**3. Membrane process**

Membrane filtration is a process used to separate dissolved substances and fine particles from solutions. Membrane acts as a semipermeable and selective barrier that separates particles based on molecular or physical size. Solutes smaller of solution than the membrane pore size are able to pass through the membrane as permeate flux while particles and

30 5000

COD (mgO2.L-1)

> 1800 150 550

100 000-200 000

3 000 7 800

5 000-7 000 4 000-14 000 4 000-20 000

BOD5 (mgO2.L-1)

> 8 100 1 400

> 4 500

2 500 6 500

42 000 860

(m3ton-1) of product

0.3 1.7 0.7

5.0 1.2

0.7 1.5

90 1.5

Alcohol plant 900-1 200

Beer production 4.2 2 500 1 800

\*adapted from 1Iaquinta et al., 2009; 2Noronha et al., 2002; 3Mantzavinos & Kalogerakis, 2005; 4Madaeni & Mansourpanah, 2006; 5 Matthiasson, 1983; 6Kuca & Szaniawska, 2009; 7Walha et al., 2009; 8Scharnagl

Primary and secondary treatments are often used to decompose the high organic contents of

After of traditional treatment of wastewater, general requirements are covered by regulations of each country, usually complemented by consent limits based on avoidance of pollution. Discharge licenses may include maxima for flow, temperature, suspended solids, dissolved solids, BOD5, nitrogen, phosphorous and turbidity. According at quality of water, in most cases, final disposal of treated waste water is into a water course where it will be diluted by the existing flow. However, subsequently one advanced process of effluent treating can be an option desirable to recycle water within a factory of food processing.

Membrane filtration is a process used to separate dissolved substances and fine particles from solutions. Membrane acts as a semipermeable and selective barrier that separates particles based on molecular or physical size. Solutes smaller of solution than the membrane pore size are able to pass through the membrane as permeate flux while particles and

wastewater of food industry by aerobic and anaerobic fermentation processes.

Feed processing Wastewater

Meat processing - Scalding tube -Chiller showers - Cooling tanks


Vegetable processing -frozen carrots -Olive mill3

Potato starch -Shower Starch rinsing

Fish industry - Unloading fish5

Dairy industry -Whey



et al., 2000; 9Gésan-Guiziou et al., 2007

**3. Membrane process** 

Table 1. Wastewater from food industry


Fruit juice -Orange -Apple

molecules larger than the membrane pore size are retained. The two fluxes at outlet of membrane are important because this process has a high efficiency in the separation.

The majority of commercial membranes are made usually of organic polymers (polysulfones and polyamides) and inorganic materials (ceramic membranes based on oxides of zirconium, titanium, silicium and aluminum).

The membranes are implemented in several types of modules. The membrane configuration determines the manner in which the membrane is packed inside the modules. Four main types of membrane configurations are used in the industry. These are: plate-and-frame, spiral wound, tubular and hollow-fiber configurations. The membrane geometry is planar in the first two and cylindrical in the two others. Figure 1 shows schematically a typical hollow fiber module (Okokchina, 2010).

The membrane system is operated in a cross-flow feed mode. The concentrated stream passes parallel to the membrane surface as opposed to perpendicular flow that is used traditionally in filtration. This operating mode allows that accumulation of solute molecules at the membrane surface decreases and the permeate flux remains constant for a long time due to decreased hydrodynamic resistance at the membrane surface by cross-flow induced hydraulic turbulence. Flow direction is usually inside-out, i.e. the concentrate flux inside the fibers and the permeate flux is collected at the shell-side. It is often possible to reverse the flow (outside-in) for cleaning and unclogging of the membrane. Cylindrical configuration provides the possibility of maintaining high tangential velocity in the feed stream and is therefore particularly suitable for applications where the feed contains a high proportion of suspended solids or must be strongly concentrated.

The choice for a certain kind of membrane system is determined by a great number of aspects, such as costs, risks of plugging of the membranes, packing density and cleaning opportunities. The effects of the feed properties, the membrane properties, and the filtration conditions are obviously very important for the success of a membrane filtration process. Principal limitation of membrane lies in membrane fouling which is mainly associated with the deposition of a biosolids cake layer onto the membrane surface (McCutcheon & Elimelech, 2006; Mi & Elimelech, 2008). However, everal alternatives have been implemented to enhance this problem (Al-Akoum et al., 2002; Jaffrin et al., 2004).

#### **3.1 Membrane applications in food industry for wastewater treatment**

Membrane separation process has special recognition in food wastewater treatment, applied to the end of conventional treatment systems (Vourch et al., 2008). The process is used

membrane and compounds with molecular weights less than the MWCO will pass through the membrane as permeate. Table 3 shows size range of particles retained with range of

**Retained solutes** 

Bacteria, fat, oil, grease, colloids, organics microparticles

Proteins, pigments, oils, sugar, organics microparticles

Pigments, sulfates, divalent cations, divalent anions, lactose, sucrose, sodium chloride

and inorganic ions

**Application in effluents treatment of food industry** 

251

Oil, Cereal, Dairy, Beverage

Dairy, Cereal, Oil, Tomato puree, Beer, Wine, Fish, Meat, Pickled vegetables

Olive oil, Dairy, Beverage, Meat canning, Pickled vegetable

Dairy, Cereal, Fish, Meat, Pickled vegetables

MWCO membranes for treatment of wastewater of food industry.

**Retained diameters particle (µm range)** 

RO 0.2-2 10-4- 10-3 Salts, sodium chloride

by membrane process in wastewater treatment of food industry.

membrane affect rejection (Guizard & Amblard, 2009).

Table 3. Typical range of application of MWCO, diameter particle and retained solutes type

Retention is obviously affected by the pore size due to the sieving effect, especially when using MF and UF membranes. With tighter (NF and RO) membranes retention will be governed more and more by the electrostatic forces as well as by other interactions between membranes and solutes. Thus MWCO is only a rough indication of the membrane's ability to remove a given compound as molecular shape because polarity and interaction with the

Respect to pore diameter, it has frequently been seen that the membrane with the most open pores does not usually give the highest permeate flux in filtration process. Porosity (ratio of void space to total membrane volume in porous membrane) and pore size distribution may influence the apparent size of particles retained. Typical microporous membranes have average porosities in the range 30%–70%. Porosity can also be measured by analyzing processed images obtained from microscopic analyses such as scanning electron microscopy (SEM). Figure 2 shows SEM image of an asymmetric porous structure of a ceramic membrane. It may be noted that the membrane has fine pores through which raw water is filtered (Figure 2a). The most of ceramic membrane elements are constructed from

Carbon macroporous material is used as support for ceramic membrane deposition (Figure 2b and 2c). Multiple layers are usually resulting from residual spaces created between ceramic particles during sintering. The bottleneck geometry is representative of pores resulting from sintering of almost spherical particles, for example, this is the case of porous structures obtained with titania, zirconia (Guizard et al., 2002; Guizard & Amblard, 2009). The porous sites are uniformly distributed in the membrane and effective diameter of the membrane pore can be determined assuming pores are circular in shape. However, pore geometry (tortuosity; *τ*) can also affect the retention of molecules by a membrane. Tortuosity reflects the length of the average pore compared to the membrane thickness. Cylindrical

supported multiple ceramic layers constituting an asymmetric porous structure.

**MWCO membrane (kilo Daltons range)** 

MF 100-500 10-1-10

UF 20-150 10-3 – 1

NF 2-20 10-3 – 10-2

**Membrane Process** 

primarily to reduce the volume of the food wastewater that is achieved by recovering of two fluxes: permeate water flux having the majority of the original volume, and concentrated flux in a lesser volume (constituents of effluents retained).

The membranes used in food wastewater treatment differ widely in their structure and function. Mainly they are operated in four membrane processes: microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO). Solvent permeability and separation selectivity are the two main factors characterizing at these membranes. Transport mechanisms and operating membrane conditions can also explain the pass of species through membranes. Particle size is practically the sole criterion for describer the permeation or rejection of membranes. However, microporous membranes (NF and RO) have ability of separate particles at molecular level and their selectivity is mainly based on the chemical nature of the species.

Several works have been focused on these factors to explain separation selective of residues of food wastewater. Effluents treatment of dairy industry by RO and NF membranes are reported in many investigations, however, a strong development and growth of membrane technology can be observed in the results from the other food industries (Turano et al., 2002). Food industry standards specify that, spent process water intended for reuse (even for cleaning purposes) must be at least of drinking quality. Regulations for other applications, such as boiler make-up water or warm cleaning water, are even more stringent. There has been a study on the possibilities for reuse of vapor condensate in a milk processing company (dried milk production) as boiler make-up water (Hafez et al., 2007), and the reuse of chiller shower water in a meat processing company (sausage production) as warm cleaning water (Mavrov & Bélières, 2000).

#### **3.2 Membrane characteristics**

Generally membranes are characterized by pore flow or molecular weight of particle that is retained or is filtered by the membrane. However, important membrane properties such as structure, porosity, thickness, wettability surface and operating conditions, are also studied because affect rejection of solutes. The electrostatic repulsion between the membrane surface and the contaminant may be particularly analyzed to enhance waste solute retention and to increase water flux.

The smallest particle size present in the feed is very important for the selection of membrane pore size. However, currently the feed properties can be changed by pretreatments such as pH adjustment, thermal treatment, addition of chemicals, and pre-filtration. The pH adjustment (Luo et al., 2010) and thermal treatment can decrease the precipitation of certain substances. In addition, chemicals can be added to the feed to increase the particle size through aggregation, and the retention of specific substances can be enhanced through micellation or complexation (Wu et al., 2007). The salt concentration of the feed and the valence of the salt present can also be important to select membrane type (Muro et al., 2009; Lefebvre & Moletta, 2006)

#### **3.2.1 Pore-flow and material membranes**

Membrane pore flow is differentiated by the size of particles diameter that they can separate (micrometers, µm) and nominal molecular weight cutoff MWCO (kilo Daltons), which is a performance-related parameter, defined as the lower limit of a solute molecular weight for which the rejection is 95-98% (Boerlage et al., 2004). In theory, compounds having a molecular weight greater than the molecular weight cut off (MWCO) will be retained by the

primarily to reduce the volume of the food wastewater that is achieved by recovering of two fluxes: permeate water flux having the majority of the original volume, and concentrated

The membranes used in food wastewater treatment differ widely in their structure and function. Mainly they are operated in four membrane processes: microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO). Solvent permeability and separation selectivity are the two main factors characterizing at these membranes. Transport mechanisms and operating membrane conditions can also explain the pass of species through membranes. Particle size is practically the sole criterion for describer the permeation or rejection of membranes. However, microporous membranes (NF and RO) have ability of separate particles at molecular level and their selectivity is mainly based on

Several works have been focused on these factors to explain separation selective of residues of food wastewater. Effluents treatment of dairy industry by RO and NF membranes are reported in many investigations, however, a strong development and growth of membrane technology can be observed in the results from the other food industries (Turano et al., 2002). Food industry standards specify that, spent process water intended for reuse (even for cleaning purposes) must be at least of drinking quality. Regulations for other applications, such as boiler make-up water or warm cleaning water, are even more stringent. There has been a study on the possibilities for reuse of vapor condensate in a milk processing company (dried milk production) as boiler make-up water (Hafez et al., 2007), and the reuse of chiller shower water in a meat processing company (sausage production) as

Generally membranes are characterized by pore flow or molecular weight of particle that is retained or is filtered by the membrane. However, important membrane properties such as structure, porosity, thickness, wettability surface and operating conditions, are also studied because affect rejection of solutes. The electrostatic repulsion between the membrane surface and the contaminant may be particularly analyzed to enhance waste solute retention and to

The smallest particle size present in the feed is very important for the selection of membrane pore size. However, currently the feed properties can be changed by pretreatments such as pH adjustment, thermal treatment, addition of chemicals, and pre-filtration. The pH adjustment (Luo et al., 2010) and thermal treatment can decrease the precipitation of certain substances. In addition, chemicals can be added to the feed to increase the particle size through aggregation, and the retention of specific substances can be enhanced through micellation or complexation (Wu et al., 2007). The salt concentration of the feed and the valence of the salt present can also be important to select membrane type (Muro et al., 2009;

Membrane pore flow is differentiated by the size of particles diameter that they can separate (micrometers, µm) and nominal molecular weight cutoff MWCO (kilo Daltons), which is a performance-related parameter, defined as the lower limit of a solute molecular weight for which the rejection is 95-98% (Boerlage et al., 2004). In theory, compounds having a molecular weight greater than the molecular weight cut off (MWCO) will be retained by the

flux in a lesser volume (constituents of effluents retained).

the chemical nature of the species.

**3.2 Membrane characteristics** 

increase water flux.

Lefebvre & Moletta, 2006)

**3.2.1 Pore-flow and material membranes** 

warm cleaning water (Mavrov & Bélières, 2000).

membrane and compounds with molecular weights less than the MWCO will pass through the membrane as permeate. Table 3 shows size range of particles retained with range of MWCO membranes for treatment of wastewater of food industry.


Table 3. Typical range of application of MWCO, diameter particle and retained solutes type by membrane process in wastewater treatment of food industry.

Retention is obviously affected by the pore size due to the sieving effect, especially when using MF and UF membranes. With tighter (NF and RO) membranes retention will be governed more and more by the electrostatic forces as well as by other interactions between membranes and solutes. Thus MWCO is only a rough indication of the membrane's ability to remove a given compound as molecular shape because polarity and interaction with the membrane affect rejection (Guizard & Amblard, 2009).

Respect to pore diameter, it has frequently been seen that the membrane with the most open pores does not usually give the highest permeate flux in filtration process. Porosity (ratio of void space to total membrane volume in porous membrane) and pore size distribution may influence the apparent size of particles retained. Typical microporous membranes have average porosities in the range 30%–70%. Porosity can also be measured by analyzing processed images obtained from microscopic analyses such as scanning electron microscopy (SEM). Figure 2 shows SEM image of an asymmetric porous structure of a ceramic membrane. It may be noted that the membrane has fine pores through which raw water is filtered (Figure 2a). The most of ceramic membrane elements are constructed from supported multiple ceramic layers constituting an asymmetric porous structure.

Carbon macroporous material is used as support for ceramic membrane deposition (Figure 2b and 2c). Multiple layers are usually resulting from residual spaces created between ceramic particles during sintering. The bottleneck geometry is representative of pores resulting from sintering of almost spherical particles, for example, this is the case of porous structures obtained with titania, zirconia (Guizard et al., 2002; Guizard & Amblard, 2009). The porous sites are uniformly distributed in the membrane and effective diameter of the membrane pore can be determined assuming pores are circular in shape. However, pore geometry (tortuosity; *τ*) can also affect the retention of molecules by a membrane. Tortuosity reflects the length of the average pore compared to the membrane thickness. Cylindrical

potential can be expected to be strongly negative even at low pH values (pH 2–3); while when the membrane contains weakly acidic groups, the zeta potential can be expected to become more negative from the point the groups start to dissociate to the point where the groups are totally dissociated. Similarly, strongly basic groups give positive potentials in most of the pH range, while weakly basic groups have no positive charges at pH values

The isoelectric point (IP) (pH where net charge is zero) of a membrane is also a referent to determinate the behavior of their surface charge, depending on the pH of the wastewater in contact with the membrane. (Cheng et al., 2008). For example, typically NF polimeric membranes are negatively charged at neutral pH, with IP around pH 3-4, while ceramic

The IP of a membrane can be evaluated from the pH dependence of the zeta potential (Martín et al., 2003). However other experiments can also describe this parameter. Figure 3

Fig. 3. Isoelectric point of a ceramic membrane of UF. (a) pH of permeate water is measured during an operation time range. (b) pH differential is determinate when pH feed solution is adjustment at 4-8 range, intersection of line with horizontal axe denotes isoelectric point at 6.2. (c) The values of zeta potential are measured in dependence with pH of ionic solution. Figure 3a, denotes pH determination of pure water during water filtration by membrane of UF. In this membrane the pH value of permeate does not change with operation time. It shows that isoelectric point is around pH 6. Figure 3b shows the pH differential permeate water dependent of pH feed solution. Intersection of line with horizontal axe denotes isoelectric point at pH 6.2 for the same ceramic membrane. Figure 3c shows the values of zeta potential and pH of ionic solution; the values were measured through the pores of ceramic membrane. Intersection line with pH axis, indicates that isoelectric point is 6.2.

pH feed solution (b)

123456789

**4. Influence of engineering aspects in food industrial wastewater treatment**  Systematic studies of membrane applications in food wastewater treatment are focused on membrane functionality and performance filtration, under different operation condition. Several researches are specifically focused on optimizing crossflow hydrodynamics and/or membrane filter geometry to increment performance of water flux and maximum rejection or recovering species from effluents. Hydrodynamic factors affecting the membrane

) and transmembrane pressure (TMP). Permeate


Zeta potential

0 2 4 6 8 10

253

pH ionic solution (c)

shows isoelectric point of a ceramic membrane of zirconium and titanium oxide.

higher than 8 (Kim et al., 2005).

membranes have a IP around pH 6-7

0 25 50 75 100 125 150

Time (min)

(a) -1.5


pH diferential

5.5 5.7 5.9 6.1 6.3 6.5

pH water permeate flux

functionality, are cross-flow velocity (

pores at right angles to the membrane surface have a tortuosity of one, that is, the average length of the pore is the membrane thickness (Cho et al., 2000; Zhao et al., 2000; Vrijenhoek et al., 2001).

Fig. 2. SEM image of cross section of a ceramic membrane porous structure of MF, with cut off of 300 kDa and 5 µm pore size, used in wastewater treatment of food industry (From Escobar. PhD thesis, Institute Technological of Toluca, México, 2010)

Chemical composition, hydrophilicity/hydrophobicity, charge, and morphology have also significantly effect on permeability and stability of the membrane (Khayet et al., 2005). Particularly, ceramic membranes have a composite structure, which is used to increase the permeability for small pore size membranes by decreasing the overall hydraulic resistance (Peng et al., 2005; Yu et al., 2006) while polymeric membranes can be modified to make them more hydrophilic and achieve less fouling and better cleaning efficiency.

#### **3.2.2 Surface pore charge. Isoelectric point**

Membrane charge affects membrane efficiency in food wastewater treatment, particularly when low cutoff membranes are used for treatment effluents with high salts concentration. The charging occurs due to, for instance, dissociation of functional groups, adsorption of ions from solution, and adsorption of polyelectrolytes, ionic surfactants, and charged macromolecules. Generally, membrane materials carry a negative charge or are modified to have a negative charge because natural organic matter in water is negatively charged at neutral pH, due to phenolic and carboxylic functional groups (Kaeselev et al., 2002). A negatively charge membrane, therefore, prevents rapid deposition of foulants on the membrane surface by charge repellence An increase in the flux of a relatively dense membrane at a high pH may result from an increase in membrane hydrophilicity due to the dissociation of the functional groups in the membrane structure (Schaep & Vandecasteele, 2001; Zhao et al., 2005). Many polymeric membranes are amphoteric, having both negatively and positively charged functional groups in the polymer matrix. Ceramic membranes can also show in water amphoteric behavior and thus their surface charge is pH dependent (Cho et al., 2000).

Membrane charge, as well as hydrophilicity property, can be predicted based on known membrane chemical structure. However, membrane surface/pore charge can be measure by electrical potential (Martín et al., 2003). When the membrane contains strongly acidic groups, the dissociation of the groups occurs immediately at a low pH, and the zeta

pores at right angles to the membrane surface have a tortuosity of one, that is, the average length of the pore is the membrane thickness (Cho et al., 2000; Zhao et al., 2000; Vrijenhoek

(a) (b) (c) Fig. 2. SEM image of cross section of a ceramic membrane porous structure of MF, with cut off of 300 kDa and 5 µm pore size, used in wastewater treatment of food industry (From

Chemical composition, hydrophilicity/hydrophobicity, charge, and morphology have also significantly effect on permeability and stability of the membrane (Khayet et al., 2005). Particularly, ceramic membranes have a composite structure, which is used to increase the permeability for small pore size membranes by decreasing the overall hydraulic resistance (Peng et al., 2005; Yu et al., 2006) while polymeric membranes can be modified to make them

Membrane charge affects membrane efficiency in food wastewater treatment, particularly when low cutoff membranes are used for treatment effluents with high salts concentration. The charging occurs due to, for instance, dissociation of functional groups, adsorption of ions from solution, and adsorption of polyelectrolytes, ionic surfactants, and charged macromolecules. Generally, membrane materials carry a negative charge or are modified to have a negative charge because natural organic matter in water is negatively charged at neutral pH, due to phenolic and carboxylic functional groups (Kaeselev et al., 2002). A negatively charge membrane, therefore, prevents rapid deposition of foulants on the membrane surface by charge repellence An increase in the flux of a relatively dense membrane at a high pH may result from an increase in membrane hydrophilicity due to the dissociation of the functional groups in the membrane structure (Schaep & Vandecasteele, 2001; Zhao et al., 2005). Many polymeric membranes are amphoteric, having both negatively and positively charged functional groups in the polymer matrix. Ceramic membranes can also show in water amphoteric behavior and thus their surface charge is pH dependent (Cho

Membrane charge, as well as hydrophilicity property, can be predicted based on known membrane chemical structure. However, membrane surface/pore charge can be measure by electrical potential (Martín et al., 2003). When the membrane contains strongly acidic groups, the dissociation of the groups occurs immediately at a low pH, and the zeta

Escobar. PhD thesis, Institute Technological of Toluca, México, 2010)

more hydrophilic and achieve less fouling and better cleaning efficiency.

**3.2.2 Surface pore charge. Isoelectric point** 

et al., 2001).

et al., 2000).

potential can be expected to be strongly negative even at low pH values (pH 2–3); while when the membrane contains weakly acidic groups, the zeta potential can be expected to become more negative from the point the groups start to dissociate to the point where the groups are totally dissociated. Similarly, strongly basic groups give positive potentials in most of the pH range, while weakly basic groups have no positive charges at pH values higher than 8 (Kim et al., 2005).

The isoelectric point (IP) (pH where net charge is zero) of a membrane is also a referent to determinate the behavior of their surface charge, depending on the pH of the wastewater in contact with the membrane. (Cheng et al., 2008). For example, typically NF polimeric membranes are negatively charged at neutral pH, with IP around pH 3-4, while ceramic membranes have a IP around pH 6-7

The IP of a membrane can be evaluated from the pH dependence of the zeta potential (Martín et al., 2003). However other experiments can also describe this parameter. Figure 3 shows isoelectric point of a ceramic membrane of zirconium and titanium oxide.

Fig. 3. Isoelectric point of a ceramic membrane of UF. (a) pH of permeate water is measured during an operation time range. (b) pH differential is determinate when pH feed solution is adjustment at 4-8 range, intersection of line with horizontal axe denotes isoelectric point at 6.2. (c) The values of zeta potential are measured in dependence with pH of ionic solution.

Figure 3a, denotes pH determination of pure water during water filtration by membrane of UF. In this membrane the pH value of permeate does not change with operation time. It shows that isoelectric point is around pH 6. Figure 3b shows the pH differential permeate water dependent of pH feed solution. Intersection of line with horizontal axe denotes isoelectric point at pH 6.2 for the same ceramic membrane. Figure 3c shows the values of zeta potential and pH of ionic solution; the values were measured through the pores of ceramic membrane. Intersection line with pH axis, indicates that isoelectric point is 6.2.

#### **4. Influence of engineering aspects in food industrial wastewater treatment**

Systematic studies of membrane applications in food wastewater treatment are focused on membrane functionality and performance filtration, under different operation condition. Several researches are specifically focused on optimizing crossflow hydrodynamics and/or membrane filter geometry to increment performance of water flux and maximum rejection or recovering species from effluents. Hydrodynamic factors affecting the membrane functionality, are cross-flow velocity () and transmembrane pressure (TMP). Permeate

0 2 *i*

Where Pi is pressure at the inlet of the membrane module; P0 is pressure at the outlet of the

The permeate flux depends directly on the applied TMP for a given surface area under uniform operational conditions. The flux of the pure water is linearly pressure dependent. However when food wastewater is treated by membrane system the flux is more complex. The behavior depends of wastewater composition, membrane type and crossflow velocity. In food wastewater treatment, one has to keep in mind that the permeate flux will be determined by the combination of crossflow velocity and TMP, due to contaminants (Sarkar

Figure 4a and 4b show the effect of crossflow velocity and TMP on permeate flux using two membranes of different MWCO (300 kDa and 15 kDa). The experiments were performed by Escobar, 2010. The results indicated that the flux enhancement caused by increasing crossflow velocity was particularly pronounced at range values of the TMP (3-5 bar) and crosflow velocity of 3 ms-1. Fouling occurred over a range of TMPs of 5-6 bar and crossflow velocities at 3.5 ms-1. The permeate flux decreased with time during the development of the fouling layer, but once the fouling layer was established, the permeate flux became constant for a given set of experimental conditions. Therefore these results indicate that at moderate values of TMPs and high flow rates at the membrane surface are operating conditions that conduce at high permeate fluxes in these experiments. Besides, figure 4c shows an overall positive effect of enhanced flow hydrodynamic conditions (TMP = 4 bar) on the average permeate flux, although in the turbulent regime (Re>3,000) a weaker correlation and more data scattering were observed. Therefore a clear correlation between the 3 h flux and Re in

Fig. 4. Effect of crossflow velocity and TMP on the 3 h permeate flux in wastewater treatment of a cereal industry using membranes of MF and UF (a) 300 kDa. (b) 15 kDa. (c)

Flux (Lh-1m-2)

The interdependence between average flux and hydrodynamic conditions for two

membranes in a wide range of Re numbers at TMP = 4 bar (From Escobar 2010. PhD thesis,

3.5 ms-1 3 ms-1

012345678

TMP (bar) (b)

0.5 ms-1

1.5 ms-1 Average flux at 3h (Lm-2h-1)

0 2000 4000 6000

300 kDa 15 kDa

255

Reynolds (c)

Particularly, operational membrane conditions in wastewater treatment show moderate TMP and high flow rates at the membrane surface are conducive of high permeate fluxes in

membrane module and Pp is permeate pressure.

the transient regime (Re<3000) could be expected.

0.5 ms-1 2.5ms-1 1.5 ms-1

Institute Technogical of Toluca).

01234567

TMP (bar) (a)

3ms-1

Flux ( Lh-1m-2)

et al., 2006; Blöcher et al., 2002; Oktay et al., 2007; Avula et al., 2009).

*p*

*P P TMP <sup>P</sup>* (2)

flux can increase or decrease due to simultaneous influence of these variables. Temperature, dilution and pH are also variables involved in the membrane efficiency in membranes filtration. Permeate flux increases with increasing feed temperature due to a decrease in viscosity and/or due to an increase in solubility of suspended solids (Galambos et al., 2004). The exception is the presence of calcium and magnesium salts that might precipitate when temperature is increased. This problem can be avoided at least in some cases through feed pretreatment (Sarkar et al., 2006). The pH has a significant influence on the permeation rate especially around the isoelectric point of certain colloids where they tend to destabilize and precipitate. It also has an effect because of the changes in surface charge of the membrane either due to the amphoteric nature of the surface or due to the specific adsorption of species as presented earlier (Vourch et al., 2008).

#### **4.1 Cross-flow velocity**

A hydrodynamic variable of membranes in cross-flow filtration systems is essentially the velocity at which the feed flow is passed across the surface of the membrane. Crossflow velocity ( ) is linear velocity (m/s-1) of the feed flow circulating tangentially across the membrane. This parameter is described by relation of feed flow rate (Qw; m3/s-1) and the cross sectional area of feed membrane (As; m2).

Turbulent flow conditions are recommended to maintain the flow tangential to the membrane, thereby reducing the phenomenon of concentration polarization and, consequently, the accumulation of solute near the membrane and inducing acceptable permeate flux for long time. Shear effects induce hydrodynamic filtration of the particles from the boundary layer back into the bulk, with a positive effect on the permeate flux. However, as feed concentration increases, it becomes more difficult to maintain a high recirculation velocity due to an increase in feed viscosity (Muro et al., 2009). In addition, if foods waste water containing macromolecular solutions with flexible solutes, thus also a high velocity can cause deformation of the polymer chains, which favors certain macromolecules that pass through the pores.

The hydrodynamics flow can also be characterized by calculating the Reynolds (Re) number by equation (1).

$$\text{Re} = \nu \frac{d\_h}{\mu} \tag{1}$$

Where is crossflow velocity, dh hydraulic diameter of membrane module and µ the dynamic viscosity of fluid.

Normally, Re>2100 guarantees a turbulent flow in the module and a minimum thickness for the concentration polarization layer. Prevention of reversible fouling layer formation is sufficiently achieved by a crossflow velocity of around 2.0 ms-1 in UF membranes (McKeown et al., 2005). In practical applications, one has to keep in mind that the permeate flux will be determined by the combination of crossflow velocity and TMP (See Figure 5).

#### **4.2 Transmembrane pressure**

The driving force for transport behind membrane process MF, UF, NF and RO, is the pressure difference between feed and permeate flux of the membrane (TMP; bar, psi). TMP is defined as the difference in pressure between the filtrate side of the membrane and the permeate side of the membrane. The average TMP is in general calculated as follows:

254 Membrane Separation Process in Wastewater Treatment of Food Industry 261

260 Food Industrial Processes – Methods and Equipment

flux can increase or decrease due to simultaneous influence of these variables. Temperature, dilution and pH are also variables involved in the membrane efficiency in membranes filtration. Permeate flux increases with increasing feed temperature due to a decrease in viscosity and/or due to an increase in solubility of suspended solids (Galambos et al., 2004). The exception is the presence of calcium and magnesium salts that might precipitate when temperature is increased. This problem can be avoided at least in some cases through feed pretreatment (Sarkar et al., 2006). The pH has a significant influence on the permeation rate especially around the isoelectric point of certain colloids where they tend to destabilize and precipitate. It also has an effect because of the changes in surface charge of the membrane either due to the amphoteric nature of the surface or due to the specific adsorption of species

A hydrodynamic variable of membranes in cross-flow filtration systems is essentially the velocity at which the feed flow is passed across the surface of the membrane. Crossflow

membrane. This parameter is described by relation of feed flow rate (Qw; m3/s-1) and the

Turbulent flow conditions are recommended to maintain the flow tangential to the membrane, thereby reducing the phenomenon of concentration polarization and, consequently, the accumulation of solute near the membrane and inducing acceptable permeate flux for long time. Shear effects induce hydrodynamic filtration of the particles from the boundary layer back into the bulk, with a positive effect on the permeate flux. However, as feed concentration increases, it becomes more difficult to maintain a high recirculation velocity due to an increase in feed viscosity (Muro et al., 2009). In addition, if foods waste water containing macromolecular solutions with flexible solutes, thus also a high velocity can cause deformation of the polymer chains, which favors certain

The hydrodynamics flow can also be characterized by calculating the Reynolds (Re) number

Re *<sup>h</sup> <sup>d</sup> <sup>µ</sup>* 

Normally, Re>2100 guarantees a turbulent flow in the module and a minimum thickness for the concentration polarization layer. Prevention of reversible fouling layer formation is sufficiently achieved by a crossflow velocity of around 2.0 ms-1 in UF membranes (McKeown et al., 2005). In practical applications, one has to keep in mind that the permeate flux will be determined by the combination of crossflow velocity and TMP (See Figure 5).

The driving force for transport behind membrane process MF, UF, NF and RO, is the pressure difference between feed and permeate flux of the membrane (TMP; bar, psi). TMP is defined as the difference in pressure between the filtrate side of the membrane and the permeate side of the membrane. The average TMP is in general calculated as follows:

is crossflow velocity, dh hydraulic diameter of membrane module and µ the

(1)

) is linear velocity (m/s-1) of the feed flow circulating tangentially across the

as presented earlier (Vourch et al., 2008).

cross sectional area of feed membrane (As; m2).

macromolecules that pass through the pores.

**4.1 Cross-flow velocity** 

by equation (1).

dynamic viscosity of fluid.

**4.2 Transmembrane pressure** 

Where

velocity (

$$TMP = \frac{P\_i + P\_0}{2} - P\_p \tag{2}$$

Where Pi is pressure at the inlet of the membrane module; P0 is pressure at the outlet of the membrane module and Pp is permeate pressure.

The permeate flux depends directly on the applied TMP for a given surface area under uniform operational conditions. The flux of the pure water is linearly pressure dependent. However when food wastewater is treated by membrane system the flux is more complex. The behavior depends of wastewater composition, membrane type and crossflow velocity.

In food wastewater treatment, one has to keep in mind that the permeate flux will be determined by the combination of crossflow velocity and TMP, due to contaminants (Sarkar et al., 2006; Blöcher et al., 2002; Oktay et al., 2007; Avula et al., 2009).

Figure 4a and 4b show the effect of crossflow velocity and TMP on permeate flux using two membranes of different MWCO (300 kDa and 15 kDa). The experiments were performed by Escobar, 2010. The results indicated that the flux enhancement caused by increasing crossflow velocity was particularly pronounced at range values of the TMP (3-5 bar) and crosflow velocity of 3 ms-1. Fouling occurred over a range of TMPs of 5-6 bar and crossflow velocities at 3.5 ms-1. The permeate flux decreased with time during the development of the fouling layer, but once the fouling layer was established, the permeate flux became constant for a given set of experimental conditions. Therefore these results indicate that at moderate values of TMPs and high flow rates at the membrane surface are operating conditions that conduce at high permeate fluxes in these experiments. Besides, figure 4c shows an overall positive effect of enhanced flow hydrodynamic conditions (TMP = 4 bar) on the average permeate flux, although in the turbulent regime (Re>3,000) a weaker correlation and more data scattering were observed. Therefore a clear correlation between the 3 h flux and Re in the transient regime (Re<3000) could be expected.

Fig. 4. Effect of crossflow velocity and TMP on the 3 h permeate flux in wastewater treatment of a cereal industry using membranes of MF and UF (a) 300 kDa. (b) 15 kDa. (c) The interdependence between average flux and hydrodynamic conditions for two membranes in a wide range of Re numbers at TMP = 4 bar (From Escobar 2010. PhD thesis, Institute Technogical of Toluca).

Particularly, operational membrane conditions in wastewater treatment show moderate TMP and high flow rates at the membrane surface are conducive of high permeate fluxes in

membrane permeability because of hydraulic resistance increment by the fouling phenomena. Increment of crossflow velocity, dilution of wastewater, change of temperature of feed and using turbulence promoters such as backflow techniques, feed pulsation and rotation of filter elements, are hydrodynamic methods to increment permeate flux and

The best measure of the ability of a membrane to separate molecules (i) of wastewater, is the ratio of their permeability αj, called the membrane selectivity, which can be written in terms

> *ip if C C*

(5)

257

(6)

i

Cip is concentration of specie (i) in the permeate flux and Cif is the concentration of specie (i)

The selectivity of a membrane depends on its ability to transmit different species to different extents. Factors that affect solute transmission are solute type, membrane type, solution pH, solution ionic strength, the permeate flux, and the hydrodynamic conditions on the feed side. Membrane selectivity is most often expressed as the membrane retention, R, toward the species to be separated. R is dimensionless parameter, with variation range of 0-100 %.

> <sup>R</sup> i 1 *if ip if C C <sup>C</sup>*

MF/150 140 148 23.0 25.1 16.2 16.2 6.4 9.9 UF/15 133 135 22.2 20.1 13.1 13.0 5.7 2.1

Rejection of neutral organic solutes generally increases with the molecular weight (or diameter) of the solute. Species will be retained by the membrane according to their size (sieving effect). For a mixture of multivalent and monovalent co-ions in the feed, mulitivalent co-ions are retained due to their higher electrical charge, while a part of monovalent co-ions pass through the membrane with counter ions to fulfill charge equilibrium criterion on both sides of the membrane (Lefebvre et al., 2003). However, the absolute values of the salt rejection vary over a wide range; the ranking for the different salts is the same for all membranes (Rautenbach & Albrecht, 1989). A high TMP value also affects the selectivity of some ions species. Table 4 shows the effect of TMP conditions on permeability of some ions by two ceramic membranes (Muro et al., 2009). Ions were identified in wastewater of a food industry. The experiments were performed to determine

Table 4. Effect of (TMP) on the permeability of some ions by MF y UF membranes

**Ions concentration (mgL-1) in water permeate**  Na1+ K1+ Ca2+ Fe3+ TMP (bar) 4 5 4 5 4 5 4 5

reduce the hydraulic resistance due to fouling (Jaffrin et al., 2004; Luo et al., 2010).

**4.4 Selectivity factor** 

in the feed flow.

**Membrane/Cutt off (kDa)** 

of the apparent sieving coefficient:

the MF and UF. An increase in TMP is required to maintain a particular water flux (constant-flux operation) independently of the membrane type and MWCO. However, an increasing flux could lead to an increase in polarization and fouling, which will limit the permeate flux (Abbasi et al., 2011; Simate et al., 2011).

High pressure can also allow membrane compaction, ultimately resulting in the formation of a denser membrane with smaller pores, or one possible enlargement of membrane pores with time, which enables particles to penetrate through the membrane matrix. Choi et al., (2005) showed clearly that pore sizes are modified in the membrane matrix increased with increasing TMP.

#### **4.3 Permeate flow rate**

The functionality of a membrane in wastewater treatment is determined by water permeation capacity and retention of solutes. Although permeate flux depends of the characteristics of the membrane and quality of wastewater, the average pore size and poresize distribution is important since it will give an indication of which transport mechanism can be expected to be dominant for a given specie mixture in a defined material and at given process conditions. There are two theory models to describe the mechanism of permeation in membrane process; one is the solution-diffusion model, in which permeants are diffused through the membrane down a concentration gradient. The other model is the pore-flow model, in which permeants are transported by pressure-driven convective flow through tiny pores. Separation in this case, occurs by excluding of some particles of the pores in the membrane. Fick´s law describes the mass flux through an area perpendicular to the flow direction (Miyoshi, 1998):

$$\frac{dV\_i}{Adt} = \mathbf{J}\_{\text{pi}} = -D\_i K\_i \frac{dC\_i}{d\mathbf{x}} \tag{3}$$

Where Jpi is the linear fluid velocity (ms-1) of component (i) or permeability flux (Lm-2 h-1). The diffusion coefficient Di (ms-1) reflects the mobility of individual molecules in membrane material and the molecule sorption coefficient Ki reflects the number of molecules dissolved in the membrane material. The product DiKi is membrane permeability and is a measure of the membrane's ability to permeate species. dCi/dx is the concentration gradient (molL-1) for component (i) over the length x (m). Vi is the volume of substance (i) transferred (L), t is time (h) and A is perpendicular area (m2).

Permeability flux Jpi = Vi/At is obtained by equation integration (3) and applied for dx = x (membrane thickness or membrane resistance for the pure water transport). Ci0 and Cif are the concentration of component (i) on the feed side and concentration of component (i) on the permeate side respectively. Solution-diffusion model is often used to describe the transport in RO membranes.

$$\frac{V\_i}{At} = -L\_p \frac{\Delta P}{\chi} \tag{4}$$

Lp is the hydraulic permeability coefficient (Lm-2 bar h-1); ∆P is gradient pressure TMP (bar) in membrane system. Information about porous structure and viscosity of the filtrated liquid is contained in Lp factor.

Membrane resistance (x) is a measure of the hydraulic resistance to flow through a pore channel. However, when wastewater is fed, increment of TMP can cause a decreasing of membrane permeability because of hydraulic resistance increment by the fouling phenomena. Increment of crossflow velocity, dilution of wastewater, change of temperature of feed and using turbulence promoters such as backflow techniques, feed pulsation and rotation of filter elements, are hydrodynamic methods to increment permeate flux and reduce the hydraulic resistance due to fouling (Jaffrin et al., 2004; Luo et al., 2010).

#### **4.4 Selectivity factor**

262 Food Industrial Processes – Methods and Equipment

the MF and UF. An increase in TMP is required to maintain a particular water flux (constant-flux operation) independently of the membrane type and MWCO. However, an increasing flux could lead to an increase in polarization and fouling, which will limit the

High pressure can also allow membrane compaction, ultimately resulting in the formation of a denser membrane with smaller pores, or one possible enlargement of membrane pores with time, which enables particles to penetrate through the membrane matrix. Choi et al., (2005) showed clearly that pore sizes are modified in the membrane matrix increased with

The functionality of a membrane in wastewater treatment is determined by water permeation capacity and retention of solutes. Although permeate flux depends of the characteristics of the membrane and quality of wastewater, the average pore size and poresize distribution is important since it will give an indication of which transport mechanism can be expected to be dominant for a given specie mixture in a defined material and at given process conditions. There are two theory models to describe the mechanism of permeation in membrane process; one is the solution-diffusion model, in which permeants are diffused through the membrane down a concentration gradient. The other model is the pore-flow model, in which permeants are transported by pressure-driven convective flow through tiny pores. Separation in this case, occurs by excluding of some particles of the pores in the membrane. Fick´s law describes the mass flux through an area perpendicular to the flow

> pi J *i i i i*

Where Jpi is the linear fluid velocity (ms-1) of component (i) or permeability flux (Lm-2 h-1). The diffusion coefficient Di (ms-1) reflects the mobility of individual molecules in membrane material and the molecule sorption coefficient Ki reflects the number of molecules dissolved in the membrane material. The product DiKi is membrane permeability and is a measure of the membrane's ability to permeate species. dCi/dx is the concentration gradient (molL-1) for component (i) over the length x (m). Vi is the volume of substance (i) transferred (L), t is

Permeability flux Jpi = Vi/At is obtained by equation integration (3) and applied for dx = x (membrane thickness or membrane resistance for the pure water transport). Ci0 and Cif are the concentration of component (i) on the feed side and concentration of component (i) on the permeate side respectively. Solution-diffusion model is often used to describe the

*<sup>i</sup>*

*p <sup>V</sup> <sup>P</sup> <sup>L</sup> At x*

Lp is the hydraulic permeability coefficient (Lm-2 bar h-1); ∆P is gradient pressure TMP (bar) in membrane system. Information about porous structure and viscosity of the filtrated

Membrane resistance (x) is a measure of the hydraulic resistance to flow through a pore channel. However, when wastewater is fed, increment of TMP can cause a decreasing of

*Adt dx* (3)

(4)

*dV dC D K*

permeate flux (Abbasi et al., 2011; Simate et al., 2011).

increasing TMP.

**4.3 Permeate flow rate** 

direction (Miyoshi, 1998):

time (h) and A is perpendicular area (m2).

transport in RO membranes.

liquid is contained in Lp factor.

The best measure of the ability of a membrane to separate molecules (i) of wastewater, is the ratio of their permeability αj, called the membrane selectivity, which can be written in terms of the apparent sieving coefficient:

$$\alpha\_i = \frac{\mathcal{C}\_{ip}}{\mathcal{C}\_{if}} \tag{5}$$

Cip is concentration of specie (i) in the permeate flux and Cif is the concentration of specie (i) in the feed flow.

The selectivity of a membrane depends on its ability to transmit different species to different extents. Factors that affect solute transmission are solute type, membrane type, solution pH, solution ionic strength, the permeate flux, and the hydrodynamic conditions on the feed side. Membrane selectivity is most often expressed as the membrane retention, R, toward the species to be separated. R is dimensionless parameter, with variation range of 0-100 %.

$$\mathbf{R}^\* = \frac{\mathbf{C}\_{if} - \mathbf{C}\_{ip}}{\mathbf{C}\_{if}} = \begin{pmatrix} 1 - \alpha\_i \end{pmatrix} \tag{6}$$


Table 4. Effect of (TMP) on the permeability of some ions by MF y UF membranes

Rejection of neutral organic solutes generally increases with the molecular weight (or diameter) of the solute. Species will be retained by the membrane according to their size (sieving effect). For a mixture of multivalent and monovalent co-ions in the feed, mulitivalent co-ions are retained due to their higher electrical charge, while a part of monovalent co-ions pass through the membrane with counter ions to fulfill charge equilibrium criterion on both sides of the membrane (Lefebvre et al., 2003). However, the absolute values of the salt rejection vary over a wide range; the ranking for the different salts is the same for all membranes (Rautenbach & Albrecht, 1989). A high TMP value also affects the selectivity of some ions species. Table 4 shows the effect of TMP conditions on permeability of some ions by two ceramic membranes (Muro et al., 2009). Ions were identified in wastewater of a food industry. The experiments were performed to determine

the wall shear force. For soluble species and fine colloids, the critical flux can be considered as the flux below which the wall concentration does not initiate fouling (Cho & Fane, 2000). Choi et al., 2005). High capacity of the concentration of species from wastewater can also be achieved in this regime and the critical flux may either be identical to the clean water flux at the same TMP (Hwang et al., 2006). However, outside the limiting flux, operation at sustained permeability and selectivity is not possible due to the accumulation and compaction of the fouling layer on the membrane. Finally flux decline in time-dependent with high pressure above the critical TMP, are identified in regime III due to increment membrane fouling. Their removal is necessary for stable membrane operation (Espinasse et

Fig. 6. Critical flux regimes in membranes of 300, 150 and 15 kDa: (I) Subcritical operation

**6. Membrane fouling control in food industry for wastewater treatment** 

Fouling is the most important issue affecting the development of membrane filtration-as it worsens membrane performance and shortens membrane life (Boerlage et al., 2004). Membrane fouling by food wastewater filtration is attributed to depositation of species from effluents onto the membrane surface or within membrane porous, it causes a permeate flux decline with time because the filtration resistance is significant increased (Foley, 2006). Fouling studies on membranes are based in proteins depositation and their interaction in membranes surface. Polydispersity of naturally occurring macromolecules such as polysaccharides and humic substances, have also added a particular complexity on investigation to the fouling membrane mechanisms. Advances in understanding fouling of

The critical flux value depends largely on the hydrodynamic conditions in the process, the membrane pore size, and the feed physicochemical condition (Mänttäri & Nyström, 2000). Appropriate manipulation of these parameters, specifically the hydrodynamic condition, may lead to increment of flux and the reduction or even the elimination of both reversible and irreversible fouling of the membrane. The critical flux can be experimentally identified through constant flux filtration experiments by incrementing the flux until the TMP is no

01234567

Subcritical operation Critical operation Decline flux

I II III

300 kDa

259

150 kDa

15 kDa

TM**P** (bar)

al., 2002).

longer steady.

(II) Critical operation (III) Decline flux.

**Fux (Lh-1m-2)** 

Pure water flux

the effect of pressure increment on selectivity of these membranes for these ions. The results indicate that for all PTM values, the ions Fe3+ and Ca2+ were slightly declined, while ions Na+ and K+, were filtered by both membranes.

For other hand, exceptional selectivity for a number of important separations in wastewater treatment of food industry are mentioned in several reports (Vourch et al., 2008; Muro et al., 2010, Escobar et al., 2011; Simate et al., 2011).

Figure 4a and 4b show the difference between selectivity of two ceramic membranes of MF (300 and 150 kDa) and one of UF (50 and 15 kDa) for various TMP values. The data were obtained by experimental study of organics species in micelles with two colorants (a) Brilliant blue. (b) Tartrazine. Membranes denote a low selectivity for the colorants and a high permeability for water. Particularly, membrane of 15 kDa shows the lowest selectivity for two colorants for all TMP values. SEM image denotes, particles deposited on membrane surface, showing a low selectivity of a membrane of 300 kDa for tartrazine colorant.
