temperature are
vestment and operating costs, the most cost-effective design corresponds
displayed in Table 3. The table includes both pervaporation designs,
to heat integration at 45 ◦ C.
with (PVHI) and without heat integration (PV). From the presented re­
Finally, Table 4 also shows annualized values corresponding to each
sults, it is noted that the volume of the feed and permeate vessels re­
of the process alternatives for the total cost of the dealcoholization
mains equal for all designs. The membrane area requirement decreases
process, the credit from the sale of the ethanol-rich permeate to a wine
as the temperature increases, reaching its lowest value at 45 ◦ C with two
distillery for their material valorization, the resulting earnings and the
membrane modules of 13 m2 each arranged in parallel. The heat ex­
profit per liter of the final product with an ethanol content of 13.5% v/v.
change area and the steam consumption for retentate warming increase
Economic figures for each variant in Table 4 are estimated based on
as the operating temperature deviates from the ambient temperature.
the production cost and the credit from the annual sale of 350,000 L of
Since the condensation duty is practically the same for all temperatures,
ethanol-rich permeate to the local market at a price of 1 US$/L, a value
the condenser area is inversely proportional to the operating time, with
that is in accordance not only with the high alcohol content of the
the lowest requirement observed for the design at 45 ◦ C. As expected,
permeate [34] but also with its aroma components content [1] as will
regardless of the analyzed temperature, steam and cooling water con­
show in Section 4.4. Figures also reflect the assumption that that the cost
sumption is significantly reduced in the design alternatives that incor­
associated with the reduction in wine volume is absorbed by the price of
porate heat integration. The electricity consumption of the recirculating
the final blend product. Note that the cost associated with the reduction
pump, which depends on the number of membrane modules in parallel,
in wine volume represents the cost of wine before bottling of about 324
is highest for the designs at 35 ◦ C. For the vacuum pump, electricity
thousand liters of wine (3% of yearly production). Adopting a dilution
consumption varies by approximately 10% across the different alter­
ratio of 70/30 for fresh wine to dealcoholized wine, the production of
natives compared to the lowest consumption observed at 40 ◦ C.
blended wine increases to 10.9 million liters per year.
Presented in thousands of dollars, Table 4 provides the investment
Fig. 6 shows both payback period and internal rate of return for
and operating costs, which vary with process temperature and are
variants with (PVHI) and without heat integration (PV) as a function of
associated with the designs mentioned in the previous table. For all the
process temperature. Attractive indicators, such as payback periods
temperatures analyzed, the highest investment cost is associated with
between 2 and 3 years and internal return rates between 25 and 36%
the feed vessel, which is constructed from stainless steel. The second-

Table 4
Investment and operating costs as a function of process temperature. Pervaporation alternatives with (PVHI) and without heat integration (PV) are considered.
PV PVHI

Operating temperature ( C)
◦
35 40 45 35 40 45

Capital Investment (Thousand US$) Feed vessel 133.2 133.2 133.2 133.2 133.2 133.2
Permeate vessel 39.6 39.6 39.6 39.6 39.6 39.6
Membrane module 120.3 90.2 60.1 120.3 90.2 60.1
Retentate heat exchanger 16.3 17.8 19.6 16.3 17.8 19.6
Condenser 101.3 102.4 91.7 101.4 102.4 91.7
Refrigerant condenser 30.4 30.6 28.6 17.5 16.9 16.2
Recirculating pump 31.3 31.2 31.1 31.3 31.2 31.1
Vacuum pump 19.4 17.8 15.6 19.4 17.8 15.6
Total investment cost 491.9 462.9 419.5 479.0 449.2 407.2
Annualized capital investment (Thousand US$/year) 80.1 75.3 68.3 78.0 73.1 66.3
Operating cost (Thousand US$/year) Electricity 2.8 2.3 2.1 2.8 2.3 2.1
Steam 10.7 11.7 12.7 2.1 3.2 4.2
Refrigeration 9.6 9.6 9.6 9.6 9.6 9.6
Cooling water 2.2 2.2 2.2 0.6 0.6 0.7
Membrane annual replacement 2.7 2.0 1.4 2.7 2.0 1.4
General cost 52.1 50.0 46.7 51.1 48.9 45.8
Total operative cost 80.1 77.8 74.7 68.9 66.6 63.8
Total cost of dealcoholization process (Thousand US$/year) 160.2 153.1 143.0 146.8 139.8 130.0
Credit from the sale of permeate (Thousand US$/year) 350.1 350.1 350.1 350.1 350.1 350.1
Earnings (Thousand US$/year) 189.9 196.9 207.1 203.2 210.3 220.0
Profit per liter of blended wine (US$ cents/L) 1.7 1.8 1.9 1.9 1.9 2.0

10
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

peratures at laboratory scale. Subsequently, in order to perform the

sensory analysis, additional runs were made at each of the selected
temperatures until three liters of partially dealcoholized wine were
accumulated in two bottles (A and B). It should be noted that the
operating times of the additional runs differed by average values of 8.4,
16.1 and 12.0% with respect to the values obtained through model
simulations at temperatures of 35, 40 and 45 ◦ C; that is, 16.6, 11.9 and
9.6 h, respectively.
Table 5 shows the overall parameters analyzed in the initial, deal­
coholized and blended wines. The initial alcohol was reduced by 32.43,
27.9 and 28.24% at 35, 40 and 45 ◦ C respectively. Blended wines reach
approximately 13.5% of ethanol by 70:30 mixing of initial and deal­
coholized wines. The pH of wines, which has a direct impact on the
sensory quality, was affected by the pervaporation process but not by the
different temperatures. Total and volatile acidity showed statistical
differences between initial and dealcoholized wines. Dealcoholized
wines at high temperatures had a higher content of both parameters.
Blended wines at 35 ◦ C did not show differences concerning the total
Fig. 6. Internal rate of return and payback period as a function of process acidity, but at 45 ◦ C had 2% more organic acids (expressed as tartaric
temperature. Pervaporation alternatives with (PVHI) and without heat inte­ acid) than initial Malbec. Volatile acidity showed no statistical differ­
gration (PV) are considered. ences between the different temperatures applied. Glucose is the prin­
cipal reducing sugar metabolized by yeast as a carbon and energy source
that were obtained. The best figures corresponded to the design with and is better fermentable than fructose [35]. For this reason, it is ex­
heat integration at 45 ◦ C. pected the glucose concentration to be lower than fructose in finished
wines. Dealcoholized wines showed four times more fructose than the
initial MB, and higher contents were observed at low operation tem­
4.3. Effect of dealcoholization temperature on chemical parameters of peratures. These results were in agreement with Motta et al. [36] in Rosé
wines (Step 5 of Fig. 1) and Barbera and Pelaverga red varieties dealcoholized by membrane
contactor and distillation under vacuum. On the other hand, our pro­
The following sections, which correspond to stages 5, 6 and 7 of cedure showed differences concerning the other dealcoholization tech­
Fig. 1, focus on the initial Malbec wine MB, the dealcoholized wines at niques. For example, reverse osmosis-evaporative pertraction applied to
the three selected temperatures (MB 35, MB40, MB45 and the blended reduce the ethanol in Verdelho, Petit Verdot, and Cabernet sauvignon
wines (MB blend 35, MB blend 40, MB blend 45), which were obtained did not significantly affect their general parameters [7,37].
by blending 30% of the partially dealcoholized wines with 70% of fresh Phenolic compounds are responsible of color, astringency, and
wine. This resulted in wines with an alcohol content of 13.5% v/v. bitterness in wines. Anthocyanins are hydrosoluble pigments respon­
As shown in the previous sections, the conceptual design of the sible for the red/violet color of red wines and are affected by tempera­
processes corresponding to the operating temperatures of 35, 40 and tures during the grape growing, winemaking and conservation [38,39].
45 ◦ C was carried out from integration of the process model until The differences of anthocyanins between initial and dealcoholized wines
reaching the required value of the pervaporation fractional advance ηreq would be due that the low concentration of ethanol in dealcoholized
of 0.09. The model was built up from simple runs for the selected tem­

Table 5
One way analysis (ANOVA) of basic general parameters and phenolics of initial Malbec (MB), dealcoholized at 35, 40 and 45 ◦ C (MB35, MB40, MB45), and blend
wines.
Parameter MB Dealcoholized wines Blend wines

MB35 MB40 MB45 MB35 MB40 MB45

Alcohol content (% v/v) 14.80 (0.01)d 10.00 (0.28)a 10.67 (0.15)b 10.62 (0.05)b 13.53 (0.04)c 13.52 (0.12)c 13.51 (0.12)c
pH 4.04 (0.00)b 3.97 (0.01)a 3.99 (0.00)a 3.98 (0.01)a 4.00 (0.01)a 4.00 (0.00)a 4.00 (0.00)a
Total acidity (tartaric acid g/L) 3.80 (0.00)a 4.08 (0.07)c 4.24 (0.02)d 4.39 (0.05)e 3.82 (0.02)a 3.90 (0.02)ab 4.00 (0.04)bc
Volatile acidity (acetic acid g/L) 0.37 (0.00)a 0.45 (0.02)bc 0.47 (0.00)c 0.48 (0.00)c 0.41 (0.00)ab 0.41 (0.00)ab 0.40 (0.00)ab
Glucose (g/L) nd nd nd nd nd nd nd
Fructose (g/L) 0.24 (0.00)a 1.11 (0.07)d 0.95 (0.02)c 0.87 (0.05)bc 0.76 (0.00)b 0.84 (0.04)bc 0.87 (0.00)bc
Density (g/L) 0.992 (0.00)a 0.995 (0.00)c 0.996 (0.00)c 0.995 (0.00)c 0.996 (0.00)b 0.993 (0.00)b 0.993 (0.00)b
Total anthocyanins (mg malvidin-3- 322.33 (2.62)c 286.23 (1.78)b 274.47 (5.13) 262.91 (0.74)a 326,53 (10.72) 325.02 (13.35) 328.27 (3.189)
glucoside/L) ab c c c
Total phenols (mg galic acid/L) 1320.30 1374.34 1358.76 1349.90 1349.04 1302.75 1308.20
(10.75)a (48.23)a (40.77)a (25.46)a (25.46)a (17.86)a (11.53)a
Total tannins (mg catechin/L) 576.46 (4.69)c 321.819 391.44 (9.96) 437.55 (25.89) 396.75 (33.16) 363.22 (8.06) 434.06 (18.34)
(75.85)a ab b ab ab b
LPP 0.60 (0.00)a 1.10 (0.12)b 1.07 (0.04)b 1.21 (0.28)b 1.00 (0.02)ab 0.97 (0.18)ab 0.935 (0.21)ab
SPP 2.40 (0.01)ab 2.23 (0.01)ab 2.49 (0.02)ab 2.57 (0.19)ab 2.13 (0.01)a 2.55 (0.36)ab 2.66 (0.13)b
TP 3.01 (0.02)a 3,34 (0.11)bc 3.57 (0.06)cd 3.78 (0.134)d 3.13 (0.01)b 3.52 (0.23)cd 3.60 (0.10)cd
L* 68.83 (0.17)c 65.75 (0.22)a 65.11 (0.22)a 65.36 (0.32)a 67.64 (0.07)b 67.28 (0.41)b 67.19 (0.25)b
C*ab 29.14 (0.77)a 31.59 (0.37)b 33.04 (0.24)c 33.26 (0.04)c 30.21 (0.24)ab 30.78 (0.44)bc 30.76 (0.28)bc
hab 8.09 (0.03)e 5.66 (0.34)a 6.29 (0.38)ab 6.46 (0.08)bc 6.51 (0.01)bc 7.94 (0.10)d 7.24 (0.31)cd
ΔEMB/MB35-40-45 4.53 5.42 5.77
ΔEMB/MBblend35-40-45 2.85 2.68 2.82
ΔEMB35-40-45/MBblend35-40-45 1.63 1.67 1.73

Mean (SD) for n = 2. Different letters indicate significant differences among treatments (Tukey HSD test, p < 0.05).

11
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

wines (~10.5%) promote the condensation between anthocyanins and shown different results for the anthocyanin content. Banvolgyi et al.
tannins, generating polymeric pigments. It could be seen because the [45] and Bogianchini et al. [46] observed that the anthocyanin content
LPP (formed by anthocyanins that have reacted with tannins) was higher of dealcoholized wines was similar to that observed in the initial wines
in dealcoholized respect to the initial wines, and total tannins had the and higher for Belisario-Sánchez et al. [47], Motta et al. [36] and Pham
opposite behavior. Polymeric pigments have a positive effect on the et al. [37]. Therefore, based on the analytical results, brown hues were
color stability during wine aging and mouth feel [40–42]. Another not detected in the dealcoholized wines. The results could indicate that
plausible explanations for this result could be related to the adsorption/ the decrease in anthocyanin content (and tannins with the increase of
absorption phenomena of anthocyanins on the PDMS membrane [43] LPP) in dealcoholized wines is more closely related to the formation of
and/or the effect of the temperature on the anthocyanin stability at high polymeric pigments. On the other hand, blended wines showed no sig­
temperatures. Romero and Bakker [44] observed the degradation of nificant differences in the total anthocyanins about the initial Malbec.
anthocyanins in model solutions from 20 ◦ C with the presence of brown Surprisingly, the total content of phenolic compounds did not show
precipitates. Also, Giuffrida de Esteban et al. [31] evaluated the impact differences between all treatments.
of temperature on anthocyanin profile during wine aging and observed Wine color is described employing specific attributes such as light­
that at 25 ◦ C the total content fell 255 times more than 15 ◦ C. There is no ness (L*), chroma (C*ab), and hue (hab). Data showed the lightness (L*)
reference concerning the impact of pervaporation with PDMS membrane was slightly higher in the initial MB for the dealcoholized and blended
on the phenolic composition. Other dealcoholization strategies have wines, indicating that the initial wine is lighter than the other wines.

Table 6
Volatile organic compounds (μg/L) in initial Malbec wines (MB), dealcoholized wines (MB 35-MB40-MB45) and blend wines (MBblend35-MBblend40-Mblend45)
analyzed by HS-SPME/GC–MS.
Compound MB Dealcoholized wines Blend wines

MB35 MB40 MB45 MB35 MB40 MB45

Higher alcohols
2,3-Butanediol 32.67 (0.96)d 10.26 (0.97)a 15.037 (2.43) 18.66 (4.42)bc 17.01 (0.01)c 12.30 (0.42)ab 29.48 (0.67)e
bc
S-3,4-Dimetilpentanol 98.58 (2.59)d 16.41 (0.24)a 16.59 (0.62)a 13.70 (0.90)a 41.38 (2.37)b 44.96 (0.02)bc 45.85 (1.22)c
2-Ethyl-1-hexanol 6.56 (0.19)e nd 1.53 (0.10)a 3.81 (0.10)c 3.68 (0.15)c 2.23 (0.06)b 5.63 (0.02)d
(E)-2-Nonen-1-ol 14.71 (0.37)d 11.60 (0.91)c 4.06 (0.20)a 6.18 (0.06)b 12.23 (0.03)c 5.29 (0.17)b 20.96 (0.48)e
Phenylethyl alcohol 1442.46 (54.00)e 1365.54 (19.61)d 952.37 (7.63)b 1243.94 (20.71)c 1306.27 (12.19) 728.68 (12.41)a 1254.80 (19.06)c
cd
3-Methyl-2-pentanol nd nd 2.75 (0.21)b nd nd 1.67 (0.03)a nd
Total 1595.30 (58.13) 1404.57 (17.74) 960.72 (8.16)b 1282.36 (19.98) 1380.58 (9.65)c 795.14 (13.12) 1356.71 (10.80)
d c c a c
MCFA Ethyl Esters
Ethyl butanoate 24.19 (0.24)d nd nd nd 13.64 (0.07)b 12.19 (0.38)a 16.65 (0.27)c
Ethyl 2-methylbutanoate 13.74 (0.23)c nd nd nd nd 1.76 (0.01)b 1.00 (0.01)a
Ethyl 3-methylbutanote 2.58 (0.01)b nd nd nd nd 0.56 (0.01)a nd
Ethyl hexanoate 454.80 (7.22)g 89.63 (1.04)c 22.74 (1.38)b 7.41 (0.17)a 334.97 (7.60)e 220.72 (3.12)d 363.02 (9.05)f
Ethyl 2-hexenoate 6.16 (0.06)c nd nd nd 3.11 (0.02)a 4.77 (0.16)b nd
Ethyl octanoate 1100.08 (34.63)f 649.46 (44.24)d 179.58 (2.56)b 114.79 (3.85)a 1083.72 (9.51)ef 442.02 (11.58)c 1024.86 (13.63)e
3-Methylbutyl hexanoate 1.69 (0.07)b nd nd nd nd 1.61 (0.03)b 0.62 (0.01)a
Ethyl nonanoate 21.68 (0.55)c 11.78 (0.02)ab 12.01 (0.08)ab 12.92 (0.73)b 11.88 (0.16)ab 12.05 (0.08)ab 10.98 (1.11)a
Ethyl decanoate 329.58 (20.06)e 271.35 (10.00)d 128.47 (7.59)b 65.41 (2.64)a 264.31 (22.74)d 127.52 (6.39)b 223.73 (13.44)c
Ethyl trans-4-decenoate 49.73 (3.74)e 21.81 (2.04)b 34.60 (2.34)c 11.58 (0.11)a 42.29 (1.90)d 13.99 (0.61)a 48.75 (1.80)e
3-Methylbutyl octanoate 11.15 (0.22)e 4.57 (0.16)c 2.75 (0.22)a nd 9.45 (0.07)d 4.11 (0.33)bc 3.81 (0.08)b
Total 2015.42 (65.48)f 1048.62 (37.51) 380.17 (1.01)b 212.13 (7.20)a 1763.39 (38.01)e 841.31 (19.21) 1693.45 (1.98)e
d c
Higher alcohols acetates
3-Methylbutanol acetate 204.44 (4.89)e 14.56 (0.03)a nd nd 154.74 (4.16)d 95.60 (2.13)b 110.51 (2.89)c
2-Methyltubutanol acetate 57.24 (1.95)d 19.74 (0.35)a nd nd 36.54 (1.02)b 19.22 (1.02)a 49.30 (2.67)c
2-Phenylethyl acetate 60.46 (1.62)c 57.35 (3.88)c 30.41 (2.51)a 33.42 (0.28)a 46.36 (2.17)b 44.87 (2.91)b 46.91 (1.88)b
Total 322.92 (8.47)f 91.65 (4.22)b 30.41 (2.51)a 33.42 (0.28)a 237.64 (5.81)e 159.70 (4.40)c 206.73 (7.14)d
Medium chain fatty acids
Hexanoic acid 38.31 (2.39)d 25.58 (0.85)bc 15.01 (2.26)a 12.97 (2.76)a 29.94 (2.32)c 23.54 (1.73)b 30.36 (0.89)c
Octanoic acid 108.32 (1.20)f 18.57 (0.99)a 32.24 (1.53)b 35.54 (7.96)bc 74.10 (1.64)e 42.30 (1.23)c 63.18 (3.33)d
Nonanoic acid 124.15 (1.53)d 53.06 (4.51)c 31.50 (4.05)b 38.74 (2.01)b 31.75 (1.48)c 20.15 (0.61)a 29.83 (6.68)ab
Decanoic acid 39.36 (2.78)f 20.69 (0.61)ab 28.09 (3.02)cd 32.64 (2.68)de 18.62 (2.34)a 35.97 (1.85)ef 25.21 (1.10)bc
Total 310.15 (7.71)d 117.90 (5.67)a 106.85 (7.75)a 119.90 (11.40)a 174.42 (2.30)c 121.98 (2.89)a 148.59 (12.00)b
Norisoprenoids
β-Ionone 17.00 (0.96) nd nd nd nd nd nd
β-Damascenone 51.99 (2.93)d 15.46 (0.67)b 1.31 (0.01)a 1.40 (0.09)a 23.70 (0.81)c 2.18 (0.04)a 1.25 (0.04)a
Total 68.99 (3.90)d 15.46 (0.67)b 1.31 (0.01)a 1.40 (0.09)a 23.70 (0.81)c 2.18 (0.04)a 1.25 (0.04)a
Terpenoids
Limonene nd nd 0,26 (0,02)a nd nd 0,28 (0,01)a nd
2,7-Dimehtyl 4,5-octadien-1- 2,01 (0,02)c nd nd nd 1,15 (0,01)b 0,64 (0,02)a nd
ol
α-Linalool 10,00 (0,21)f 2,23 (0,03)b 0,99 (0,01)a 0,89 (0,06)a 6,23 (0,41)e 3,57 (0,13)c 5,18 (0,74)d
β-Citronellol 69,24 (6,27)e 22,97 (2,75)c 14,31 (3,32)a 15,25 (0,78)ab 38,82 (1,25)d 19,56 (0,27)ab 34,98 (0,03)d
Nerol 15,48 (0,48)d 4,48 (0,30)b nd 8,47 (0,76)c 8,54 (0,11)c 1,61 (0,03)a 9,50 (0,95)c
Nerolidol 6,15 (0,35)b nd nd nd 3,03 (0,01)a 2,24 (0,03)a 2,48 (0,95)a
Total 102,88 (7,30)d 29,68 (2,48)b 15,56 (3,35)a 24,61 (0,32)ab 57,85 (1,78)c 27,91 (0,09)b 52,51 (1,93)c

Values are means of two replicates and the standard deviation in parenthesis. Different letters in each row indicate statistical differences between all treatments
employing Tukey’s (HSD) test (α = 0.05).

12
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

Temperature showed no effect on dealcoholized wines and blended shows for each family of components the overall percentage losses in the
wines concerning this parameter. As shown in Table 5, the initial wine alcohol-rich permeate, the concentration in the permeate and the sep­
showed a lower intensity of color (related to the C*ab parameter) con­ aration factor β defined as the ratio between a component concentration
cerning the dealcoholized and blended wines and was higher in deal­ in the permeate and in the raw wine. Note that the dealcoholization
coholized wines at 40 and 45 ◦ C. Hue (hab) is a CIELAB parameter process leads to a decrease in volume characterized by the pervapora­
associated with the red/blue tonality in red wines. In dealcoholized tion fractional advance η.
wines, hab showed a direct linear relationship with SPP (r2 = 0.9994), Longo et al. [7] evaluated the impact of partial dealcoholization by
which is associated with red pigments. At higher dealcoholization reverse osmosis-evaporative pertraction (RO-EP) and early harvest (EH)
temperatures, the wines showed higher hab values, indicating a shift on the volatile and sensory profile of Petit Verdot and Verdelho wines.
towards redder hues, with the violet hue preserved at lower Reductions in alcohol content of 4.4% points (Verdelho) and 2.5%
temperatures. points (Petit Verdot) with respect to the content of the corresponding
The total color difference (ΔE*ab) is an important parameter for the late harvested wines were considered. The authors demonstrated the
wine industry because it expresses the ability of human eyes to reduction of VOCs in dealcoholized wines, independently of the adopted
discriminate the color between two wines, considering the visual technology. For instance, for the case of dealcoholization of the Ver­
discrimination threshold between 3 and 5 units [42]. Data showed that delho wine via membrane technology they observed the loss of 60% of
only the initial and dealcoholized wines could discriminate among them the total esters content, 34% of higher alcohols, 36% of terpenoids and
by color for the human eye. An important point to observe is that the 53% of norisoprenoids. In a more recent work, Sun et al. [1] address the
blended wines cannot be differentiated from the initial MB or the total dealcoholization of a red wine by pervaporation with a PDMS
dealcoholized wines, because the color difference was less than three in membrane. In Table 6 of the aforementioned manuscript, the authors
both cases. It would indicate that adopting a dilution ratio of 70/30 for analyze the selectivity of the membrane for species belonging to the HA,
fresh wine to partially dealcoholized wine is adequate to reduce the MCFAEE and MCFA groups in terms of concentration in the permeate
alcohol content without affecting the visual quality of the wines. and separation factor. For comparison purposes, the percentage losses of
each group can be calculated by simply grouping them by family and
considering the pervaporation fractional advance of 0.33 reported in
4.4. Effect of dealcoholization temperature on volatile composition of
Table 9 [1]. Thus, the percentage losses are 39.5, 48.9 and 28.8 % for the
wines (Step 6 of Fig. 1)
HA, MCFAEE and MCFA groups, respectively. Rightly, the authors
justify the low losses of the MCFA group components on a low driving
Table 6 reports the evaluation of the volatile compounds in the initial
force for mass transfer across the membrane triggered by both the low
Malbec wine (MB), the dealcoholized wines at the three selected tem­
partial pressures of these components and the relatively high permeate
peratures (MB 35, MB40, MB45) and the blended wines (MB blend 35,
pressure (5 kPa).
MB blend 40, MB blend 45). A total of thirty-two free volatile organic
The results above are in accordance with those found in our work as
compounds (VOCs) including higher alcohols (HA), medium chain fatty
all dealcoholized wines showed less volatile compounds than the initial
acid ethyl esters (MCFAEE), higher alcohol acetates (HAA), medium
wine. The reduction of the total content of MCFAEE in dealcoholized
chain fatty acids (MCFA), norisoprenoids (NI) and terpenoids (TP) were
wines was 52.7, 82.8, and 90.45% at 35, 40, and 45 ◦ C respectively,
identified and quantified. In all the Malbec wines analyzed, HA and
while the total of HA showed percentages of reduction of 19.9, 45.2 and
MCFAEE were the most abundant VOCs as vividly shown in Fig. 7 for the
26.9% for the three considered variants. Additionally, it was observed
cases of the Malbec and the blend linked with the runs at 45 ◦ C.
the loss of 74–91% of HAA, approximately 66% of MCFA, 80–98% of NI,
MCFAEE, HA, HAA, MCFA and certain TP are synthetized by yeast
and 74–86% of TP for all three dealcoholized wines. Of note is the
during the alcoholic fermentation. NI and some TP are considered the
separation factor of 2.75 for ethanol achieved by the group of Nanjin
varietal aroma of wines and are related to the typicity of red and white
Tech University at a permeate pressure of 5 kPa (see Table 5 in Sun et al.
wines.
[1]) relative to the value of 3.7 achieved in our work at a permeate
Partial dealcoholization through pervaporation at the three selected
pressure of 1.3 kPa. Thus, since the membranes used in both works have
temperatures resulted in the content reduction of all VOCs with MCFAEE
active layers made of PDMS, it seems clear that other structural and
showing the highest losses compared with the other chemical groups.
chemical factors as, for example, active layer thickness or porosity, as
Table 7, which uses the mass balance of the feed tank-membrane system,
well as possible chemical modifications, could play a relevant role in the
transport of aroma compounds [48]. The operating conditions, the hy­
drodynamic conditions at the retentate side of the membrane and the
desired degree of dealcoholization are also relevant.
NP in wines are synthesized from carotenoids during grape ripening
and contribute to the varietal character of certain wine varieties. These
compounds are considering into the varietal volatile composition as well
as certain terpenoids. β-ionone contributes to the floral-like violet notes
of the wines at low concentrations (90 ng/L in model wine solution)
[49]. There are poor references about the impact of the dealcoholization
procedure on the NP or TP in wines. The loss of β-ionone was observed in
the dealcoholized wines at all three temperatures, while at 40 and 45 ◦ C,
there was a 98% of loss of β-damascenone (floral and/or fruity attri­
bute). TP in wines comes from two sources: from the grape (varietal
aroma) and yeast metabolism during alcoholic fermentation (fermen­
tative aroma) [50]. All temperature treatments showed the reduction of
these compounds, and there was a total loss of 2,7-dimethyl 4,5-octa­
dien-1-ol and nerolidol.
As can be seen, the blended wines also showed differences in the total
content of esters, acetates, acids, alcohols, terpenoids, and nor­
Fig. 7. Overall composition (μg/L) by compound family in initial MB and MB isoprenoids regarding MB. The total content of MCFAEE did not show
blend 45 wines. statistical differences at 35 and 45 ◦ C. The same behavior can be seen in

13
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

Table 7
Calculated overall values for components % losses in permeate, permeate concentrations and separation factors corresponding to dealcoholized wines.
Groups Raw and dealcoholized wines concentrations Components % losses in permeate Permeate concentrations Separation factors β

MB MB35 MB40 MB45 MB35 MB40 MB45 MB35 MB40 MB45 MB35 MB40 MB45

HA 1595.3 1404.57 960.72 1282.36 19.88 45.20 26.85 3269.50 7433.45 4416.00 2.05 4.66 2.77
MCFAEE 2015.42 1048.62 380.17 212.13 52.65 82.83 90.42 10939.96 17210.98 18787.44 5.843 8.54 9.32
HAA 322.92 91.65 30.41 33.42 74.17 91.43 90.58 2469.26 3043.78 3015.54 7.65 9.43 9.34
MCFA 310.15 117.9 106.85 119.9 65.41 68.65 64.82 2091.35 2195.02 2072.59 6.74 7.08 6.68
NI 68.99 15.46 1.31 1.4 79.61 98.27 98.15 566.20 698.95 698.10 8.21 10.13 10.12
TP 102.88 29.68 15.56 24.61 73.75 86.24 78.23 782.18 914.64 829.74 7.60 8.89 8.07

HAA and TP, while NI did not show differences between 40 and 45 ◦ C
being almost ten times less than MB blend 35 ◦ C treatment. Considering
the most abundant chemical families, the MCFAEE/HA ratio for initial
Malbec was 1.26, <1 for dealcoholized wines and higher than 1 for
blended wines; more specifically, 1.28, 1 and 1.25 for MB35, MB40 and
MB45, respectively.
Seen from the perspective of the grape spirits produced at the three
selected operating temperatures, these are characterized both by their
high ethanol content and by their aromatic character, presenting values
of the MCFAEE/HA ratio of 3.34, 2.31 and 4.25 for the variants at 35, 40
and 45 ◦ C, respectively.
Both dealcoholized and blended wines did not show the presence of
volatile compounds related to chemical changes by temperature
(oxidized aroma, hydrolysis). The increase in the temperature process
did not lead to an increase in the NI profile. NI could be present in free
and glycosylated forms in wines. Temperature affects the hydrolysis rate
of glycosylated precursors (volatile + aglycone), a behavior commonly
observed during wine aging at temperatures over 25 ◦ C. The tempera­
ture could also affect the hydrolysis ratio of esters and acetates, releasing
medium-chain fatty acids and higher alcohols, respectively [31,51].
Additionally, due to the use of nitrogen gas to reduce the impact of
oxygen during the dealcoholization process, wines did not show the
presence of oxidized aromas like furfurals [39].

4.5. Effect of dealcoholization temperature on sensory characteristics of

wines (Step 7 of Fig. 1)

Wines were subjected to descriptive sensory analysis employing

thirteen attributes: color intensity, aroma intensity, fresh red fruit,
cooked red fruit, chocolate, smoke, spicy, floral, herbaceous, sweet,
acid, bitterness, astringency, pungency, and fullness. Different attributes
were present in wines, and the final list was selected by consensus. PCA
analysis shows the spatial distribution of sensory attributes with statis­
tical significance (Fig. 8a), and the spatial distribution of all wines
evaluated (Fig. 8b). The first two components explained the 93.69% of
the total variance (dimension 1 76.73% and dimension 2 16.96%). The
non-overlapping ellipses show the statistical differences between the
treatments, while the overlapped ellipses indicate the non-statistical
differences. Thus, judges differentiated the samples by the type of
wine (initial, dealcoholized, and blend) but could not discriminate them
by the temperature.
Although the chemical analyses showed significant differences in Fig. 8. Principal component analysis (PCA) of descriptive sensory data of
total acidity, the content in each sample was not enough to be differ­ Malbec wines: (a) sensory attributes loadings, and (b) wine factor maps with
entiated by the panel of trained judges. As previously mentioned, the 95% of confidence ellipses. MB: initial Malbec (14.8% ethanol), MB 35-MB40-
phenolic compounds impact the sensory quality of wines and contribute MB45: partial dealcoholized wines (~10.5% ethanol), and MBblend35-
to color, astringency, and bitterness [52]. Astringency and bitterness MBblend40-MBblend45: blend of MB and partial dealcoholized wines
were directly associated with the tannin content. As can be seen, initial (~13.5% ethanol).
wines were perceived as more astringent and bitter than dealcoholized
and blended wines. This behavior could be explained by the higher analysis (Section 3.3, Table 5) about the C*ab parameter, showing a
contents of alcohol in wines exerting strong bitter and astringent per­ positive correlation (r2: 0.86).
ceptions [7]. As noted, these results are in agreement with the total Olfactory attributes like aroma intensity, fresh and cooked red fruit,
tannin content (Section 3.3, Table 5). The perception of the color in­ chocolate, smoke, spicy, and herbaceous were perceived in all wines but
tensity was higher in blended and dealcoholized wines compared to did not show statistical differences. For this reason, it cannot contribute
initial Malbec. These results are following the interpretation of chemical to the PCA analysis. The floral descriptor was the olfactory attribute

14
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

representative of the blended wines. It is important to take into account consumption was significantly reduced in the design variants incorpo­
that is difficult to correlate the sensory and volatile composition due to rating heat integration. The trend verified for steam consumption trig­
the synergic/antagonist effect between volatile compounds and/or gers a better environmental performance for PVHI alternatives as
volatile and non-volatile compounds [53]. This phenomenon is influ­ explained in a previous work [11].
enced by the loss of a hydrophobic environment due to the alcohol The annual cost of producing 3.3 million liters of wine with an
reduction [54]. Villamor et al. [55] evaluated the effect of ethanol on alcohol content of 10.5% v/v was, for the variants with heat integration,
twenty sensory parameters in model wines and found that high ethanol 146.8, 139.8 and 130.0 thousand US$ at 35, 40 and 45 ◦ C, respectively.
concentration reduced the perception of fruity and floral notes. On the The reported values represent costs per liter of partially dealcoholized
contrary, Baker and Ross [56] observed that high ethanol content pro­ wine (10.5% v/v) of 4.48, 4.27 and 3.97 or 1.34, 1.28 and 1.19 US$
motes the persistency and the intensity of the floral attribute of the 2- cents when considering the 10.9 million liters of wine with an alcohol
phenylethanol. Li et al. [57] demonstrated the synergic effect (addi­ content of 13.5% v/v produced by blending fresh wine with partially
tive effect) of certain esters on the other esters, terpenoids, and nor­ dealcoholized wine in a 70/30 ratio.
isoprenoids on the floral and fruity descriptors and the sweet mouthfeel. Finally, highly attractive economic indicators are presented based on
Therefore, although dealcoholization reduced the content of esters, ac­ both the sale of the permeate product at a price of 1 US$ per liter and the
etates, and higher alcohols (some of which related to the floral aroma), assumption that the cost associated with the reduction in wine volume is
the interactions between the different volatile compounds improved in absorbed by the price of the final blend product. On the one hand, the
the blended wines, which were perceived as more floral than the initial price assumed for the grape spirits produced is reasonable both for their
wine. high alcohol content (54% v/v) and their aroma character with values
The results observed could indicate that the relation 70:30 of for the MCFAEE/HA ratio of 3.34, 2.31 and 4.25 for the variants at 35,
blended wines was enough to perceive these wines differently than the 40 and 45 ◦ C, respectively. On the other hand, the quality of the blended
original, and could improve some mouthfeel attributes (reduce the wines, measured in terms of both their aroma profile and sensory
astringency, pungency, and bitter and increase the floral sensations). In characteristics, make appropriate the second assumption.
addition, it is necessary to emphasize that all wines did not show the Emphasizing the complex nature of the wine matrix, the process of
presence of brown color or undesirable volatile compounds. selecting the optimum operating temperature should include, in addi­
tion to economic and environmental aspects, those related to volatile
5. Conclusions and future work species profiles, chemical parameters and sensory analysis of the wines
obtained.
The development of a local flux model from single points experi­ The results showed the impact of partial dealcoholization on the
ments with binary ethanol–water mixtures at definite values of ethanol volatile profile, demonstrating the effect on the compounds with
composition and operation temperatures allowed obtaining parameters different chemical natures. The reduction of the total content of
that were then used to build a model of the dealcoholization process MCFAEE in dealcoholized wines through pervaporation was 52.7, 82.8,
considering the wine as a pseudo-ternary mixture composed of ethanol, and 90.45% at 35, 40, and 45 ◦ C respectively, while the total of HA
water and non-volatile solids. Experimental data of wine deal­ showed percentages of reduction of 19.9, 45.2 and 26.9% for the three
coholization compared well with results from the model of the process at considered variants. Additionally, it was observed the loss of 74–91% of
lab-scale using the adjusted parameters at local flux for the binary HAA, approximately 66% of MCFA, 80–98% of NI, and 74–86% of TP for
mixture. The model made it possible to capture the essence of the pro­ all three partially dealcoholized wines. However, the reduction in the
cess, plan the necessary repetitions to achieve the minimum volumes of alcohol content and changes in the volatile profile favored the percep­
partially dealcoholized wine required for sensory analysis, scale-up the tion of floral notes in the blended wines. Cooked fruit, chocolate,
process and make preliminary economic estimates for each of the smoked, spicy, and herbaceous notes were also common descriptors in
selected operating temperatures. all wines (initial, dealcoholized, blend) but did not carry statistical
From the conceptual modeling stage the required value of the frac­ weight for their differentiation.
tional pervaporation advance ηreq of 0.09 was obtained, as well as the Wine dealcoholization also affected the mouthfeel: the blended
average values of permeate overall fluxes at 35, 40 and 45 ◦ C; i.e., 1.02, wines exhibited less bitterness, astringency, and pungency than the
1.38 and 1.82 kg m− 2 h− 1, respectively. For all variants, a permeate initial Malbec wines. The color difference established that initial and
product with an alcohol content of 54% v/v was estimated. Experi­ blended wines could not be differentiated suggesting that, although
mental values agreed reasonably well with those obtained from the there are statistical differences in the chromatic parameters, they are not
model simulation. Regarding repetitions, it should be noted that the relevant for the human eye to differentiate them. On the other hand, the
operating times of the additional runs differed by average values of 8.4, dealcoholization favored the appearance of large polymeric pigments,
16.1 and 12.0% with respect to the values obtained through model which could positively affect the color stability of the wines. It should be
simulations at temperatures of 35, 40 and 45C; that is, 16.6, 11.9 and noted that negative olfactory and visual attributes associated with
9.6 h, respectively. Experimental tasks required about 280 h of labora­ oxidative processes were not perceived in dealcoholized and blended
tory work. wines. It is also important to highlight that the partially dealcoholized
The following points regarding both the scaling-up and economic wines could not be sensory discriminated based on the dealcoholization
analysis stages deserve to be highlighted: temperature.
The membrane area required at 35, 40 and 45 ◦ C; that is, 52, 39 and Considering the results obtained in this work, we could infer that the
26 m2 followed the trend of the average overall fluxes of 1.02, 1.38 and production of partially dealcoholized wines by pervaporation at 45 ◦ C
1.82 kg m− 2 h− 1 with operating times of 10.2, 10.1 and 11.4 h, and their mixing with fresh wine at a 70/30 ratio is very promising
respectively. taking into account the new scenarios of climate change and their
Since for all the variants analyzed the desired degree of deal­ impact on grape ripeness.
coholization was achieved for a pervaporation fractional advance value As future works, two topics of interest are raised:
ηreq of 0.09 with a permeate product with an alcohol content of 54% v/v, On the one hand, the inclusion of aspects related to the PDMS
condensation duties were practically the same (i.e., about 222 kWh/ membrane cleaning policies between batches, its long-term stability and
batch). As a result, the required permeate condenser area for each finally the sensory analysis of the alcohol-rich permeate is planned from
variant was inversely proportional to the corresponding operating time, experiments at pilot plant level.
with the lowest area requirement observed for the design at 45 ◦ C. On the other hand, the comparison of the performance of deal­
Regardless of the analyzed temperature, steam and cooling water coholization by pervaporation with those of other alternatives;

15
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

especially with that of fermentation from early harvests is also [12] J. Espinosa, E. Salomone, K. A Torres, CBD Toolkit 4.0 User Manual, Argentina,
2012.
projected.
[13] R.J. Sánchez, D.A. Figueroa Paredes, D.S. Laoretani, Y. Villada, M. Fuentes,
J. Espinosa, On the conceptual design of the hybrid nanofiltration/distillation
Funding process in the production of alcohol-free beers, Sep. Purif. Technol. 267 (2021),
118625, https://doi.org/10.1016/j.seppur.2021.118625.
[14] A.W. Sleeswijk, L.F. van Oers, J.B. Guinée, J. Struijs, M.A. Huijbregts,
This work was made in the framework of project PIP 779 “Concep­ Normalisation in product life cycle assessment: An LCA of the global and European
tual design of membrane-based hybrid processes” with the financial economic systems in the year 2000, Sci. Total Environ. 390 (1) (2008) 227–240,
https://doi.org/10.1016/j.scitotenv.2007.09.040.
support from CONICET (Argentina). [15] 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. Report I:
CRediT authorship contribution statement
Characterization, RIVM Report 2016-0104, National Institute for Public Health and
the Environment, The Netherlands.
Danilo Alexander Figueroa Paredes: Writing – original draft, [16] Instituto Nacional de Vitivinicultura: https://www.argentina.gob.ar/inv/vinos/est
Investigation, Validation, Methodology, Software, Conceptualization. adisticas.
[17] M. C. Goldner, M. C. Zamora, Sensory characterization of vitis vinifera cv. Malbec
Mariela Assof: Writing – original draft, Validation, Supervision, Re­ wines from seven viticulture regions of Argentina; Wiley Blackwell Publishing, Inc;
sources, Investigation, Formal analysis, Conceptualization. Ramiro J. Sensory Stud. 22(5) (2007) 520–532. https://doi.org/10.1111/j.1745-459X.200
Julián Sánchez: Investigation, Validation, Conceptualization. Santiago 7.00123.x.
[18] A.C. Aruani, C.I. Quini, H. Ortiz, R. Videla, M. Murgo, S. Prieto, Argentinean
Sari: Writing – review & editing, Resources, Conceptualization. Aníbal commercial Malbec wines: regional sensory profiles, J. Life Sciences ISSN 1934-
Catania: Software, Resources, Methodology. Martín Fanzone: Writing 7391, USA 8(2) (2014) 134–141.
– review & editing, Resources, Investigation, Conceptualization. José [19] D. Pérez, M. Assof, E. Bolcato, S. Sari, M. Fanzone, Combined effect of temperature
and ammonium addition on fermentation profile and volatile aroma composition
Espinosa: Writing – review & editing, Writing – original draft, Valida­ of Torrontés Riojano wine, LWT Food Sci. Technol. 87 (2018) 488–497, https://
tion, Supervision, Resources, Project administration, Investigation, doi.org/10.1016/j.lwt.2017.09.020.
Funding acquisition, Formal analysis, Conceptualization. [20] M.L. Fanzone, S.E. Sari, M.V. Mestre, A.A. Catania, M.J. Catelén, V.P. Jofré, M.
L. González-Miret, M. Combina, F. Vazquez, Y.P. Maturano, Combination of pre-
fermentative and fermentative strategies to produce Malbec wines of lower alcohol
and pH, with high chemical and sensory quality, Oeno One 54 (4) (2020)
Declaration of competing interest 1041–1058, https://doi.org/10.20870/OENO-ONE.2020.54.4.4018.
[21] V. Jofré, M. Assof, E. Vargas, M. Fanzone, Determinación simultánea de
carotenoides y compuestos fenólicos en subproductos agroalimentarios:
The authors declare that they have no known competing financial
Microextracción Líquido-Sólido Asistida por Ultrasonido optimizada por
interests or personal relationships that could have appeared to influence metodologías de diseño experimental, ICU. Investigación, Ciencia Y Universidad 4
the work reported in this paper. (5) (2020) 20–28, https://doi.org/10.59872/icu.v4i5.341.
[22] S.V. Sumbhate, S. Nayak, D. Goupale, A. Tiwari, R.S. Jadon, Colorimetric method
for the estimation of ethanol in alcoholic-drinks, J. Anal. Tech. 1 (2012) 1.
Data availability [23] Hysys User Manual, Hyprotech Ltd., Calgary, Canada, 1999.
[24] W.D. Seider, J.D. Seader, D.R. Lewin, Product and Process Design Principles:
Data will be made available on request. Synthesis, Analysis, and Evaluation, 2nd intern, John Wiley & Sons, New York,
2004.
[25] A. Fahmy, Membrane Processes for the Dehydration of Organic Compounds,
References University of Hannover, Germany, 2002. Ph.D. Thesis.
[26] G.D. Ulrich, P.T. Vasudevan, How to estimate utility costs, Chem. Eng. 113 (4)
(2006) 66–69.
[1] X. Sun, G. Dang, X. Ding, C. Shen, G. Liu, C. Zuo, X. Chen, W. Xing, W. Jin,
[27] A. Massera, M. Assof, S. Sari, I. Ciklic, L. Mercado, V. Jofré, M. Combina, Effect of
Production of alcohol-free wine and grape spirit by pervaporation membrane
low temperature fermentation on the yeast-derived volatile aroma composition and
technology, Food Bioprod. Process. 123 (2020) 262–273, https://doi.org/10.1016/
sensory profile in Merlot wines, Lwt 142 (February) (2021), https://doi.org/
j.fbp.2020.07.006.
10.1016/j.lwt.2021.111069.
[2] L. Takács, G. Vatai, K. Korány, Production of alcohol free wine by pervaporation,
[28] P.J. Linstrom, W.G. Mallard (Eds.), NIST Chemistry WebBook, NIST Standard
J. Food Eng. 78 (1) (2007) 118–125, https://doi.org/10.1016/j.
Reference Database Number 69, National Institute of Standards and Technology,
jfoodeng.2005.09.005.
Gaithersburg MD, 20899, https://doi.org/10.18434/T4D303.
[3] Brueschke et al., US Patent 5,385,647. Process for the reduction of the alcohol
[29] ISO 3591:1977. Sensory analysis — Apparatus — Wine-tasting glass.
content of alcoholic beverages, 1995.
[30] G.E. Ahumada, A. Catania, M.L. Fanzone, M.J. Belmonte, C.V. Giordano, C.
[4] M. Catarino, A. Mendes, Dealcoholizing wine by membrane separation processes,
V. González, Effect of leaf-to-fruit ratios on phenolic and sensory profiles of Malbec
Innov. Food Sci. Emerg. Technol. 12 (2011) 330–337, https://doi.org/10.1016/j.
wines from single high-wire-trellised vineyards, J. Sci. Food Agric. 101 (4) (2021)
ifset.2011.03.006.
1467–1478, https://doi.org/10.1002/jsfa.10760.
[5] N. Diban, A. Arruti, A. Barceló, M. Puxeu, A. Urtiaga, I. Ortiz, Membrane
[31] M.L. Giuffrida de Esteban, C. Ubeda, F.J. Heredia, A.A. Catania, M.V. Assof, M.
dealcoholization of different wine varieties reducing aroma losses. Modeling and
L. Fanzone, V.P. Jofre, Impact of closure type and storage temperature on chemical
experimental validation, Innov. Food Sci. Emerg. Technol. 20 (2013) 259–268,
and sensory composition of Malbec wines (Mendoza, Argentina) during aging in
https://doi.org/10.1016/j.ifset.2013.05.011.
bottle, Food Res. Int. 125 (February) (2019), https://doi.org/10.1016/j.
[6] R. Duc-Truc Pham, V.J. Ristic, D.W. Stockdale, J. Jeffery, K.W. Tuke, Influence of
foodres.2019.108553.
partial dealcoholization on the composition and sensory properties of Cabernet
[32] A. Kujawska, K. Knozowska, J. Kujawa, W. Kujawski, Influence of downstream
Sauvignon wines, Food Chem. 325 (2020), 126869, https://doi.org/10.1016/j.
pressure on pervaporation properties of PDMS and POMS based membranes, Sep.
foodchem.2020.126869.
Purif. Technol. 159 (2016) 68–80, https://doi.org/10.1016/j.seppur.2015.12.057.
[7] R. Longo, J.W. Blackman, G. Antalick, P.J. Torley, S.Y. Rogiers, L.M. Schmidtke,
[33] D.A. Figueroa Paredes, R.J. Sánchez, M. Magariño, J. Espinosa, Pervaporative
A comparative study of partial dealcoholisation versus early harvest: effects on
recovery of aroma compounds in the production of non-alcoholic beers:
wine volatile and sensory profiles, Food Chem. 261 (2018) 21–29, https://doi.org/
incorporation of different condensation strategies into the conceptual design of the
10.1016/j.foodchem.2018.04.013.
process, Sep. Purif. Technol. 290 (2022), 120821, https://doi.org/10.1016/j.
[8] C.M. Salgado, E. Fernández-Fernández, L. Palacio, F.J. Carmona, A. Hernández, P.
seppur.2022.120821.
Prádanos, Application of pervaporation and nanofiltration membrane processes for
[34] F. Huerta-Pérez, J.R. Pérez-Correa, Optimizing ethanol recovery in a spinning cone
the elaboration of full flavored low alcohol white wines, food and bioproducts
column, J. Taiwan Inst. Chem. Eng. 83 (2018) 1–9.
processing 101 (2017) 11–21, doi:10.1016/j.fbp.2016.10.001.
[35] B. Magasanik, C.A. Kaiser, Nitrogen regulation in Saccharomyces cerevisiae, Gene
[9] N. García-Martín, S. Pérez-Magariño, M. Ortega-Heras, C. González-Huerta,
290 (1–2) (2002) 1–18, https://doi.org/10.1016/s0378-1119(02)00558-9.
M. Mihnea, M.L. González-Sanjosé, L. Palacio, P. Prádanos, A. Hernández, Sugar
[36] S. Motta, M. Guaita, M. Petrozziello, A. Ciambotti, L. Panero, M. Solomita,
reduction in white and red musts with nanofiltration membranes, Desalin. Water
A. Bosso, Comparison of the physicochemical and volatile composition of wine
Treat. 27 (2011) 167–174, https://doi.org/10.5004/dwt.2021.2039.
fractions obtained by two different dealcoholization techniques, Food Chem. 221
[10] M. Margallo, R. Aldaco, A. Barceló, N. Diban, I. Ortiz, A. Irabien, Life cycle
(2017) 1–10, https://doi.org/10.1016/j.foodchem.2016.10.046.
assessment of technologies for partial dealcoholisation of wines, Sustain. Product.
[37] D.T. Pham, V.J. Stockdale, D. Wollan, D.W. Jeffery, K.L. Wilkinson, Compositional
Consumpt. 2 (2015) 29–39, https://doi.org/10.1016/j.spc.2015.07.007.
consequences of partial dealcoholization of red wine by reverse osmosis-
[11] D.A. Figueroa Paredes, R.J. Sánchez, B. Morero, M.B. Fernández, J. Espinosa,
evaporative perstraction, Molecules 24 (7) (2019) 1–16, https://doi.org/10.3390/
Enriching the conceptual modelling approach with environmental considerations:
molecules24071404.
application to the partial dealcoholization of wines, Separ. Purif. Technol. 308C
(2023) 122950, https://doi.org/10.1016/j.seppur.2022.122950.

16
D.A. Figueroa Paredes et al. Separation and Purification Technology 335 (2024) 126076

[38] M. Fanzone, F. Zamora, V. Jofré, M. Assof, A. Peña-Neira, Phenolic composition of [48] A.I. Paz, C.A. Blanco, C. Andrés-Iglesias, L. Palacio, P. Prádanos, A. Hernández,
Malbec grape skins and seeds from Valle de Uco (Mendoza, Argentina) during Aroma recovery of beer flavors by pervaporation through polydimethylsiloxane
ripening: effect of cluster thinning, J. Agric. Food Chem. 59 (11) (2011) membranes, J. Food Process Eng 40 (2017) 1–9, https://doi.org/10.1111/
6120–6136, https://doi.org/10.1021/jf200073k. jfpe.12556.
[39] N. Scrimgeour, S. Nordestgaard, N.D.R. Lloyd, E.N. Wilkes, Exploring the effect of [49] E. Tomasino, S. Bolman, The potential effect of β-ionone and β-damascenone on
elevated storage temperature on wine composition, Aust. J. Grape Wine Res. 21 sensory perception of pinot noir wine aroma, Molecules 26 (5) (2021), https://doi.
(2015) 713–722, https://doi.org/10.1111/ajgw.12196. org/10.3390/molecules26051288.
[40] K. Bindon, S. Kassara, Y. Hayasaka, A. Schulkin, P. Smith, Properties of wine [50] F.M. Carrau, K. Medina, L. Farina, E. Boido, P.A. Henschke, E. Dellacassa,
polymeric pigments formed from anthocyanin and tannins differing in size Production of fermentation aroma compounds by Saccharomyces cerevisiae wine
distribution and subunit composition, J. Agric. Food Chem. 62 (47) (2014) yeasts: effects of yeast assimilable nitrogen on two model strains, FEMS Yeast Res.
11582–11593, https://doi.org/10.1021/jf503922h. 8 (7) (2008) 1196–1207, https://doi.org/10.1111/j.1567-1364.2008.00412.x.
[41] R. Boulton, The copigmentation of anthocyanins and its role in the color of red [51] M. Wicks, A.L. Robinson, H. Heymann, Effect of simulated shipping conditions on
wine: A critical review, Am. J. Enol. Vitic. 52 (2) (2001) 67–87, https://doi.org/ sensory attributes and volatile compounds in six wines, Am. J. Enol. Vitic. 60 (3)
10.5344/ajev.2001.52.2.67. (2009) 409a–a.
[42] M. Fanzone, A. Catania, M. Assof, V. Jofré, J. Prieto, D.G. Quiroga, J.L. Sottano, [52] M. Fanzone, I. Coronado, S. Sari, A. Catania, M. Gil i Cortiella, M. Assof, V. Jofré,
S. Sari, Application of vine-shoot chips during winemaking and aging of Malbec C. Ubeda, A. Peña-Neira, Microwave-assisted maceration and stems addition in
and Bonarda wines, Beverages 7 (3) (2021) 1–20, https://doi.org/10.3390/ Bonarda grapes: effects on wine chemical composition over two vintages, Food Res.
beverages7030051. Int. 156 (March) (2022), https://doi.org/10.1016/j.foodres.2022.111169.
[43] A. Gambuti, A. Rinaldi, M.T. Lisanti, R. Pessina, L. Moio, Partial dealcoholisation of [53] L. Garcia, C. Perrin, V. Nolleau, T. Godet, V. Farines, F. Garcia, S. Caillé, C. Saucier,
red wines by membrane contactor technique: Influence on colour, phenolic Impact of acetaldehyde addition on the sensory perception of syrah red wines,
compounds and saliva precipitation index, Eur. Food Res. Technol. 233 (4) (2011) Foods 11 (12) (2022), https://doi.org/10.3390/foods11121693.
647–655, https://doi.org/10.1007/s00217-011-1553-2. [54] D.T. Pham, R. Ristic, V.J. Stockdale, D.W. Jeffery, J. Tuke, K. Wilkinson, Influence
[44] C. Romero, J. Bakker, Effect of storage temperature and pyruvate on kinetics of of partial dealcoholization on the composition and sensory properties of Cabernet
anthocyanin degradation, vitisin A derivative formation, and color characteristics Sauvignon wines, Food Chem. (2020) 325, https://doi.org/10.1016/j.
of model solutions, J. Agric. Food Chem. 48 (6) (2000) 2135–2141, https://doi. foodchem.2020.126869.
org/10.1021/jf990998l. [55] R. Villamor, M. Evans, C.F. Ross, Effects of ethanol, tannin, and fructose
[45] S. Banvolgyi, K. Savaş Bahçeci, G. Vatai, S. Bekassy, E. Bekassy-Molnar, Partial concentrations on sensory properties of model red wines, Am. J. Enol. Vitic. 64 (3)
dealcoholization of red wine by nanofiltration and its effect on anthocyanin and (2013) 342–348, https://doi.org/10.5344/ajev.2013.12118.
resveratrol levels, Food Sci. Technol. Int. 22 (8) (2016) 677–687, https://doi.org/ [56] A.K. Baker, C.F. Ross, Wine finish in red wine: the effect of ethanol and tannin
10.1177/1082013216642331. concentration, Food Qual. Prefer. 38 (2014) 65–74, https://doi.org/10.1016/j.
[46] M. Bogianchini, A.B. Cerezo, A. Gomis, F. López, M.C. García-Parrilla, Stability, foodqual.2014.05.014.
antioxidant activity and phenolic composition of commercial and reverse osmosis [57] N. Li, G. Li, A. Li, Y. Tao, Synergy effect between fruity esters and potential
obtained dealcoholised wines, Lwt 44 (6) (2011) 1369–1375, https://doi.org/ odorants on the aroma of hutai-8 rose wine revealed by threshold, s-curve, and σ− τ
10.1016/j.lwt.2011.01.030. plot methods, J. Agric. FoodChem 71 (37) (2023) 13869–13879, https://doi.org/
[47] Y.Y. Belisario-Sánchez, A. Taboada-Rodríguez, F. Marín, A. Iguaz, A. López-Gómez, 10.1021/acs.jafc.3c03733.
Aroma recovery in wine dealcoholization by SCC distillation, Food Bioproc. Tech. 5
(6) (2009) 1–11, https://doi.org/10.1007/s11947-011-0574-y.

17