Improvement on the concentrated grape juice physico-chemical characteristics by an enzymatic treatment and Membrane Separation Processes

In this work, the improvement on the concentrated grape juice physico-chemical characteristics by using an enzymatic treatment followed by Membrane Separation Process (MSP) has been investigated. By using Novozym 33095® and Ultrazym AFP L® enzymes varying three operating parameters, the best result on the grape pulp characteristics was attained for the Novozym 33095® performed at 35 C, 15 min. and 50 mgL. In micro/ultra fi ltration processes after enzymatic pretreatment, the best performance of the MSP with high permeate fl ux value and suitable grape juice characteristics was attained using 0.05 μm membrane pore size, 1 bar pressure and 40 C treatment temperature. When reverse osmosis process is operated at 40 bar and 40 C, high soluble solid and low turbidity values are attained. An enzymatic treatment along with MSP has shown an alternative and effi cient grape juice processing system, being possible to extend to other foods.


INTRODUCTION
Innovative technologies are required to meet quality standards and other demands of the consumer market as for production and marketing of different fruit juices.In order to attend the consumer's preference such as fl avor, aroma, appearance and mouth feel, the juice industry has developed new techniques for retaining these characteristics of freshly squeezed juices in concentrate and in the reconstituted juice, but still distinguishable from fresh juice (Jiao et al. 2004).
Usually, fresh fruit juices consist largely of water (around 80%), with a high concentration of colloids that are rich on polysaccharides such as pectin, cellulose, hemicellulose, and lignin, among other substances (Vaillant et al. 2001).In particular, fresh grape juice presents an elevated acidity due to the presence of tartaric, malic and citric acids, ensuring a low pH value and equilibrium between acidic and sweet tastes (Gurak et al. 2010).Besides these characteristics, the fresh grape juice quality is also associated to a high amount of phenolic PLÍNIO R.F.CAMPOS et al. compounds that are responsible by affecting colour and astringency (Girard and Mazza 1998).Due to be a differentiated beverage with positive energetic, nutritional and bioactive effects, grape juice was also reported as benefi c to human health, reducing or preventing a wide range of diseases such as cancer (Jang et al. 1997, Thomasset et al. 2006, Gurak et al. 2010).Furthermore, technological advances on the phenolic compounds-enriched grape juice production as well as the impact of winemaking processes on phenolic extraction in wine have been reported (González-Barrio et al. 2009, Pérez-Lamela et al. 2007, Sacchi et al. 2005).Nonetheless, regarding its rich constitution, the cloudiness of the fresh fruit juice is mainly related to the presence of pectins, which are diffi cult to remove except by enzymatic treatment using pectinases (Vaillant et al. 1999, Kashyap et al. 2001).Many works on the optimization of enzymatic pretreatment for clarification of fruit juice have been reported in this regard (Lee et al. 2006, Rai et al. 2004, Sin et al. 2006).In addition, performing a pectin degradation, it is expected to reduce the membrane fouling, which is mainly caused by the colloidal constituents of the fi ltered media, resulting in a consequently drop on the fl ux in fi ltration processes (Balischi et al. 2002, Barros et al. 2003, Habert et al. 2006, Peter-Verbanets et al. 2011, Qu et al. 2012).
Many efforts have been devoted to improve methods such as freeze concentration, sublimation concentration and membranes for concentrated juice processing (Chen et al. 1993, Köseóglu et al. 1990).However, the Membrane Separation Process (MSP) including microfi ltration, ultrafi ltration and reverse osmosis is an advanced technique that has been widely applied to the dairy, food and beverage industry, allowing to clarify, concentrate, fractionate, desalt and purify fruit juice with low thermal damage to product, reduction in energy consumption and lower capital investments, because MSP is performed at low temperatures as well as it does not involve phase change for water removal (Jiao et al. 2004, Merson et al. 1980).
In last decades, traditional fi ltrations methods have been replaced by cross-fl ow microfi ltration in oenology.After the firsts trials of cross-flow microfiltration with unsuitable results on the wine quality, the development of new filtration membrane along with a better understanding of the compounds involved in the membrane fouling have brought the selection of membrane suitable for wine fi ltration.In spite of progress made, some technological and economical barriers associated to the membrane fouling are still a limitation of the widespread application of this technique (El Rayess et al. 2011).Unfortunately, even though macromolecules present in clarified fruit juice are much smaller than the pore size of typical microfi ltration membranes, they cause signifi cant fouling (Czekaj et al. 2000).
The purpose of this paper is to study the improvement of the clarifying and concentrating grape juice quality concomitant with a minimization of the membrane fouling when a selected enzyme and better operating conditions are used in an integrated filtration system.Two different commercial enzymes (pectin lyase and polygalacturonase) were tested as a pretreatment step of the grape pulp in order to reduce the fouling phenomenon.Selection of the best enzyme along with fi nding better experimental condition were performed on the basis of the total acidity, soluble solids ( o Brix), color, total solids, and turbidity of the grape pulp as indicators of grape juice quality.For micro/ultra fi ltration processes, an inorganic tubular ceramic membrane operated under crossfl ow mode was used to assess the infl uence of the processing parameters (pressure, temperature and pore size) on permeate fl ux and quality of clarifi ed grape juice.By using a reverse osmosis unit, the improvement on the quality of concentrated grape juice was also assessed.In addition, the fouling mechanism was evaluated for designing more effi cient fi ltration processes for the fruit industry.

COLLECTION AND PREPARATION OF THE PULP
In Brazil, the most common grape used to the manufacture of grape juice is the American species Vitis labrusca, commercially known as Isabel grape.A great amount of Isabel grape was purchased from a local harvest in the region of the city of Toledo, located in the Brazilian Paraná State.In lab, all grapes were washed, removed all type of particulates on them and submitted to a pulping by an industrial mechanical pulper.In addition, a mixture of grape pulps was performed and immediately distributed in several samples of 15 kg, packing in clean plastic bags and stored at -4 o C for posterior analysis and treatment.

TESTED ENZYMES
For a previous enzymatic treatment to MSP, both pectin lyase (Novozym 33095) and cellulase polygalacturonase (Ultrazym AFP L®) enzymes were tested.In order to assess the performance of both enzymes, concentration values of 50, 100 and 150 mgL -1 were chosen, considering temperature values within the best pectolytic activity region (35 and 45 o C for Novozym 33095 and 25 -35 o C for Ultrazym AFP L), according to the manufacturer manual (Novozymes Manual 2001).

PHYSICO-CHEMICAL ANALYSES
As response physic-chemical parameters on the grape juice quality, the titratable acidity (TA), soluble solids, color, total solids, and turbidity were considered for non-and treated grape pulp samples as well as clarified and concentrated grape juice samples.For pH determination, a pH-meter (Digimed, model DM20) was used.Total solids were determined as AOAC (AOAC 1990).Color of grape juice samples coming from pulping and treatments was related to the light absorbance at a 440 nm wavelength and determined by using a spectrophotometer UVvis (Shimadzu, model HACH DR/2010).By using the same spectrophotometer and setting the light absorbance at 860 nm, the turbidity of each grape juice sample was measured.A titrimetric method (AOAC 1990), consisting in titrating 10 mL sample with 0.1 N NaOH to pH 8.1 and using phenolphthalein as indicator, was applied to measure the TA as expressed in g tartaric acid per 100 mL of sample.Concentration of soluble solids (in o Brix) was determined with a Shimadzu Abbe type refractometer Model 3L.

ENZYMATIC TREATMENT
Based on a completely randomized design with 24 experiments in triplicates for each enzyme, a set of enzymatic experiments was performed for assessing the best experimental condition.According to the enzymatic treatment procedure, reported by Balischi et al. (2002), samples of 100 mL grape pulp placed in 250 mL-Erlenmeyer fl asks were stirred in a thermostatic bath.Regarding two temperature for each enzyme, grape pulp samples were submitted to enzymatic treatment by adding concentrations of 50, 100 and 150 mg enzyme L -1 .Such treatments were performed at pectolytic activity temperatures (35 and 45 o C for the Novozym 33095® and 25 and 35 o C for the Ultrazym AFP L®) during adequate treatment times, ranging from 15 to 90 min at intervals of 15 min.After each enzymatic treatment, the thermostatic bath temperature was then increased to 70 o C at least 20 minutes for the enzymatic deactivation.Control experiments without adding enzyme were also included.After performing all enzymatic treatments, aliquots of each treated sample were collected and analyzed, obtaining a set of response physic-chemical parameter data for each condition of treatment.In order to highlight the signifi cant statistical difference between multiples mean values of response physic-chemical parameters, F-PLÍNIO R.F.CAMPOS et al.
and Tukey tests were applied to the experimental data and their results validated by ANOVA.

MSP TREATMENTS
Regarding the best enzymatic pretreatment experimental condition for the Novozym enzyme, a completely randomized design with several experiments was applied to all treatments of micro/ ultrafi ltration.In order to retain the particles of the grape juice after enzymatic treatment, a set of 18 micro/ultrafi ltration experiments in triplicates were performed by using an ultrafi ltration pilot unit (Netzsch, model 027.06-1C1/07-0005/AI).Such pilot unit consists of a feed tank of 5 L connected to a tubular module with 25 cm length and 0.7 cm internal diameter, being used ceramic membranes (α-Al2O3/TiO2).This MSP unit was operated under cross-fl ow mode with membranes of 0.05, 0.1 and 0.2 μm and 4.17 m/s maximum crossfl ow velocity.By using a pumping, enzymatically pretreated sample was introduced to the MSP module that was operated at three pressures (1, 2 and 3 bar) and two temperatures (30 and 40 o C), determining the permeate fl ux and analyzing the permeated grape juice for posterior treatment by reverse osmosis.Applying the Tukey test and ANOVA to the micro/ultrafi ltration experimental data, significant statistical differences between multiples mean values of response physic-chemical parameters were highlighted.
Clarified grape juice samples, which were obtained from the micro/ultrafi ltration unit under its better experimental condition, were considered to perform experiments based on the reverse osmosis process.A module of reverse osmosis consisting in a feed tank with a 15-L effective volume connected to a tubular module with 25 cm length and 0.7 cm internal diameter, being used composed fi lm membranes of spiral type with 63.5 mm diameter and 355.6 mm length (FILMTEC, model BW30-2514).Samples from the feed tank were pumped into tubular module by a 4 HP motor, being the temperature and pressure of samples controlled.
The reverse osmosis module was operated at a pressure of 40 bar and two temperatures (30 and 40 o C), determining the permeate fl ux and collecting the concentrated grape juice for physic-chemical analysis.

DETERMINATION OF THE FOULING MECHANISM
By regarding the transient build-up of a yielded particulate layer of cake type on the membrane upstream interface in membrane processes, the permeate flux is negatively affected.Related to such a phenomenon, named as concentration polarization, the tendency is to occur a drastically reduction on the permeate fl ux at early fi ltration stage and drive gradually towards to a steady or nearly steady-state limit value after a long flux reduction.The physico-chemical interactions of the layer of rejected particulates with the membrane have been referred to another aspect of concentration polarization phenomenon.Another adversely phenomenon causing fouling at the interface is related to the adsorption on the membrane pore walls and pore plugging by solute penetrate (Barros et al. 2003).The membrane fouling is considered the main disadvantage of MPS application, due to the frequent cleanings or replacements of membranes and especially by increased operating costs caused by the higher power consumption, caused by the reduced permeate fl ux along time (Giorno et al. 1998).
Regarding constant pressure blocking fi ltration laws, earlier proposed by Hérmia (Hérmia 1982) who applied to power-law non-Newtonian fl uids, and reformulated by Field et al. (1995) for crossfl ow microfi ltration in critical fl ux, the decay on the permeate fl ux in cross-fl ow fi ltration at constant pressure was proposed by Barros et al. (2003), being mathematical expressed by the Eq. 1. Attributing a suitable value for the general index (n) in Eq. 1 a kind of fouling mechanism involved during the fi ltration process could be evidenced (Barros et al. 2003).A complete blocking of membrane pores is expected to n value equal 2. Beside this, an internal blocking of pores might occur as n assumes a value of 1.5.When n value is equal to 1, a partial blocking of pores is expected, while for n value equal to 0 a cake type layer is formed.
( ) where J is the permeate fl ux, J* is the critical fl ux and t is the time.As reported by Todisco et al. (1996), k and n are phenomenological coeffi cient and general index, respectively, both depending on fouling mechanism.
A stochastically optimized global method, called a Particle Swarm Optimization (PSO), was applied to search the best modeling parameters (k and n) through a non-linear fitting of the experimental data.The basic principle of the PSO method is to seek a set of potential solutions located in a wide search hyperspace that is randomly scanned under different kinematic conditions of bird flocking according to some considerations based on local (c 1 ) and global (c 2 ) accelerations and swarm inertia (ω).An insight into these better values was reported earlier in other works (Espinoza-Quiñones et al. 2009, Trigueros et al. 2010a, b).Best results of attaining the near-global solution have been reported when both global and local collective accelerations have assumed the same value and are equal to 1.5 (Módenes et al. 2012, Trigueros et al. 2012).In this work, an initial particle swarm (at least 500) is defi ned as well as the number of iterations (at least 25) in order to scan a wide search hyperspace where potential solutions are identifi ed and stored.The performance of each particle is related to a built-in objective function (OF), determined by the least square statistical method, as shown in Eq. 2 of the present work.Besides these parameters, the critical fl ux (J*) was included as a parameter to be determined during the search procedure.PSO method was implemented in the software Maple 14 ® , and executed within a Windows 7 environment, using a microcomputer Intel ® CoreTM i7-930, 2.8 GHz and 8 GB RAM.
2 exp exp 1 where exp J is the fl ux value obtained experimentally and pred J is the fl ux value predicted by the model.

GRAPE PULP CHARACTERIZATION
Analyzing the characteristics of the grape pulp sample in nature, values of 13.9 g tartaric acid per 100 mL of grape juice, 11.0 o Brix, 31,000 mg Pt-Co L -1 , 3.1, 7.7 and 8,700 FAU for titratable acidity, soluble solids, color, pH, total solids and turbidity, respectively, were found.The Brazilian standard for the grape juice quality is not well defined because it depends on achieving a set of specifi c characteristics such as fl avor, taste, and appearance for fi nal acceptance by the consumer.In general, these characteristics are mainly related to the grape origin, applied treatment systems to produce grape juice among factors, being possible to manufacture a great variety of grape juice.
In this work, an improvement on the grape juice quality by applying an enzymatic pretreatment followed by two Membrane Separation Processes was monitored by titratable acidity, total solids, soluble solids, color, pH and turbidity.However, some initial characteristics, such as pH of grape pulp samples in nature, are advisable to be maintained unchangeable.Meanwhile other characteristics, for instance, high concentrations of total solids and turbidity, are recommendable to reduce their values to lower ones after a series of grape pulp treatments, by applying more effective and low-cost processes, searching a grape juice of high quality along with a high acceptance by the consumer.PLÍNIO R.F.CAMPOS et al.

ENZYMATIC PRETREATMENT ANALYSIS
The most effective enzyme and the best experimental conditions for the enzymatic treatment were searched on the analysis of the physicochemical parameter data obtained at three enzyme concentrations, and two pectolytic activity temperatures for each enzyme, regarding treatment time range from 0 to 90 min.From a completely randomized experimental design, the set of results for the titratable acidity, soluble solids ( o Brix), color, pH, total solids and turbidity is obtained and summarized in Tables I and II as Novozym 33095 and Ultrazym AFP L enzymes, respectively, were used.According to the null hypothesis test, all response physico-chemical parameter (titratable acidity, total solids, soluble solids, color, pH and turbidity) data have followed normal distributions.In addition, the application of F-and Tukey tests on the experi-mental data have been performed, showing very similar results (data not shown) related to the comparison between multiple mean values of response parameters (RP).For this reason, results of the Tukey test are only being reported in the present work.The Tukey test was ran within the software SAS®, version 9.1, introducing as criterion of comparison among all multiple mean values of RP within a 95% confi dence level.All Tukey results were validated by ANOVA (data not shown), providing the least mean value for each RP as well as allowing to highlight the best experimental condition for the set of enzymatic treatments.
Performing the Tukey test analysis of RP data obtained for the two tested enzyme types at the two temperatures for better pectolic activity, three enzyme concentrations, and seven treatment times, the lowest RP values were attained as Novozym 33095 enzyme is used at 35 o C temperature, 15 min.treatment time, and 50 mgL -1 concentration.Under the best experimental condition, the enzymatically treated grape juice was characterized by the lowest values of 12.82 g tartaric acid per 100 mL of grape juice, 11.0 o Brix, 31,550 mg Pt-Co L -1 , 3.1 and 8,700 FAU for titratable acidity, soluble solids, color, pH and turbidity, respectively.A set of enzymatically treated grape juice samples was posterior submitted to experimental designs for assessing micro, ultra fi ltration and reverse osmosis processes.

MEMBRANE SEPARATION PROCESS ANALYSIS
Regarding the set of treatments of enzymatically treated grape juice samples based on micro and ultra filtration processes within a completely randomized experimental design, a set of mean values of permeate fl ux (in kg m -2 h -1 ) and response physico-chemical parameters (RP) was obtained, as summarized in Table III.The fi ltration system unit was operated at three pressures (1, 2 and 3 bar), two temperatures (30 and 40 o C) and by using three tubular ceramic membranes with pore diameter of 0.2 and 0.1 μm, for micro fi ltration, and 0.05 μm, for ultra fi ltration.It can be noticed that an increasing on the pore size from 0.05 to 0.2 μm has resulted in a decreasing permeate fl ux for both considered temperatures.A change on the temperature of the treated grape juice at high pressure has driven to a strong decay on the permeate fl ux when the ultra filtration system was operated with lower membrane pore size.Besides this, an improvement on the permeate fl ux value was achieved when low pressure and small pore size were used.It could be explained by the polarization effect present in the cake layer formation.By analyzing the behavior of the permeate fl ux with the mean pore size and applied pressure, a reduction on the permeate fl ux was observed with an increasing on both parameter pore size and pressure values (see Table III).Experimental permeate fl ux data obtained from a micro and ultra fi ltration unit were fi t by the model described by Eq. 1.The PSO method was applied in order to search the globally optimized values for the phenomenological coeffi cient (k) and the general index (n).All PSO results are shown in Table IV, while the behavior of the permeate fl ux values as function of time in ultrafi ltration process, performed at 40 °C, 0.05 μm pore size and three pressures (1, 2 and 3bar), is shown in Fig. 1.
As a function of the solid/solute size and shape in relation to the membrane pore size, several types of modes are expected to occur, according to the n-value.According to Field et al. (1995) the critical fl ux corresponds to the permeate fl ux before the fouling, i.e., it consists on the highest permeate fl ux for which no fl ux reduction in time is observed.Regarding a temperature of 30 o C in processes of micro and ultra fi ltration, it can be    noticed that the same mode of fouling associated to a cake layer formation (n=0) is expected to be present, regardless of tested membrane pore size and pressure values.However, considering a temperature of 40 o C, it was observed a systematic progress in the fouling mechanism mode, appearing fi rst a cake layer formation (n=0) for low pressure value (1 bar) and ending with a complete blocking of membrane pores (n=2) for high pressure value (3 bar).In addition, as a consequence of an increasing on the pore size an increasing on the k value and a reduction of the critical fl ux (J*) value are expected.
A similar response is expected to occur when an increasing on the temperature is considered, except to pore size of 0.05 μm for which the results for permeate fl ux were kept unaltered.The pressure effect on the complete blocking of membrane pores become more evident when greater membrane pore sizes are used.Although the permeate fl ux is positively favored as a consequence of a reduction of the viscosity at high temperatures, a complete blocking of pores could occur when the pressure is increased.Relatively scarce information on detailed studies of fouling mechanisms caused by polysaccharides and polyphenols is found in literature (Czekaj et al. 2000).In earlier works (Belleville et al. 1990(Belleville et al. , 1992)), performing microfi ltration of red wine, membrane fouling has been attributed to high levels of polysaccharides and polyphenols.In ultrafi ltration processes the interaction between inorganic particles and biopolymers has resulted in a fouling cake of signifi cantly reduced porosity (Jermann et al. 2008).An irreversible fouling by organic matter takes place due to internal pore adsorption, affecting negatively the ultrafi ltration process (Katsoufi dou et al. 2005).In the case of the use of enzymatically treated grape juice samples, the pectin component might also agglomerate other particulates forming a fouling cake.Furthermore, there is a greater contribution in the transversal fl ux with an increasing on the pressure, reducing thus the permeate fl ux.As a consequence of an increasing on the temperature, the permeate fl ux is positively favored due to a reduction on the viscosity value of the grape juice, allowing carrying on particulates with high feasibility through a less viscous medium.
A statistical analysis of micro and ultra fi ltration processes was performed by the Tukey test, aiming to fi nd out the least mean value for each RP as well as allowing highlighting the best experimental condition among all tested MSP treatments.According to the Tukey test, the best result for the permeate fl ux (136.38 kg m -2 h -1 ), among all tested MSP conditions, was attained by using a membrane with pore diameter of 0.05 μm under a pressure of 1 bar, at 40 o C (see Table III).Regarding the same experimental condition, grape pulp sample in nature without a previous enzymatic treatment was also tested, exhibiting undoubtedly a lower mean permeate fl ux (65.16 kg m -2 h -1 ).An enzymatic treatment prior to Membrane Separation Process becomes an important pretreatment step, allowing reducing the impact of insoluble particles and suspended solids onto membrane interface with an increasing on the MSP performance along with a suitable clarifi ed grape juice.At least one of set of operating variables has affected signifi cantly  the RP values.Nonetheless, taking altogether, a reduction on the RP values was attained by using the membrane of 0.05 μm, under a pressure of 1 bar, regardless of temperature.With regard to the mean permeate fl ux and physico-chemical parameters, the best operating condition was verified when using the membrane of 0.05 μm, at 40 o C, under the pressure of 1 bar.By using the clarifi ed grape juice, originating from the ultrafi ltration (0.05 μm, 40 o C, 1 bar), in the reverse osmosis module, operated at a pressure of 40 bar, the permeate fl ux exhibited values of 8.51 and 4.65 kg m -2 h -1 for temperatures of 30 and 40 o C, respectively, suggesting that low temperatures are recommended to be used for improving the permeate flux.In addition, the quality of the concentrated grape juice was characterized by the set of fi ve physico-chemical parameters.For a temperature of 30 o C, values of 22.4 g tartaric acid per 100 mL of grape juice, 18.4 o Brix, 10,600 mg Pt-Co L -1 , 3.62, and 900 FAU were obtained for titratable acidity, soluble solids, color, pH, and turbidity, respectively.Meanwhile for a temperature of 40 o C, values of 23.8 g tartaric acid per 100 mL of grape juice, 20.4 o Brix, 13,000 mg Pt-Co L -1 , 3.43, and 780 FAU were obtained for titratable acidity, soluble solids, color, pH, and turbidity, respectively.Two grape juice characteristics have been improved at temperature of 40 o C, showing an increasing of the soluble solids and a reduction on the turbidity, reinforcing also the color.
In comparison, the value of soluble solid of grape juice that was pretreated with enzymes followed by two Membrane Separation Processes is twice above those values attributed to three Brazilian commercial grape juices (see Table V).In addition, another positive characteristic was a lower turbidity value than that for commercial grape juices.As a whole, an enzymatic treatment along with micro/ultrafi ltration and reverse osmosis has shown a great performance on the production of concentrated grape juice.

CONCLUSIONS
According to the assessment of enzymatically treated grape pulp characteristics by using the Tukey test an ANOVA, the Novozym 33095 enzyme has shown better result on response physicchemical parameters than Ultrazym AFP L® one as experiments were performed at 35 o C pectolytic activity temperature, 15 min.treatment time and 50 mgL -1 enzyme concentration.The permeate fl ux of enzymatically treated-grape juice, which were submitted to micro/ultra fi ltration processes, has shown a behavior depending on the pressure, pore size and temperature.The pressure effect on permeate fl ux has become more evident on high pressures.A decay on the permeate fl ux is started with a cake layer formation at 1 bar pressure (0.05 μm) and ended with a complete blocking of membrane pores at 3 bar (0.2 μm), according to a tested fouling mechanism model.An increasing on the temperature has caused an improvement on the permeate fl ux due to a reduction on the medium viscosity.High temperature could also contribute to reinforce the fouling mechanism, blocking easily small membrane pores.Nonetheless, the best performance of the MSP with high permeate fl ux value and suitable grape juice characteristics was attained using 0.05 μm membrane pore size, 1 bar pressure and 40 o C treatment temperature.An improving on the grape juice characteristics was observed at 40 o C, with an increasing on the amount of soluble solids and a reduction on the turbidity, reinforcing thus the color.Based on an enzymatic pretreatment of grape pulp followed by MSP, desirable characteristics of the processing grape pulp could be maintained, undesirable characteristic could be reduced by microfi ltration and others could be reinforced by reverse osmosis, suggesting that this is an alternative and potential grape juice processing system for application on other types of foods.

Figure 1 -
Figure 1 -Behavior of the permeate fl ux values as function of time in ultrafi ltration processes performed at 40 °C, 0.05 μm pore size and pressures of a) 1, b) 2 and c) 3 bar, along with the respective fi ts of the tested model, proposed by Field et al. (1995).