4.3.4.1 Basic Principle
OD is a membrane-based separation technique with great potential for the production of high-quality concentrated fruit and vegetable juices such as orange juice, tomato juice (Durham and Nguyen, 1994), apple and grape juices (Sheng et al., 1991), pas-sion fruit juice (Vaillant et al., 2001), and other fruit juices (Girard et al., 2000a;
Petrotos et al., 2001). This process is also known as osmotic evaporation, osmotic TABLE 4.9
Determination of TSS, Density, TAC, and AC in Samples of Black Currant Juice Concentrated by Membrane distillation at ΔT = 15°C
Sample TSS (°Bx) Density (g/cm3) TAC (%)
Anthocyanin Content C (g/L)
Raw juice 15.0 1.049 3.65 1.201
MD feed 22.0 1.059 5.12 1.868
MD product 58.2 1.292 16.00 3.805
Source: Reprinted from Kozak, A. et al., Desalination, 241, 309, 2009. With permission.
concentration by membrane, membrane evaporation, isothermal MD, or gas mem-brane extraction. OD is usually operated under atmospheric pressure and at ambient temperature, thus avoiding thermal and mechanical degradation with the preservation of the aroma and flavor contents of the fresh juice. The OD process uses a micropo- rous hydrophobic membrane to separate the liquid feed (juice) by a hypertonic solu-tion (generally concentrated brine) flowing downstream a membrane. The water vapor pressure gradient across the hydrophobic membrane is maintained by lowering the vapor pressure on the downstream side relative to that on the upstream side by using a concentrated brine stripper such as salt solution of MgSO4, CaCl2, KHPO4, etc.
The hydrophobic nature of the membrane prevents infiltration of aqueous solutions to the pores, resulting in air entrapped within the membrane pores forming the gas membrane. The difference in solute concentration, and so in water activity between the two sides of the membrane, induces at the vapor–liquid interfaces a vapor pressure difference causing a diffusion of water vapor across the pores (Gostoli, 1999). The mass transfer mechanism occurs in three steps: evaporation at the dilute vapor–liquid interface, diffusion of vapor through the membrane pores from higher-vapor pressure side to the lower one, and condensation of the vapor at the membrane-brine interface.
OD is very close to the mechanism of MD, although there are some significant differ-ences. Both the processes involve sustaining a water vapor pressure gradient, that is, chemical potential gradient across the membrane pores, in order to get the thermo- dynamic force causing the diffusion process. In the case of MD, a temperature dif-ference induces a corresponding vapor pressure gradient across the membrane pores, whereas, in the case of OD, this gradient is due to the difference in the composition of the bulk liquid phases adjoining the membrane at both sides. The major problems associated with RO is the osmotic pressure limitation which restricts the concentra-tion of juice up to 25–30 °Bx, while in MD some loss of volatile aroma compounds and thermal degradation due to the heat requirement for the feed stream in order to maintain the water vapor pressure gradient. These problems can be circumvented by the use of OD with the achievable concentration levels close to the values presently obtained by evaporation.
4.3.4.2 Apparatus Design
The OD processes are carried out in batch as well as continuous mode of opera-tion for the concentration of fruit juices. In both modes of operations, the most important design parameters include (1) plant capacity in terms of volume of feed to be concentrated, (2) concentration of solute in feed and product, (3) water vapor pressure–concentration relationship for both feed and strip solution, and (4) membrane characteristics (hydrophobicity, thickness, pore size, porosity, tor-tuosity, etc.). Sheng et al. (1991) concentrated various fruit juices (orange, apple, and grape) in the range of 65–70 °Bx by OD process in a specially designed Syrinx plate and frame configuration SR-72 membrane module fitted with PTFE membrane (area: 0.7 m2; thickness: 100 μm, pore size: 0.2 μm) with 3 mm spac-ing between the membranes under batch mode of operation. An OD laboratory bench plant supplied by Hoechst-Celanese Corporation (Wiesbaden, Germany) equipped with a Liqui-Cell Extra-Flow 2.5 × 8 in. membrane contactor (Hoechst-Celanese Corporation, Wiesbaden, Germany) (Table 4.10) was used to concentrate
the clarified fruit juices such as citrus and carrot (Cassano et al., 2003), kiwifruit (Cassano et al., 2004), and cactus pear (Cassano et al., 2007). The clarified juice was recirculated through the shell side of the membrane module, while the strip-ping solution, 60 w/w% calcium chloride dehydrate solution, was recirculated in a tube side with countercurrent mode. The clarified juice was concentrated by this process up to a total soluble content of about 60 °Bx at an average throughput of about 1 kg/m2 h. The schematic of OD laboratory plant is shown in Figure 4.24.
4.3.4.3 Effect of Different Process Parameters on Permeate Flux and Retention 4.3.4.3.1 Effect of Pretreatment
It was found that the pretreatment of fresh juice (apple, kiwifruit, passion, carrot, etc.) resulted in an increase in the efficiency of OD process (Vaillant et al., 2001;
Cassano et al., 2003, 2004). The pretreatment of juice was usually carried out using enzyme, UF, MF, etc., and sometimes a combination of both. Pretreatment of grape juice by UF was shown to result in an improved flux during subsequent con-centration of permeate by OD. The flux increase was attributed to a reduction in the viscosity of the concentrated juice–membrane boundary layer due to removal of pectin, protein, etc. (Bailey et al., 2000). The enzymatic treatment followed by the UF of apple juice was shown to decrease the levels of pectin and proteins significantly, thus reducing the deposition on the hydrophobic surface and thereby membrane fouling. Such a deposition improved membrane wetting and could even-tually result in an undesired convective flow of liquid through the membrane in the OD process.
TABLE 4.10
Data Sheet of Liquid-Cell Extra-Flow 2.5 × 8 in.
Membrane Contactor
Fiber Characteristics
Fiber type Celgard microporous polypropylene hollow fiber
External diameter 300 μm
Internal diameter 220 μm
Wall thickness 0.03 mm
Porosity 40%
Length 0.16 m
Cartridge Operating Limits
Max. transmembrane differential pressure 4.2 kg/cm2 (60 psi) Max. operating temperature range 40°C (104°F) Cartridge Characteristics
Cartridge dimensions (D × L) 8 × 28 cm (2.5 × 8 in.) Effective surface area 1.4 m2 (15.2 ft2) Effective area/volume 29.3 cm2/cm3 Fiber potting material Polyethylene
4.3.4.3.2 Effect of Solute Content
Cassano et al. (2004) performed OD experiments during the concentration of kiwi-fruit juice to observe the effect of the brine concentration on the evaporation fluxes (Figure 4.25). They observed that the OD flux was significantly affected by the salt content of the brine as a 50% mass fraction reduction of CaCl2 2H2O (from 60.0 w/w%
(aw = 0.28) to 30.0 w/w% (aw = 0.80)) leads to a 70% vapor flux decline in OD flux from 1.16 to 0.35 kg/m2 h. Similar trends were also observed by Courel et al. (2000) in the concentration of sucrose solution by OD. They observed that decreasing the salt con-tent in the brine of about 30% resulted in 64% flux decline from 10.3 to 3.7 kg/m2 h.
0.025 0.2 0.4 0.6 0.8 1.0 1.2
1.4 0.9
0.8 0.7 0.6 0.5 0.4 0.3
30 35 40 45 50 55 60 650.2
xb (w/w%) Jw (kg/m2h)
Jw
awb awb
FIGURE 4.25 Evaporation flux (Jw) and water activity of the brine (awb ) vs. brine solute con-tent (xb). Operating conditions: T = 25°C; juice flow rate = 29.8 L/h; brine flow rate = 37.8 L/h.
(Reprinted from Cassano, A. et al., Food Res. Int., 37(2), 139, 2004. With permission.) 1
2
3 6 5
4 7
8 9 10
12 11 Brine solution (tube side)
Juice (shell side)
FIGURE 4.24 Schematic of OD laboratory plant. (1) Extracting solution tank, (2) brine pump, (3,5,7,8) manometer, (4) OD membrane module, (6,9) flow meter, (10) feed pump, (11) feed tank, (12) digital balance. (Reprinted from Cassano, A. et al., J. Food Eng., 57, 153, 2003. With permission.)
This result may be explained by the strong dependence of the water activity of the brine on salt content. This activity effect was more pronounced than the flux increase that could be expected from improvement of transport properties: both density and vis-cosity of the brine decreased with dilution, while the diffusion coefficient increased.
This reduction of mass transfer resistance in the salt solution rendered into a vapor flux improvement, which was in fact masked by the activity effect.
Courel et al. (2000) used OD for the concentration of sucrose solution using a flat sheet membrane. They observed that with increasing sucrose content from 0 to 65 w/w%, the OD flux decreased from 10.3 to 1.1 kg/m2 h (Figure 4.26). Unlike the salt effect, the water activity of the sugar solution decreased only by 13.2% in the range of sucrose concentration considered, and the flux decay was not attributed to a water activity effect. The viscosity of the sugar solution increased exponen-tially with the solute content while the diffusion coefficient strongly decreased. The presence of sugar at higher concentration resulted in vapor flux decay by reduc-ing the transport properties of the solutions and specifically by increasing viscosity.
During the concentration of cactus pear juice from 11 to 61.4 °Bx by OD, Cassano et al. (2007) reported that at low TSS concentration of the feed, the flux decline was more attributable to the dilution of the stripper while at higher TSS concentration, it was mainly due to exponential increase in juice viscosity with juice concentration.
Similar observations are also reported in the literature during concentration of pas-sion fruit and kiwifruit juice by OD (Vaillant et al., 2001; Cassano et al., 2004).
Vaillant et al. (2001) observed similar trends in the concentration of microfiltered passion fruit juice on an industrial scale by OD. A pilot plant equipped with a module containing 10.2 m2 of polypropylene hollow fibers was used to concentrate passion fruit juice up to TSS content higher than 60 °Bx at 30°C. An average evaporation flux of almost 0.75 kg/h was obtained with water, 0.65 kg/h when juice was concentrated to 40 °Bx, and 0.50 kg/h when it reached 60 °Bx. OD can also be carried out in a two-stage process to concentrate juice from 14 to 60 °Bx with a constant evaporation flux of around 0.62 kg/h. Two-stage process showed almost 20% saving in membrane’s surface area.
00 0
0.2 0.4 0.6
2 10 4 6 8 10 12
20 30 40 50 60 70
∆aw
∆aw
N
Xsug (w/w%)
N (kg/m–2 h–1)
FIGURE 4.26 Effect of sucrose content of the dilute solution on mass transfer: vapor flux (N) and water activity difference between sucrose solution and brine (Δaw) vs. sucrose content (Xsug).
(Reprinted from Courel, M. et al., J. Membr. Sci., 170, 281, 2000. With permission.)
4.3.4.3.3 Effect of Velocity
At high concentration level (>60 °Bx), evaporation rate was found to be strongly depen- dent on tangential velocity (Vaillant et al., 2001). With decreasing the tangential veloc-ity of juice from 0.24 to 0.09 m/s, flux declined by almost 20%. This can be explained by the fact that at higher concentration, viscosity increases exponentially with severe concentration polarization near the membrane surface, which is highly sensitive to tan-gential velocity and, consequently, to increase flux with increase in tangential velocity.
The circulation velocity of brine also has a significant effect on the improvement of OD flux. Courel et al. (2000) observed that with increasing circulation velocity of brine from 0.2 to 2.2 m/s, flux increased more than twofold. At higher velocity (>1.7 m/s), improvement was not significant. This flux increase can be attributed to the strong shear stress along the condensation side of the membrane imposed by the circulation velocity of brine. At higher circulation velocity, the effect of concentration polarization became negligible resulting in no further significant increase in flux.
4.3.4.3.4 Effect of Water Vapor Pressure Difference between Juice and Stripping Solution
The effect of operating conditions (juice flow rate, juice concentration, and tem-perature) on the OD flux during the concentration of apple, orange, and grape juice through a PTFE membrane (0.7 m2) with pore size of 0.2 μm and an overall thick-ness of 100 μm was studied by Sheng et al. (1991) under batch mode operation. Their results revealed that OD flux decreased with the increase in juice concentration, and it was shown to depend strongly on the osmotic pressure difference (Δπ) between the feed juice and the stripping solution. When decreasing Δπ from 416 atm (low juice concentration) to 280 atm (high juice concentration), that is, about 33%, a fivefold decrease in OD flux was observed. The effect of the water vapor pressure differences between juice and brine solution (ΔP) on the OD flux was also observed by Cassano et al. (2004) in the concentration of kiwifruit juice (Figure 4.27).
0.20.5 0.4 0.6 0.8 1.0 1.2
1.0 1.5 2.0 2.5
Jw (kg/m2 h)
∆P (×103 Pa)
FIGURE 4.27 Effect of water vapor pressure difference (ΔP) between juice and brine on the evaporation flux (Jw). (Reprinted from Cassano, A. et al., Food Res. Int., 37(2), 139, 2004.
With permission.)
4.3.4.3.5 Effect of OD on the Retention of Fruit Juice Components
Barbe et al. (1998) investigated the influence of different types of hydrophobic micro-porous membranes on the retention of organic volatile flavor/fragrance components during OD of Gordo grape juice and Valencia orange juice. Their study revealed that membranes with relatively large pore sizes at the surface were shown to be associated with higher organic volatiles retention per unit water removal than those with smaller surface openings. A simple model based on differences in their resistances to organic volatiles transport was proposed. Accordingly, pores with larger diameters at the membrane surface allowed greater intrusion of the liquid feed and brine streams with enhanced stagnation relative to that which occurs in membranes with smaller pore entrances. This resulted in an increase in the thickness and hence resistance of the boundary layer at the pore entrance. Shaw et al. (2001) analyzed the retention of flavor compounds during concentration of orange and passion fruit juices by OD distillation process in a pilot-scale osmotic evaporator containing 10.3 m2 of polypropylene hol-low fibers. Both juices were concentrated threefold to 33.5 and 43.5 °Bx, respectively.
Their analysis revealed that OD processing involved a loss of about 32% of volatile components in orange and about 39% in passion fruit juices. Alves and Coelhoso (2006) studied the potential of OD and MD processes for the retention of aroma compounds using model solutions containing citral and ethyl butyrate, two relevant aroma compounds of the orange juice aroma profile. A higher retention per amount of water removal was reported with the OD process. Retentions of citral and ethyl butyr-ate were about 12% and 14%, respectively, for the OD process, while 42% and 48%, respectively, for the MD process. During the concentration of UF clarified cactus pear juice (11 °Bx) by a laboratory bench plant hollow fiber module, Cassano et al. (2007) studied the potential of OD process by characterizing the concentrate in terms of TSS, total antioxidant activity (TAA), ascorbic acid, citric acid, and glutamic acid. They found the reduction of ascorbic acid and citric acid in the final retentate of the OD process of about 3.5% and 5%, respectively, with respect to UF clarified juice while the TAA and glutamic acid content in concentrate remained almost unaltered inde-pendently by the achieved level of the concentration (61 °Bx). According to Flath et al.
(1967), the compounds associated with the characteristic delicious apple-like aroma are ethyl 2-methylbutyrate, 1-hexanal, and trans-2-hexanal of which trans-2-hexanal was the most abundant volatile in the juice. The loss of trans-2-hexanal in apple juice was accelerated by increased temperature and operating time of the concentration process applied. Membrane-based concentration processes allowed to preserve trans-2-hexanal considerably, while a significant loss of trans-2-hexanal was observed in the concentration of apple juice by thermal evaporation technique. A higher retention of trans-2-hexanal was observed by Onsekizoglu et al. (2010) with the osmotic evapora-tion process (about 48%–52%) compared with MD process (about 46%–47%).