Article

pubs.acs.org/IECR

UHT Condensate Recovery by Reverse Osmosis: A Pilot-Plant Study

Adrián Suárez, Tatiana Fidalgo, Miguel A. Berdasco, and Francisco A. Riera*
Department of Chemical and Environmental Engineering, University of Oviedo, Julián Clavería 8, 33006 Oviedo, Spain

ABSTRACT: Real ultrahigh-temperature (UHT) flash cooler (FC) condensates obtained in the dairy industry were processed
in situ using a reverse-osmosis (RO) membrane (Duratherm HWS 4040 HR; 8.2 m2 filtration area). Semicontinuous trials were
performed to estimate the maximum volume concentration rate (VCR) that fulfils the permeate specifications for reuse as boiler
water. Then, long-term continuous experiments (100 h in duration) were carried out at a constant permeate flow rate (40 L h−1
m−2), and the conductivity, chemical oxygen demand (COD), pH, and Ca2+ concentration were monitored. An activated carbon
column (ACC) was used to reduce the organic charge of the permeate.

1. INTRODUCTION pump, where the pressure is controlled to restore the product’s

The food industry is one of the most extensive water-using original dry matter.29 Flash cooling also permits the removal of
industries,1 and the dairy industry can be considered one of the some volatile compounds.30 As the FC operates at the liquid
most polluting because of its high water consumption (up to 10 boiling point, when the pressure drops, some milk droplets
L of water per liter of processed milk) and its status as the could be entrained in the vapor phase, contributing to an
largest producer of food processing wastewater.2,3 These dairy increase in the pollution load of the condensates.
wastewaters are characterized by wide fluctuations in pollutant In the dairy company considered in this work as a case study,
charge and flow rate.4,5 In most cases, wastewaters generated in the vapor from the FCs is used to preheat the dairy product to
different individual operations are mixed and neutralized prior be processed and subsequently condensed and collected after
to discharge into the municipal sewage system; however, it the vacuum pump. This rotary volumetric pump uses tap water
seems that segregating streams would enable optimal treatment as an auxiliary liquid that is mixed with the condensates to
for each type of wastewater, as well as energy and cost generate the vacuum. As a consequence, the feed wastewater
savings.6,7 used in the present work is a mixture of condensates from the
Because of this extensive water consumption; stricter FCs and tap water used in the vacuum pump. This effluent can
pollution control legislation regarding waste disposal; growing be considered to have a low pollutant load.21,31
concerns for sustainable development; and the need for a safe, The amount of tap water used in the vacuum pump
reliable, and affordable supply of quality water, interest in determines, to some extent, the characteristics of the stream
developing a zero-discharge process has gained relevance, and that feeds the RO process. In general, a higher degree of
the practice of reconditioning, recycling, and reusing water is vacuum requires greater consumption of tap water, which
becoming an objective of study and research.8−12 In this causes two effects: an increase in the pollution load (at a higher
scenario, membrane processes are useful technologies in degree of vacuum, more milk drops pass into the vapor) and a
wastewater reclamation and reuse and are currently well dilution effect.
integrated in the dairy industry, as the decreasing costs of Among all possibilities for water reuse, including drinking-
installation and operation in continuous processing favor their water standards, the most restrictive specifications are given by
use.13−16 More specifically, reverse osmosis (RO) has been boiler-water usage [pH 7−10, conductivity at 25 °C < 40.0 μS/
found to be a promising separation technique for the treatment cm, chemical oxygen demand (COD) < 10 mg of O2/L, total
of dairy wastewaters.2,11,17−24 organic carbon (TOC) < 4 mg of O2/L, Ca2+ concentration <
Ultrahigh-temperature (UHT) treatment of milk and milk- 0.40 mg/L].32,33 These parameters must be guaranteed to
based products involves a heating process that ensures a sterile improve the efficiency and service lifetime of the boiler; control
product with a shelf life at room temperature of at least 6 the blow-down frequency; and avoid potential scaling,
months, while minimizing cooked flavors.25,26 A typical UHT corrosion, and formation of foams.
treatment consists of treatment at 140 °C for 4 s,26 which could The aim of this study was to evaluate the effectiveness of an
be achieved by direct or indirect heating. The benefit of the RO continuous pilot process at a moderate temperature to
direct process lies in higher rates of heat transfer and lower obtain high-quality water for boiler reuse from real dairy
changes in taste compared to the indirect process.27 In direct condensates of a UHT vacuum system. Preliminary tests were
steam injection (DSI) contact, the product is preheated and performed in semicontinuous mode to determine the optimum
pressurized to approximately 4 bar, in order to prevent boiling volume concentration rate (VCR) for long-term processing. An
of the liquid. After passing through the holding tube, the
product is flash-cooled, homogenized, cooled again, and Received: July 1, 2014
aseptically packaged.28 The flash cooling takes place in an Revised: September 1, 2014
expansion chamber, known as the flash cooler (FC), which Accepted: September 9, 2014
operates under a partial vacuum, obtained using a vacuum Published: September 9, 2014

© 2014 American Chemical Society 15237 dx.doi.org/10.1021/ie5026166 | Ind. Eng. Chem. Res. 2014, 53, 15237−15244
Industrial & Engineering Chemistry Research Article

Figure 1. Scheme of the RO equipment.

activated carbon column (ACC) was coupled after the RO step sulfate as the catalyst and Cr3+ ions were then determined
to correct the limit on organic matter in the final permeate. photometrically at 605 nm) and residual hardness (in which
calcium and magnesium ions were reacted with phthalein
2. MATERIALS AND METHODS purple in alkaline solution to form a violet colorant that was
determined photometrically at 565 nm) were carried out by
2.1. Flash Cooler Performance. In the Spanish factory means of Spectroquant cell test kits (Merck, Darmstadt,
where the trials were carried out, four FCs operate Germany) using a NOVA 60 photometer (Merck, Darmstadt,
simultaneously but at different stages of the production cycle. Germany).
When working at the same time, they generate ∼20 m3 of 2.3. Reverse-Osmosis Pilot Plant. A scheme of the
condensates per hour with different pollution loads. membrane pilot plant is shown in Figure 1. Although the plant
The UHT system works in a cyclic manner, although some had the capacity for two in-parallel 4040 spiral-wound
steps can vary depending on production needs or breakdowns. membranes, only one was used, and the other pressure vessel
Aseptic intermediate cleanings (AICs), consisting of the use of was kept closed. Tank TK-01 was used exclusively for cleaning
an alkaline-based detergent with two intermediate rinsing steps, purposes, and an additional external 1000-L capacity tank (TK-
are performed routinely. In the same way, depending on the 02) was available to collect the condensates from the FC
variability of the production schedule, a final cleaning (alkaline vacuum system. The RO rig was fed using two consecutive
and acid) is carried out. On the other hand, when the aseptic pumps [P-02 and P-03, 7.5 and 2.2 kW, respectively (Grundfos,
bottle-filling step cannot handle the production rate, the Bjerringbro, Denmark)]. A 125-μm prefilter (F-01) was
product is sent to an aseptic tank; when this tank becomes full, installed to remove coarse particles that could damage these
the system starts to work with sterile water until the aseptic pumps and the membrane. The pressure and feed flow rate
packaging system can again allow product to enter. were fixed by regulating a needle valve (FCV-03) placed on the
Condensates are continuously produced during the cycle retentate side. Transmembrane pressure (TMP) was controlled
(even when the equipment is loaded with sterile water), except by means of manometers PI-03 and PI-05, whereas the feed,
when final cleaning is performed. As a result, their character- permeate, and concentrate fluxes were measured by flow meters
istics are highly dependent on the instant in the cycle; for FI-01, FI-03, and FI-04, respectively. Finally, to maintain low
example, condensates collected during the intermediate water silt density index (SDI) values (<3)34 for the rinsing and
cleaning show high pH, whereas those accumulated during cleaning steps, a prefiltration system consisting of a bag filter
the start of or after the AICs are more concentrated and have a and 5-, 1-, and 0.2-μm cartridge prefilters was used.
higher organic load, and those occurring when sterile water is 2.4. Membrane. A spiral-wound thin-film (proprietary
circulating have low pollution levels. Finally, the type of composite) RO Duratherm HWS 4040 HR membrane (GE
product sterilized (milk, shakes, etc.) and the sterilization Water & Process Technologies, Trevose, PA) was selected on
parameters also have some influence on the condensate the basis of the recommendation of the membrane
characteristics. manufacturer to treat the type of effluent studied. The specific
2.2. Analytical Methods. In this work, pH (accuracy membrane properties are listed in Table 1. The maximum TMP
±0.05), electrical conductivity (accuracy ±0.5%), and COD and permeate flow rate (J) were set to 30 bar and 40 L h−1 m−2,
(accuracy ±3 mg of O2/L) were measured continuously, respectively, according to the manufacturer.
whereas residual hardness, expressed as Ca2+ concentration The cleaning protocol followed was suggested by a detergent
(accuracy ±0.10 mg/L), was analyzed occasionally. pH and manufacturer (Diversey, Utrecht, The Netherlands) according
electrical conductivity were measured by means of an HQ40d to the characteristics of the wastewater processed. Standard
multimeter (Hach Lange, Mechelen, Belgium). Spectrophoto- membrane cleanings consisted of acid (Divos 2, pH 1.9−2.1, 20
metric measurements of COD (in which samples were oxidized min, 45 °C) and alkaline (Divos 123, pH 11.0−11.5, 45 min, 45
with a hot sulfuric solution of potassium dichromate using silver °C) steps with intermediate water rinsings. However, only
15238 dx.doi.org/10.1021/ie5026166 | Ind. Eng. Chem. Res. 2014, 53, 15237−15244
Industrial & Engineering Chemistry Research Article

Table 1. Main Membrane Characteristics maintained at 1000 L/h [with the cross-flow velocity (CFV)
estimated to be 0.35 m/s], the permeate flow rate was fixed at
RO Duratherm HWS 4040 HR
40 L h−1 m−2, and the system was run at increasing TMP to
average salt rejectiona (%) 99.5 compensate for fouling. Feed and permeate characteristics such
active area (m2) 8.2 as temperature, TMP, feed flow rate, and permeate flow rate (J)
spacer thickness (mil) 31 were continuously recorded, as was the membrane rejection (or
maximum operating pressure (bar) reduction) (R) in terms of conductivity and COD, which can
5−50 °C 41.4 be calculated by means of the equation
51−70 °C 20.7
maximum operating temperature (°C) 70 ⎛ Cp ⎞
R (%) = ⎜1 − ⎟ × 100
maximum cleaning temperature (°C) 50 ⎝ Cf ⎠ (3)
pH operating range 4.0−11.0
pH cleaning range 2.0−11.5 where Cp and Cf are the permeate and feed (concentrated)
a
Testing conditions: 2000 mg/L NaCl solution at 15.5 bar operating values, respectively, of both parameters considered.
pressure, 25 °C, and pH 7.5.
3. RESULTS AND DISCUSSION
alkaline cleaning was carried out when intermediate cleanings 3.1. Flash Cooler Performance and Condensate
were required. To avoid microbiological growth, the membrane Characterization. A complete characterization of the
was stored between runs in acid solution (Divos 2 at pH 3.0− condensates used in this work was previously published by
4.0). The flux recovery was evaluated to determine the cleaning Riera et al.4 The values correspond to samples taken during
efficiency by measuring J at an inlet pressure of 10 bar, a feed several complete operating cycles of the FCs (eight cycles per
flow rate of 1 m3/h, and a temperature of approximately 15 °C FC), and their variability reflects the different steps explained
with prefiltered tap water (SDI < 3). before. The compositions of fresh condensates (and con-
2.5. Operating Procedure. Trials were carried out in two densates at a VCR of 6) are presented in Table 2. As can be
different modes of operation. In the first mode, after the 1000-L
feed tank had been filled, semicontinuous operation of the plant Table 2. Compositions of Initial and Concentrated Feed
was started, with the permeate being drained and the Condensates for RO Processing
concentrate returned to the feed tank until the desired VCR
initial value concentration
was reached. Once this degree of concentration was achieved, parameter range average value rangea average
the permeate stream was sent back to the feed tank, and the
T (°C) 40.5−59.9 50.2 − −
system began to run in a discontinuous closed loop for several
pH 5.31−6.49 5.83 6.34−8.24 7.14
hours. The VCR was estimated as
conductivity 76.1−245.0 184.1 447.0−1017.0 728.2
Vf Vf (μS/cm)
VCR = = COD 4−240 70 30−360 153
Vr Vf − Vp (1) (mg of O2/L)
TOC − − 32.1−85.4 56.8
where Vf, Vr, and Vp are the feed, retentate, and permeate (mg of O2/L)
volumes, respectively. The concentration tests in semi- Ca2+ concentration − − 1.22−2.38 1.67
continuous mode were performed at VCRs of 2, 6, and 10. (mg/L)
Regarding the VCR, the recovery rate (RR), defined as a
At VCR 6 for continuous processing.
VCR − 1
RR (%) = × 100
VCR (2)
seen, the condensates, before and after any previous
is an essential parameter for determining the size and economy concentration, did not fulfill the requirements for boiler water
of a large-scale membrane installation. (pH 7−10, conductivity at 25 °C < 40.0 μS/cm, COD < 10 mg
In the second part of the work, the rig was operated in of O2/L, TOC < 4 mg of O2/L, Ca2+ concentration < 0.40 mg/
continuous mode. The degree of concentration in the feed tank L),32,33 so its treatment by RO was justified.
was maintained by regulating the entrance of fresh feed and The characteristics of the fresh condensates are in agreement
controlling the amount of concentrate returned to the rig. The with those reported by Chmiel et al.,35 slightly higher than
surplus concentrate stream was drained out to the factory’s those published by Mavrov and Bélières,21 and well above those
sewage system. On the other hand, the permeate stream was described by Verheyen et al.36,37 for steam condensates in the
continuously drained out. To verify the desired VCR, the dairy industry. Detailed analyses of condensates from UHT
conductivity inside the feed tank was continuously measured. FCs are scarce, although odors due to thermal treatment and
When working in continuous mode, a model FCA-355 produced by the presence of volatile compounds from milk fat
activated carbon column (ACC; Sefiltra S.A., Madrid, Spain) and some protein degradation have previously been reported.
with 110-L capacity was coupled to the permeate outlet to However, in-depth analyses of these compounds are available.
reduce the organic matter content (RO−ACC). This ACC, Some major organic compounds found in previous research
made of polyester reinforced with fiberglass and having a suggested that acetone, ethanol, acetoine, dimethylsulfide,
capacity of 1800 L/h, was filled with 50 kg of GAC MG 1050 diacetyl, and so on are mainly responsible for organic matter
activated carbon (density ≈ 0.45 g/cm3) (Sefiltra S.A., Madrid, and odors.23,38−40
Spain). 3.2. Reverse-Osmosis Treatment. 3.2.1. Semicontinuous
In both semicontinuous and continuous trials, a cooler was Trials. The RO membrane was characterized using prefiltered
used to regulate the inlet temperature, which was fixed at tap water with an SDI of <3.34 The membrane permeability at a
approximately 45 °C in all tests. The feed flow rate was feed flow rate of 1000 L/h, temperature of 40 °C, and TMP of
15239 dx.doi.org/10.1021/ie5026166 | Ind. Eng. Chem. Res. 2014, 53, 15237−15244
Industrial & Engineering Chemistry Research Article

4−19 bar was calculated to be 5.7 L h−1 m−2 bar−1. After the The evolutions of the TMP and COD were also studied as
temperature correction, the calculated value, 0.143 L h−1 m−2 functions of time at the three VCRs tested (2, 6 and 10) (data
bar−1 °C−1, was comparable to those published by other not shown). The greater the VCR, the higher the pressure
authors who used RO for dairy wastewater processing.2,17,19,20 required to reach the same permeate flow rate, as the increase
The water recovery capacity, that is, the amount of permeate in the concentration of solutes had a negative impact on the
recovered as a percentage of the volume of condensates fed, performance of the process;41 however, the difference was not
determines the size of a large-scale membrane installation and is significant (TMP increased from 8.1 to 10.1 bar as VCR was
dependent on the VCR. RO plants are designed in stages, and increased from 2 and 10). During these experiments, the TMP
membranes face the highest concentration in the final stages. As remained constant, and fouling effects were not observed.
the feed concentration increases, the permeate flow rate is Irrespective of the VCR, the permeate exhibited COD values
reduced, the quality of the permeate is poorer, and the between 0 and 14 mg of O2/L, which exceeded the limit for
membrane cleaning frequency increases.41 For this reason, the boiler-water reuse (10 mg of O2/L) in some cases. These
maximum VCR (or RR) at which the permeate satisfies the results suggest the need to install an ACC to treat the RO
requirements must be estimated. permeate.
Semicontinuous trials were performed to determine the Considering that the permeate conductivities at a VCR of 10
behavior of the RO membrane at different VCRs, for exceeded the requirements for boiler water, continuous trials
experiments that were 12 h in duration. The trials were carried were designed at a VCR of 6, which means an RR value of
out at a constant feed flow rate of 1000 L/h and a fixed 83.3% of the water fed.
permeate flow rate of 40 L h−1 m−2, which was the value 3.2.2. Continuous Trials. Continuous 100-h long-term trials
recommended by the membrane manufacturer. The TMP was were performed at a constant VCR of 6 to simulate the last
increased during the experiments to compensate for the stages of an industrial RO installation. Such tests provide
permeate flux decay due to polarization concentration and information about the membrane behavior in terms of rejection
fouling. The temperature was maintained at an average of 45.2 and fouling frequency. When the industrial UHT sterilization
± 1.7 °C by using a cooler, as the condensates come from FCs system was undergoing cleaning and no other FC was
at temperatures between 47.8 to 55.1 °C. producing condensates, the RO pilot plant worked with
The permeate and feed conductivities and their R values are condensates that had been previously accumulated in the
plotted for VCRs of 2, 6, and 10 in Figure 2 (the conductivity 1000-L feed tank, which remained at a VCR of 6; during this
period, both the permeate and the concentrate were sent back
to the tank. The ACC coupled to the permeate line was not
used when the permeate and concentrate were recirculated.
The experiment was performed at a feed flow rate of 1000 L/h,
a constant permeate flow rate of 40 L h−1 m−2, and an
increasing pressure to compensate for the flux decay. The feed
temperature was maintained at an average of 44.9 ± 3.3 °C.
Figure 3 shows some of the results of the long-term
continuous experiment, where the temperature, TMP, con-

Figure 2. Semicontinuous RO. Feed (F) and permeate (P)

conductivities (μS/cm) and conductivity reduction (R, %) at different
operating VCRs. Boiler-water limit = 40.0 μS/cm.

requirement for boiler water is represented by the horizontal

dotted line at 40 μS/cm). The experiments were conducted on
different days, with different condensate characteristics, which
explains the discontinuity in conductivity observed at 6 h. The
R value for conductivity remained almost constant after 12 Figure 3. Long-term continuous RO−ACC test. Conditions: Constant
working hours, and the average values were estimated to be permeate flow rate (J = 40 L h−1 m−2), feed flow rate = 1000 L/h
89.5%, 95.0%, and 95.0% for VCRs of 2, 6, and 10, respectively; (CFV = 0.35 m/s). TMP (bar) and temperature (T, °C) vs time at a
VCR of 6 and as functions of feed conductivity (μS/cm) and COD
this means that the conductivity R value increased slightly with (mg of O2/L).
increasing VCR. At higher VCR, the difference in solute
concentration across the two sides of the membrane increased,
and as a result, the reduction increased as well (see eq 3).42 The ductivity, and COD of the condensates fed are plotted as
permeate conductivity varied from 20.3 to 53.5 μS/cm. functions of time. The temperature variability during the first
The conductivity requirements were satisfied (permeate part of the experiment (first 10 h) was due to the mode of
values were lower than 30 μS/cm in most cases) at VCRs of 2 operation. During the periods in which the system was working
and 6. However, the limit was exceeded at a VCR of 10, in recirculation mode (see 6−10 h, 23 h, 49−51 h, etc.), the
reaching permeate values up to 53.5 μS/cm; for this reason, the temperature control was difficult, and a drop in temperature
trial was stopped after 8 h. was recorded. During these periods, the membrane perme-
15240 dx.doi.org/10.1021/ie5026166 | Ind. Eng. Chem. Res. 2014, 53, 15237−15244
Industrial & Engineering Chemistry Research Article

ability was lower because of an increase in fluid viscosity and a

decrease of diffusivity; then, an increase in TMP was necessary
to maintain a permeate flow rate of 40 L h−1 m−2. Except in the
situations described above, the average temperature was 45 °C,
with the loss of temperature inside the equipment estimated to
be 2.5 °C by comparing the inlet and RO−ACC outlet
temperatures.
The evolution of TMP in Figure 3 provides information
about membrane fouling. Each time the rig reached the
maximum TMP (30 bar), an alkaline cleaning step was
performed. Further, complete (alkaline- and acid-based)
cleaning was carried out each weekend to avoid microbiological
growth during the weekend shutdowns (the UHT system was
stopped on weekends as well). During the entire continuous
test, seven complete cleanings and three alkaline cleanings were
performed. The original membrane permeate flow rate was
recovered after both cleaning protocols.
The COD values in Figure 3 (black columns) demonstrate
the heterogeneity of the inlet condensates, with values from 30
to 360 mg of O2/L, corresponding to the original values (at a
VCR of 1) of 4 and 117 mg of O2/L, respectively. After 12, 18,
51, 62, and 81 h of operation, the maximum TMP was reached.
However, the pressure peaks at 12, 18, 51, and 62 h were due to
a momentary increase in concentrated feed COD (black bars).
COD values higher than approximately 300 mg of O2/L caused
a very rapid increase in TMP. On the contrary, when COD
remained low (as in the period from 67 to 81 h), the increase in Figure 4. Long-term continuous RO−ACC processing. Conditions:
TMP was slower, and the membrane worked for 14 h without Constant permeate flow rate (J = 40 L h−1 m−2), feed flow rate = 1000
L/h (CFV = 0.35 m/s), temperature = 44.9 ± 3.3 °C. (a)
any cleaning or rinsing step. Conductivity (μS/cm) and reduction (R, %) and (b) COD (mg of
Values of concentrated feed conductivity during the test were O2/L) and reduction (R, %) at a VCR of 6 of permeate and after
approximately 728 ± 113 μS/cm. Conductivity did not show activated carbon (ACC) treatment.
any influence on fouling as, in general, there was neither
significant heterogeneity nor momentary peaks of high
conductivity. between 185 and 360 mg of O2/L, which are far from the
In Figure 4, the conductivity and COD values of the average COD for concentrated feed, estimated to be 85 mg of
permeate before and after ACC treatment, as well as the O2/L. As a result, these fleeting high feed CODs or the
reductions of these two parameters, are plotted as functions of combination of various consecutive high values were strongly
time. As expected, the differences in conductivity before and associated with the fouling observed and, thus, with the rapid
after ACC were negligible (Figure 4a), and only some permeate increase in TMP. Regarding the nature of the fouling observed,
samples showed values higher than the limit. Conductivity it has been demonstrated that inorganic fouling is negligible at
RO−ACC reductions were between 90.1% and 98.3% (average working pH values (neutral to alkaline), as soluble calcium,
= 95.4%) and remained almost constant throughout the tests. which appears to be the main promoter of inorganic fouling, is
It has been well demonstrated that fouling affects membrane complexed with phosphate and citrate at these pH values.44 On
selectivity and that a rise in pressure both increases the the other hand, organic fouling, mainly caused by proteins, is
hydraulic resistance of the membrane and decreases the the major factor responsible for membrane fouling in dairy
membrane selectivity.43 This behavior was observed from 76 applications.3,22,45 However, the TMP peak at 81 h was not due
to 81 h, when the permeate quality in terms of conductivity to an increase in the organic load of the feed. In that case, the
worsened as the TMP increased. As a result, the RO−ACC membrane was run for 14 h without cleaning.
outlet conductivity changed from 18.6 to 47.1 μS/cm, and the There were several significant drops in the COD reduction of
reduction decreased from 97.5% to 94.4%. This behavior was the RO permeate that matched the fouling periods correspond-
also observed in the last part of the tests, from 90 h onward, ing to peaks at 51 and 81 h. As reported by Brans et al.,46 the
although it is difficult to determine whether this stepwise fouling and increase in associated pressure reduced the
increase was related to the previous statement or was selectivity of the membrane, leading to a lower permeate
influenced by the observed rise in feed conductivity, given quality in terms of organic matter. Furthermore, in some cases,
that, although the permeate conductivity increased from 22.8 to the decreases in COD reduction corresponded to peaks in feed
39.2 μS/cm, the decrease in reduction was negligible (0.5%). COD, such as that at 41 h, where the R value was reduced by
With respect to the COD (Figure 4b), permeates showed up to 75.6%, which matched with a momentary COD inlet of
significant differences with time due to the variability of the 201 mg of O2/L. Figure 4b demonstrates the inability of RO
feed characteristics. (In Figure 4, if a permeate or an ACC membranes to completely remove the organic matter, as stated
vertical bar does not appear for a given feed, this means that the by Bódalo-Santoyo et al.,47 who showed the limited ability of
value reached was 0 mg of O2/L.) In most cases, high COD RO to reject low-molecular-weight organic compounds. On the
values corresponded to momentary inlets of highly polluted other hand, the effectiveness of the ACC in reducing the
organic matter, where the concentrated feed reached values organic charge of permeates is well demonstrated in the figure.
15241 dx.doi.org/10.1021/ie5026166 | Ind. Eng. Chem. Res. 2014, 53, 15237−15244
Industrial & Engineering Chemistry Research Article

The maximum COD reduction achieved after the RO−ACC

process was 100.0%, although the average was established as
97.9%.
During the continuous trial, 25.9 m3 of permeate was fed into
the ACC. No information was found with respect to the
maximum organic load for the commercial column used, as it is
highly dependent on the contaminant type, the characteristics
of the effluent, the feed concentration, and the type of activated
carbon.48 The average COD removal of the ACC was
calculated to be 76.8% by comparing the concentrations of
the inlet (RO permeate) and outlet streams. The total organic
load supported was 0.0057 kg of COD per kilogram of
activated carbon. At this rate, the breakthrough for this
particular case, defined as the point at which the COD in the Figure 5. Long-term continuous RO−ACC processing. pH evolution
RO−ACC outlet stream continuously exceeded the limit with time of feed and permeate and after activated carbon (ACC)
treatment. Conditions: Constant permeate flow rate (J = 40 L h−1
required and, therefore, the ACC exhaustion point, was not m−2), feed flow rate = 1000 L/h (CFV = 0.35 m/s).
reached. The ACC required some maintenance during the
trials: retro-washing with tap water at high flow rate was
performed weekly. In the case of reaching irreversible
saturation, the activated carbon should be regenerated in an inorganic compounds adsorbed on the membrane surface can
oven or completely removed and replaced. influence its charge.56 In these experiments, there was no
The average conductivity and COD of the RO−ACC outlet evidence of this effect, as the pH of the permeate remained
stream were 33.3 μS/cm and 3 mg of O2/L, respectively. Some practically constant (5.3 ± 0.4). No significant differences were
samples reached maximum values of 52.6 μS/cm and 14 mg of
found between pH samples before and after the ACC.
O2/L, respectively. Although these values were above the
When permeate pH values were outside the range required
restriction, the permeate obtained in an industrial plant will be
for boiler-water reuse (7−10), then some treatments, such as
diluted before entering the boiler. Permeates obtained in RO
degassing steps to reduce the content of CO2 or the addition of
steps with high VCRs and those from low VCRs (with better
alkalizing or antiscalant/antifouling chemicals, should be taken
permeate quality) will be mixed in the final current49,50 and the
RO−ACC outlet stream will be mixed with fresh demineralised into consideration as a final step before the reuse of permeates
water to satisfy the current necessity of the factory. This as boiler waters.
behavior could also be applied to Ca2+, which showed values of
up to 0.67 mg/L, even though the average, 0.36 mg/L, was 4. CONCLUSIONS
located below the limit (0.40 mg/L). The main difficulties when treating real UHT condensates in
The heterogeneity in organic matter found when processing the dairy industry are related to the heterogeneity of the feed in
wastewaters from the vacuum system of the UHT sterilization terms of conductivity and, especially, COD. The conductivities
process by RO advised that possible momentary high-organic- of concentrated condensates treated when working in
load streams should be controlled and segregated to reduce the continuous mode (at a VCR of 6) varied between 447 and
cleaning frequency and avoid unexpected rapid periods of 1017 μS/cm (average = 728 μS/cm), and the reduction
fouling. If this segregation is not taken into account, the obtained was between 90.1% and 98.3% (average = 95.4%). In
membrane behavior is not easily predictable and intermediate
general, the permeate conductivities were lower than the fixed
cleanings would take place when necessary, reducing the
limit (40 μS/cm). Momentary values higher than the boiler-
economic interest of the application.
water requirement were obtained when the membrane was
Finally, the pH performance during the continuous test as a
function of time for the concentrated feed, the permeate, and fouled (maximum TMP) as a result of an increase of organic
the ACC output stream are shown in Figure 5. The feed pH matter in the feed. COD was considered to be the most
values (at a VCR of 1) varied between 5.3 and 6.9, and those important parameter that affected the permeate quality because,
for the concentrated feed (at a VCR of 6) showed values when the organic load was significant (about 300 mg of O2/L
between 6.3 and 8.2. In the figure, some pH drops (clearly in this work), rapid membrane fouling took place, and an
observed after 18, 40, 51, 78, and 93 h) can be noted due to the additional cleaning had to be done. The feed COD at a VCR of
variability in the inlet waters (as a function of the degree of 6 varied between 30 and 360 mg of O2/L, and the RO
vacuum, the timing in the FC cycle, the product treated, and membrane was not able to reduce the permeate content to
the sterilization parameters of the UHT system) for different satisfy the limit established when the feed presented high COD
working days. Permeate acidification with respect to the feed values. The ACC was demonstrated to be effective in reducing
was previously observed by Suárez and Riera,42 and it was the organic load, and the RO−ACC outlet stream fulfilled the
attributed to the membrane and feed characteristics. The boiler-water specifications (10 mg of O2/L) in most of the
permeate was acidified, generally always to the same extent (it samples analyzed. When permeate pH was acidified (outside
was reduced in all cases by 1.9 ± 0.2 pH units), as a result of the limit), then a post-treatment was advisible to adjust it.
electrostatic repulsion/attraction effects between the solutes A general recommendation for treating these types of
and the membrane surface, which is negatively charged at wastewaters with the objective of producing boiler water is to
working pH values, as are most RO membranes,51 promoting segregate condensates when COD reached values higher than
the favorable passage of H+ ions.52 The acidification observed is 300 mg of O2/L. In the rest of the cases, RO followed by ACC
common in RO wastewater processing.53−55 Organic or was shown to be effective.
15242 dx.doi.org/10.1021/ie5026166 | Ind. Eng. Chem. Res. 2014, 53, 15237−15244
Industrial & Engineering Chemistry Research

■
Article

AUTHOR INFORMATION purposes of reuse: Hybrid process concept and on-site test operation
of a pilot plant. Desalination 2002, 143 (2), 183−196.
Corresponding Author (7) Casani, S.; Rouhany, M.; Knøchel, S. A discussion paper on
*Tel.: +34 985103436. Fax: +34 985103434. E-mail: far@ challenges and limitations to water reuse and hygiene in the food
uniovi.es. industry. Water Res. 2005, 39 (6), 1134−1146.
Notes (8) Sánchez, I. M. R.; Ruiz, J. M. M.; López, J. L. C.; Pérez, J. A. S.
Effect of environmental regulation on the profitability of sustainable
The authors declare no competing financial interest.

■
water use in the agro-food industry. Desalination 2011, 279 (1−3),
252−257.
ACKNOWLEDGMENTS (9) Morin-Couallier, E.; Fargues, C.; Lewandowski, R.; Decloux, M.;
This research was supported by the dairy company Lameloise, M.-L. Reducing water consumption in beet distilleries by
Corporación Alimentaria Peñasanta S.A., which provided data recycling the condensates to the fermentation phase. J. Cleaner Prod.
and valuable information. The authors also acknowledge the 2008, 16 (5), 655−663.
(10) Lee, W.; Okos, M. R. Sustainable food processing systems
continuous advice given by the membrane supplier.

■
Path to a zero discharge: Reduction of water, waste and energy. Proc.
Food Sci. 2011, 1, 1768−1777.
NOMENCLATURE (11) Sarkar, B.; Chakrabarti, P. P.; Vijaykumar, A.; Kale, V.
Wastewater treatment in dairy industriesPossibility of reuse.
Abbreviations Desalination 2006, 195 (1−3), 141−152.
ACC = activated carbon column (12) Palumbo, S. A.; Rajkowski, K. T.; Miller, A. J. Current
AIC = aseptic intermediate cleaning approaches for reconditioning process water and its use in food
CFV = cross-flow velocity manufacturing operations. Trends Food Sci. Technol. 1997, 8 (3), 69−
COD = chemical oxygen demand 74.
DSI = direct steam injection (13) Juang, L.-C.; Tseng, D.-H.; Lin, H.-Y. Membrane processes for
F = filter water reuse from the effluent of industrial park wastewater treatment
FC = flash cooler plant: A study on flux and fouling of membrane. Desalination 2007,
FCV = flow control valve 202 (1−3), 302−309.
(14) Pouliot, Y. Membrane processes in dairy technologyFrom a
FI = flow indicator
simple idea to worldwide panacea. Int. Dairy J. 2008, 18 (7), 735−740.
P = pump (15) Coskun, T.; Debik, E.; Demir, N. M. Treatment of olive mill
PI = pressure indicator wastewaters by nanofiltration and reverse osmosis membranes.
RO = reverse osmosis Desalination 2010, 259 (1−3), 65−70.
RR = recovery rate (16) Koyuncu, I.; Turan, M.; Topacik, D.; Ates, A. Application of low
SDI = silt density index pressure nanofiltration membranes for the recovery and reuse of dairy
TK = tank industry effluents. Water Sci. Technol. 2000, 41, 213−221.
TMP = transmembrane pressure (bar) (17) Akoum, O.; Jaffrin, M. Y.; Ding, L. H.; Frappart, M. Treatment
TOC = total organic carbon of dairy process waters using a vibrating filtration system and NF and
UHT = ultrahigh temperature RO membranes. J. Membr. Sci. 2004, 235 (1−2), 111−122.
VCR = volume concentration rate (18) Balannec, B.; Gésan-Guiziou, G.; Chaufer, B.; Rabiller-Baudry,
M.; Daufin, G. Treatment of dairy process waters by membrane
Symbols operations for water reuse and milk constituents concentration.
Cf = feed concentration (μS/cm or mg of O2/L) Desalination 2002, 147 (1−3), 89−94.
Cp = permeate concentration (μS/cm or mg of O2/L) (19) Balannec, B.; Vourch, M.; Rabiller-Baudry, M.; Chaufer, B.
J = permeate flow rate (L h−1 m−2) Comparative study of different nanofiltration and reverse osmosis
R = reduction (%) membranes for dairy effluent treatment by dead-end filtration. Sep.
Vf = feed volume (m3 or m3/h) Purif. Technol. 2005, 42 (2), 195−200.
Vp = permeate volume (m3 or m3/h) (20) Frappart, M.; Jaffrin, M.; Ding, L. H. Reverse osmosis of diluted
skim milk: Comparison of results obtained from vibratory and rotating
Vr = retentate volume (m3 or m3/h)

■
disk modules. Sep. Purif. Technol. 2008, 60 (3), 321−329.
(21) Mavrov, V.; Bélières, E. Reduction of water consumption and
REFERENCES wastewater quantities in the food industry by water recycling using
(1) Salehi, F. Current and future applications for nanofiltration membrane processes. Desalination 2000, 131 (1−3), 75−86.
technology in the food processing. Food Bioprod. Process. 2014, 92 (2), (22) Turan, M. Influence of filtration conditions on the performance
161−177. of nanofiltration and reverse osmosis membranes in dairy wastewater
(2) Vourch, M.; Balannec, B.; Chaufer, B.; Dorange, G. Nano- treatment. Desalination 2004, 170 (1), 83−90.
filtration and reverse osmosis of model process waters from the dairy (23) Vourch, M.; Balannec, B.; Chaufer, B.; Dorange, G. Treatment
industry to produce water for reuse. Desalination 2005, 172 (3), 245− of dairy industry wastewater by reverse osmosis for water reuse.
256. Desalination 2008, 219 (1−3), 190−202.
(3) Andrade, L. H.; Mendes, F. D. S.; Espindola, J. C.; Amaral, M. C. (24) Aydiner, C.; Sen, U.; Topcu, S.; Ekinci, D.; Altinay, A. D.;
S. Nanofiltration as tertiary treatment for the reuse of dairy wastewater Koseoglu-Imer, D. Y.; Keskinler, B. Techno-economic viability of
treated by membrane bioreactor. Sep. Purif. Technol. 2014, 126, 21−29. innovative membrane systems in water and mass recovery from dairy
(4) Riera, F. A.; Suárez, A.; Muro, C. Nanofiltration of UHT flash wastewater. J. Membr. Sci. 2014, 458, 66−75.
cooler condensates from a dairy factory: Characterisation and water (25) Tran, H.; Datta, N.; Lewis, M. J.; Deeth, H. C. Predictions of
reuse potential. Desalination 2013, 309, 52−63. some product parameters based on the processing conditions of ultra-
(5) Farizoglu, B.; Uzuner, S. The investigation of dairy industry high-temperature milk plants. Int. Dairy J. 2008, 18 (9), 939−944.
wastewater treatment in a biological high performance membrane (26) Newstead, D. F.; Paterson, G.; Anema, S. G.; Coker, C. J.;
system. Biochem. Eng. J. 2011, 57, 46−54. Wewala, A. R. Plasmin activity in direct-steam-injection UHT-
(6) Noronha, M.; Britz, T.; Mavrov, V.; Janke, H. D.; Chmiel, H. processed reconstituted milk: Effects of preheat treatment. Int. Dairy
Treatment of spent process water from a fruit juice company for J. 2006, 16 (6), 573−579.

15243 dx.doi.org/10.1021/ie5026166 | Ind. Eng. Chem. Res. 2014, 53, 15237−15244

Industrial & Engineering Chemistry Research Article

(27) Innings, F.; Hamberg, L. Steam condensation dynamics in granular activated carbon (MBR−GAC) system. Bioresour. Technol.
annular gap and multi-hole steam injectors. Proc. Food Sci. 2011, 1, 2012, 113, 169−173.
1278−1284. (49) Jirjis, B. F.; Luque, S. Practical Aspects of Membrane System
(28) Gösta, B.; López, A. Manual de Industrias lácteas TETRA PAK; Design in Food and Bioprocessing Applications. In Membrane
Ediciones Mundi Prensa: Madrid, Spain, 1994. Technology: A Practical Guide to Membrane Technology and Applications
(29) Early, R. Technology of Dairy Products; Springer: London, 1998. in Food and Bioprocessing; Cui, Z. F., Muralidhara, H. S., Eds.;
(30) Lewis, M. J. Thermal Processing. In Food Processing Handbook; Butterworth-Heinemann: Oxford, U.K., 2010; Chapter 9, pp 179−212.
Brennan, J. G., Ed.; Wiley-VCH Verlag GmbH & Co. KGaA: (50) Schwinge, J.; Neal, P. R.; Wiley, D. E.; Fletcher, D. F.; Fane, A.
Weinheim, Germany, 2006; Chapter 2; pp 33−70. G. Spiral wound modules and spacers: Review and analysis. J. Membr.
(31) Fähnrich, A.; Mavrov, V.; Chmiel, H. Membrane processes for Sci. 2004, 242 (1−2), 129−153.
water reuse in the food industry. Desalination 1998, 119 (1−3), 213− (51) Gerard, R.; Hachisuka, H.; Hirose, M. New membrane
216. developments expanding the horizon for the application of reverse
(32) Judd, S.; Jefferson, B. Membranes for Industrial Wastewater osmosis technology. Desalination 1998, 119 (1−3), 47−55.
Recovery and Re-Use; Elsevier Science Ltd.: Oxford, U.K., 2003. (52) Qin, J.-J.; Oo, M. H.; Lee, H.; Coniglio, B. Effect of feed pH on
(33) Mavrov, V.; Chmiel, H.; Bélières, E. Spent process water permeate pH and ion rejection under acidic conditions in NF process.
desalination and organic removal by membranes for water reuse in the J. Membr. Sci. 2004, 232 (1−2), 153−159.
food industry. Desalination 2001, 138 (1−3), 65−74. (53) Qin, J.-J.; Oo, M. H.; Wai, M. N.; Wong, F.-S. Effect of feed pH
(34) Mosset, A.; Bonnelye, V.; Petry, M.; Sanz, M. A. The sensitivity on an integrated membrane process for the reclamation of a combined
of SDI analysis: From RO feed water to raw water. Desalination 2008, rinse water from electroless nickel plating. J. Membr. Sci. 2003, 217
222 (1−3), 17−23. (1−2), 261−268.
(35) Chmiel, H.; Mavrov, V.; Belieres, E. Reuse of vapour condensate (54) Prats, D.; Chillon-Arias, M. F.; Rodriguez-Pastor, M. Analysis of
from milk processing using nanofiltration. Filtr. Sep. 2000, 37 (3), 24− the influence of pH and pressure on the elimination of boron in
reverse osmosis. Desalination 2000, 128 (3), 269−273.
27.
(55) Kang, M.; Kawasaki, M.; Tamada, S.; Kamei, T.; Magara, Y.
(36) Verheyen, V.; Cruickshank, A.; Wild, K.; Heaven, M. W.;
Effect of pH on the removal of arsenic and antimony using reverse
McGee, R.; Watkins, M.; Nash, D. Soluble, semivolatile phenol and
osmosis membranes. Desalination 2000, 131 (1−3), 293−298.
nitrogen compounds in milk-processing wastewaters. J. Dairy Sci. (56) Childress, A. E.; Elimelech, M. Effect of solution chemistry on
2009, 92 (7), 3484−3493. the surface charge of polymeric reverse osmosis and nanofiltration
(37) Verheyen, V.; Cruickshank, A.; Wild, K.; Heaven, M. W.; membranes. J. Membr. Sci. 1996, 119 (2), 253−268.
McGee, R.; Watkins, M.; Nash, D. Characterization of organic
particulates present in milk factory process waters used for reuse along
with aerobically digested effluent wastewater. Bioresour. Technol. 2011,
102 (2), 2118−2125.
(38) Moio, L.; Dekimpe, J.; Etievant, P.; Addeo, F. Neutral volatile
compounds in the raw milks from different species. J. Dairy Res. 1993,
60 (2), 199−213.
(39) Pereda, J.; Jaramillo, D. P.; Quevedo, J. M.; Ferragut, V.;
Guamis, B.; Trujillo, A. J. Characterization of volatile compounds in
ultra-high-pressure homogenized milk. Int. Dairy J. 2008, 18 (8), 826−
834.
(40) Valero, E.; Villamiel, M.; Miralles, B.; Sanz, J.; Martínez-Castro,
I. Changes in flavour and volatile components during storage of whole
and skimmed UHT milk. Food Chem. 2001, 72 (1), 51−58.
(41) Sagne, C.; Fargues, C.; Lewandowski, R.; Lameloise, M.-L.;
Gavach, M.; Decloux, M. A pilot scale study of reverse osmosis for the
purification of condensate arising from distillery stillage concentration
plant. Chem. Eng. Process.: Process Intensif. 2010, 49 (4), 331−339.
(42) Suárez, A.; Riera, F. A. Production of high-quality water by
reverse osmosis of milk dairy condensates. J. Ind. Eng. Chem., published
online Jun 10, 2014, http://dx.doi.org/10.1016/j.jiec.2014.06.004.
(43) Gésan-Guiziou, G.; Daufin, G.; Boyaval, E. Critical stability
conditions in skimmed milk crossflow microfiltration: Impact on
operating modes. Lait 2000, 80 (1), 129−138.
(44) Rabiller-Baudry, M.; Bouzid, H.; Chaufer, B.; Paugam, L.;
Delaunay, D.; Mekmene, O.; Ahmad, S.; Gaucheron, F. On the origin
of flux dependence in pH-modified skim milk filtration. Dairy Sci.
Technol. 2009, 89 (3−4), 363−385.
(45) Madaeni, S. S.; Mansourpanah, Y. Chemical cleaning of reverse
osmosis membranes fouled by whey. Desalination 2004, 161 (1), 13−
24.
(46) Brans, G.; Schroën, C. G. P. H.; van der Sman, R. G. M.; Boom,
R. M. Membrane fractionation of milk: State of the art and challenges.
J. Membr. Sci. 2004, 243 (1−2), 263−272.
(47) Bódalo-Santoyo, A.; G6mez-Carrasco, J. L.; Gomez-Gomez, E.;
Maximo-Martin, M. F.; Hidalgo-Montesinos, A. M. Spiral-wound
membrane reverse osmosis and the treatment of industrial effluents.
Desalination 2004, 160 (2), 151−158.
(48) Nguyen, L. N.; Hai, F. I.; Kang, J.; Price, W. E.; Nghiem, L. D.
Removal of trace organic contaminants by a membrane bioreactor−

15244 dx.doi.org/10.1021/ie5026166 | Ind. Eng. Chem. Res. 2014, 53, 15237−15244