Journal of Hazardous Materials 178 (2010) 883–889

Contents lists available at ScienceDirect

Journal of Hazardous Materials

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

Study of different pretreatments for reverse osmosis reclamation of a

petrochemical secondary effluent
C. Benito-Alcázar, M.C. Vincent-Vela ∗ , J.M. Gozálvez-Zafrilla, J. Lora-García
Department of Chemical and Nuclear Engineering, Universidad Politécnica de Valencia, C/Camino de Vera s/n, 46022 Valencia, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Conventionally treated petrochemical wastewaters contain substantial quantities of hazardous pollu-
Received 2 October 2009 tants. In addition, wastewater reuse is being enhanced as a consequence of the shortage of fresh water.
Received in revised form 2 February 2010 Advanced petrochemical wastewater treatment for water reuse will reduce hazardous pollutants dis-
Accepted 6 February 2010
charges as well as water consumption. Reverse osmosis is a suitable technology to obtain pure water.
Available online 12 February 2010
This work studies the adequacy of different pretreatments applied to a petrochemical secondary effluent
to produce a suitable feeding for reverse osmosis treatment. The permeate obtained can be used in the
Keywords:
petrochemical industry for different processes. In this work, several experiments (granulated activated
Reverse osmosis pretreatment
Granulated activated carbon filtration
carbon filtration, ultrafiltration, nanofiltration and granulated activated carbon filtration coupled with
Ultrafiltration nanofiltration) were performed to improve the conventional pretreatment. Total organic carbon, chem-
Nanofiltration ical oxygen demand, turbidity and silt density index were used to evaluate water quality for reverse
osmosis feeding. In granulated activated carbon filtration, all the measured parameters but silt density
index indicated a good filtrate quality to feed reverse osmosis membranes. Although the ultrafiltration
permeate obtained was suitable for reverse osmosis, nanofiltration and granulated activated carbon filtra-
tion coupled with NF provided a better effluent quality for reverse osmosis than the other pretreatments
studied.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction ter treatments since bioremediation of these pollutants is not

Manufacturing of many petrochemicals and other types of
products results in most cases in the generation of substantial
quantities of hazardous materials. The improper management of
hazardous wastes generated by the petrochemical industry is
considered to be one of the most important environmental prob-
lems all over the world [1]. Petrochemical effluents have had
a significant impact on the pollution of surface water systems.
Moreover, petrochemical discharges have been proved to induce
notable ecotoxicological effects [2]. In particular, some petrochem-
ical compounds, such as polyaromatic hydrocarbons (PAHs) and
heterocyclic NOS-containing compounds (NOS-compounds) have
certain physico-chemical properties that combine to make con-
taminated sites sinks for these compounds [3–5] Many of these
compounds have been considered as priority pollutants which
exert biohazardous effects on both human and other living organ-
isms in the environment [6].
Part of the hazardous organic matter present in petrochemical
wastewater is hard of removing with conventional wastewa-
always effective. Moreover, in regions with limited water resources,
such as Mediterranean countries, wastewater reclamation is being
enhanced. In Spain, several MWWTPs1 are being designed and are
improving their installations with tertiary treatments for wastew-
ater reuse [7]. Regarding specifically on industrial wastewater, it
is interesting to reuse treated effluents especially for industries
with a high consumption of water (dying, paper, petrochemical,
. . .). Reutilization of treated effluents is particularly interesting
since it both minimizes fresh water consumption and natural
resources scarcity and reduces the amount of hazardous pollutants
discharged into surface waters.
Depending on the secondary effluent composition and the water
quality required for different uses, such as agricultural (irrigation)
or industrial (cooling, process, cleaning and fire-extinction), the
tertiary treatment applied has to be more or less exhaustive. In
this work, reclaimed water will be used for cooling, process and
fire-extinction in a petrochemical factory. Table 1 shows the water
quality requirements in terms of conductivity for the different uses.

∗ Corresponding author. Tel.: +34 963877633; fax: +34 963877639.

1
E-mail address: mavinve@iqn.upv.es (M.C. Vincent-Vela). Municipal wastewater treatment plant.

0304-3894/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jhazmat.2010.02.020
884 C. Benito-Alcázar et al. / Journal of Hazardous Materials 178 (2010) 883–889
Table 1
Water quality requirements for different uses in the factory.
Use Conductivity (␮S cm−1 )
Cooling 3000
Process water (softened water) 100
Process water (demineralised water) 1 Conductivity (mS/cm)
Fire-extinction 1000
Due to the low water conductivity requirements (1 ␮S/cm), RO2
was selected as the most suitable process regarding the high quality
of the water produced at a suitable economical rate.
RO processes have been successfully employed in industrial
wastewater reclamation [8,9]. Nevertheless, controlling RO mem-
brane fouling continues to be a major challenge in desalination and
wastewater reclamation. Proper pretreatment has been stressed
repeatedly as the first line of defence in controlling membrane
fouling and assuring success for RO operation [10].
In any RO process for seawater, brackish water or wastewa-
ter, the feed quality must be appropriate to minimize membrane
fouling. In this way, the main goal of the pretreatment is
removing SS3 , colloidal mater and organic dissolved substances
which cause fouling on RO membranes surface. The efficiency of
conventional pretreatment (chlorination + coagulant + media filtra-
tion + acidification + cartridge filtration + dechlorination) had been
proved in many cases with natural feedings, either seawater or
brackish water [11]. However, when the feed comes from a sec-
ondary effluent there are some drawbacks like the variability of
sewage composition and the presence of residual pollutants [12].
This fact compels researchers to study the most effective pretreat-
ment for each effluent, mainly with industrial wastewaters whose
variability and contaminant charge are significant.
RO membrane manufacturers recommend a minimum feed
quality in order to minimize membrane fouling. Most of them sug-
gest a turbidity lower than 1 NTU and a SDI4 lower than 5 [13,14],
however an SDI lower than 3 is desired for a successful opera-
tion. The SDI provides an estimation of the fouling tendency of the
feed and it is recognised as the standard test to estimate mem-
brane fouling potential [15]. Another critical parameter on fouling
prevention is TOC5 . This parameter is related to the adsorption
of organic substances on the membrane surface which may cause
flux loss and irreversible fouling. In natural waters, pretreatment
should be considered when TOC exceeds 3 mg/L [14]. However, in
the case of wastewaters TOC acceptable values vary depending on
the substances present and their concentration. It is necessary a
case-by-case evaluation with field experiments. There are several
studies in which the membranes tested had a success operation
during long periods despite the TOC values were significant in the
feed water. Into et al. [16] investigated RO membranes operation
with an industrial wastewater containing low molecular weight
organic substances. The feed for the RO plant had a TOC value of
2500–4000 mg/l.
The selection of specific parameters to measure as criteria for
the selection of the best pretreatment process was performed
according to the above mentioned recommendations of membrane
manufacturers. Therefore, SDI, TOC and turbidity were selected as
the most representative parameters to evaluate the suitability of
the pretreatment for RO feeding.
In previous studies, conventional pretreatment was not effective
enough to achieve the reduction of these parameters to the required
values. Consequently, it had to be improved by implementation of
2 8
Reverse osmosis.
3 9
Suspended solids.
4 10
Silt density index.
5 11
Total organic carbon.
Table 2
Typical composition of the secondary effluent
of the IWWTP located in the petrochemical
factory.
Parameter Value
4.76
pH 6.55
SS (mg/L) 31
Turbidity (NTU) 3.888
COD (mg/L) 150
TOC (mg/L) 24.8
a proper treatment. Wong [17] investigated wastewater reclama-
tion for a petrochemical plant using RO. This study revealed the
effectiveness of a pretreatment consisting of chemical oxidation,
dual-media filtration, GAC6 adsorption, UF7 , and ultraviolet disin-
fection. In this paper, GAC filtration, UF and NF8 were studied in
order to enhance the quality of the secondary effluent up to RO
feed water requirements. In addition, SEM9 analysis and IR10 spec-
tra of SDI filters were carried out to examine the fouling layer and
to identify the nature of foulants.
The selection of the three pretreatments above mentioned for
this work was preformed on the basis of the following arguments.
GAC filtration was chosen because it was found in the literature [17]
that chemical or ultraviolet oxidation processes combined with
GAC filtration were successfully used as a pretreatment for RO in
the case of raw petrochemical waste streams. In our case, the petro-
chemical wastewater stream comes from a secondary effluent of
a wastewater treatment plant, therefore organic matter oxidation
has already been performed in the wastewater treatment plant and
chemical and ultraviolet oxidation may not be needed. Arviv et al.
[18] reported successful operation of RO after using a single pre-
treatment. This pretreatment consisted of an UF pretreatment in
the case of petrochemical wastewaters coming out from a biological
secondary treatment. Therefore, ultrafiltration pretreatment was
chosen as a possible pretreatment in this work. There are other
membrane technologies, such as NF, that can precede RO mem-
brane processes and prevent RO membrane fouling. However, NF
requires more energy consumption than UF. Despite this, when UF
is not able to achieve the required characteristics for the permeate
stream, NF becomes a suitable alternative. This is the reason why
NF was proposed as an alternative in this work in the case that UF
was not able to achieve the required water quality parameters for
RO feeding.
2. Materials and methods
2.1. Secondary effluent samples
In this work, the samples used as feed were taken from the sec-
ondary effluent of the IWWTP11 located in a petrochemical factory.
The typical composition of this effluent is shown in Table 2. This
effluent was pretreated at the pilot plant located in the IWWTP.
The pretreatment consisted of chlorination with sodium hypochlo-
rite, coagulated sand filtration with poly aluminium chloride and
cartridge filtration. All the experiments carried out in this research
used this pretreated secondary effluent except for the UF tests. In
this case, the sample studied came directly from the secondary
effluent without a pretreatment. Only chlorination and security
6
Granular activated carbon.
7
Ultrafiltration.
Nanofiltration.
Scanning electron microscopy.
Infrared.
Industrial Wastewater Treatment Plant.
 C. Benito-Alcázar et al. / Journal of Hazardous Materials 178 (2010) 883–889 885

Table 3
Pretreated secondary effluent characterization.

Parameter Sample 1 Sample 2 Sample 3

Conductivity (mS/cm) 4.29 3.61 6.38

pH 7.70 7.41 7.53
SS (mg/L) 23 38 32
Turbidity (NTU) 2.473 3.703 3.100
COD (mg/L) 137 147 156
TOC (mg/L) 16.4 30.5 18.6

filtration (50 ␮m) were carried out in order to delay biological

degradation of the sample and to avoid damage on the membrane
surface due to the presence of particles.
Three samples of the pretreated secondary effluent were ana-
lyzed to estimate the characterization parameters of wastewater. Fig. 1. Scheme of the UF pilot plant.
Table 3 shows the characterization results. A high presence of dis-
solved solids was observed. This may be related with the high
conductivity of the samples. There was a low quantity of SS despite duration of the each experiment was 2 h. A new membrane was
the sample came from a biological treatment. This fact is a conse- used in each experiment. Therefore, membrane cleaning was not
quence of the filtration carried out at the pretreatment pilot plant, necessary. The permeate turbidity and the SDI15 were measured
since sand filtration and cartridge filtration achieved a significant when the steady-state conditions were achieved.
SS removal. Turbidity values were similar for all the samples and It is worth to mention that fewer experimental conditions than
they revealed the presence colloids. Moreover, COD and TOC mea- in the case of GAC filtration experiments were required to evaluate
surements suggested that the samples had a significant content UF and NF as a pretreatment because UF and NF applied to petro-
in organic contaminants. On the other hand, the different values chemical wastewaters were previously studied by the authors of
obtained for conductivity and TOC confirmed the variability of the the present work [12]. Moreover, the operating conditions of UF
petrochemical factory wastewater. and NF tests were those that provided similar permeate fluxes. This
ensured the possibility of comparison between both pretreatments.
2.2. GAC filtration
2.4. NF and GAC filtration coupled with NF
Three sets of experiments were carried out with three samples
of the pretreated secondary effluent. Metcalf & Eddy [19] recom-
In this group of experiments, two different treatments were
mend a filtration rate for secondary effluents between 4.88 and
studied: NF and GAC filtration coupled with NF (GAC + NF). The flux
19.54 m/h for GAC filtration in their handbook of wastewater treat-
characteristics and the contaminant removal achieved in NF with
ment. They also mention that the typical filtration rate is 9.77 m/h
GAC pre-filtration and without it were compared. The NF mem-
for GAC filtration. Therefore, according to the usual range for GAC
brane tested was the Filmtec® NF270-2540 from Dow. It was a
filtration velocity of secondary effluents [19], five filtration rates
polyamide thin-film composite spiral wound membrane with a
(5, 10, 12.5, 15, 17 m/h) were tested in this work. Lower filtration
diameter of 6.1 cm. The diagram of the pilot plant used in the exper-
rates were not selected because they were not recommended in
iments is shown in Fig. 2. The GAC filtration was carried out at the
the literature. Moreover, the results shown in Figs. 3 and 4 confirm
filtration rate of 10 m/h. The NF tests were performed at a pres-
that there is not a significant influence of the filtration rate on the
sure of 0.4 MPa, a temperature of 25.0 ◦ C, a cross-flow velocity of
quality of the filtered effluent for the effluent studied in this work.
0.035 m/s and a feed flow of 353 L/h. The NF tests lasted 8 h and
In order to ensure the same experimental conditions, a new com-
were performed in the batch recirculation mode to maintain a con-
mercial GAC cartridge was used for each experiment. The filtrate
stant feed concentration. During the experiments, volumetric flux
was continuously returned to the feed tank which contained 10 L
and conductivity were measured. After 8 h of operation, a perme-
of sample. In this recirculation mode of operation, a filtration step
ate sample was analyzed to estimate the SDI, the COD and the TOC
was the filtration of the same volume contained in the system. The
removal.
step time was calculated for each filtration rate. During these tests
several filtrate samples were analyzed for different filtration steps.
For each sample, every filtration rate was studied for three filtra-
tion steps. The parameters studied were turbidity, COD12 , TOC and
SDI15 .

2.3. Ultrafiltration

The samples studied in this set of experiments were pre-filtered

using a 20-␮m filter. In this case, a small pilot plant (Fig. 1) equipped
with a membrane module for two flat sheet membranes was used.
The UF membranes were provided by Nadir® . Their MWCO13 were
30 and 150 kDa. All the experiments were performed at a trans-
membrane pressure of 0.2 MPa, a cross-flow velocity of 0.8 m/s, and
a temperature of 20 ◦ C. The total volume treated was 345 L and the

12
Chemical oxygen demand.
13
Molecular weight cut offs. Fig. 2. Diagram of the NF pilot plant.
886 C. Benito-Alcázar et al. / Journal of Hazardous Materials 178 (2010) 883–889

2.5. Analytical

During this research, several analyses were carried out: con-

ductivity, turbidity, COD, TOC, SS, SDI, SEM analysis and IR spectra.
Conductivity was measured with a CRISON apparatus and turbid-
ity was measured with a turbidimeter HI93761 from HANNA. COD
was determined by a photometric method with Spectroquant Nova
60 from Merck. TOC was determined using a HighTOC II analyser
from Elementar. SS and SDI were calculated according to standard
methods [20,21]. SEM analyses were carried out with a JEOL JSM-
5410 microscopy and an IR spectrum was determined with a FT-IR
Nicolet 710 spectrophotometer.

3. Results and discussion

3.1. GAC filtration

GAC filtration experiments were performed with the three sam-

ples shown in Table 3. All the filtrate samples had a turbidity lower
than 1 NTU, a COD lower than 25 mg/L and a TOC lower than 8 mg/L.
Figs. 3 and 4 illustrate the percentage of COD and turbidity removal
obtained at the filtration rates of 5, 10, 12.5, 15 and 17 m/h in three
GAC filtration steps for samples of the pretreated secondary efflu-
ent. Most of the analyzed samples showed a 70–90% of turbidity
removal and an 85–90% of COD removal.
There is not a clearly proper filtration velocity to achieve the
maximum removal of these parameters because all the filtration
rates tested showed similar results. For this reason, the filtrate
values and the removal achieved for each experiment was calcu-
lated on average. Table 4 shows all the important parameters that
control RO membrane fouling. These parameters assure a good RO
membrane operation if they are kept near the values recommended
by RO membrane manufacturers [13,14]. Therefore, conductivity,
Fig. 4. Turbidity removal in three steps at different filtration rates. (a) Sample 2 and
(b) Sample 3.
pH and suspended solids are not included in the table as they are
not the recommended parameters of RO membrane manufactur-
ers. Mean values of the above mentioned measured parameters
recommended by RO membrane manufacturers are summarized
in Table 4. The three tests that appear in Table 4 correspond to the
three samples used in GAC filtration: test 1 was performed with
Sample 1 (Table 3), test 2 was performed with Sample 2 (Table 3)
and test 3 was performed with Sample 3 (Table 3). The values shown
in the brackets correspond to the removal percentage of the char-
acterization parameters for the filtrate samples related to the feed
samples.
For Samples 1, 2 and 3, average filtrate values lower than 0.7 NTU
and 20 mg/L were obtained for turbidity and COD, respectively. TOC
values were approximately constant around 3 mg/L. These results
indicated that the GAC filtrates were a suitable feeding for RO.
Nevertheless, the analysis of TOC in Sample 2 revealed that GAC
filtration is not effective enough for TOC reduction. For Sample
2, there was a high amount of organic substances present in the
feed (TOC = 30.5 mg/L). Therefore, the analysis of the filtrate for this
sample also showed high values of TOC (4.9 mg/L) despite its high
reduction (84.4%). This fact, as well as the wastewater variability,
represents a drawback of the GAC filtration as a RO pretreatment
for the secondary effluent considered in this work.
It is important to point out that the samples used in these exper-
iments were previously filtered at the pretreatment pilot plant.
The high reduction of the parameters shown in Table 3 can be

Table 4
Filtrate mean values and removal percentage of the measured parameters.

Parameter Test 1 Test 2 Test 3

Turbidity (NTU) 0.277 (84.1%) 0.601 (83.4%) 0.692 (77.7%)

COD (mg/L) 16.5 (87.9%) 19.4 (86.9%) 19.6 (87.5%)
Fig. 3. COD removal in three steps at different filtration rates. (a) Sample 2 and (b)
TOC (mg/L) 2.7 (80.8%) 4.9 (84.0%) 3.1 (83.1%)
Sample 3.
 C. Benito-Alcázar et al. / Journal of Hazardous Materials 178 (2010) 883–889 887

Table 5
Feed and permeate values of the studied parameters for different UF membranes.

Parameter Feed UF permeate 150 kDa UF permeate 30 kDa

Turbidity (NTU) 6.275 0.125 (98.0%) 0.079 (98.7%)

SDI15 >5 3.7 4.0

a consequence of the adsorption of colloids and dissolved organ-

ics on the GAC surface. In other studies about GAC filtration of
secondary effluents similar results were obtained. Kim et al. [10]
studied different pretreatments for RO. Dual-media filtration fol-
lowed by GAC filtration of the secondary effluent was tested and
the results showed a turbidity reduction of 70%, a COD removal of
75–85% and a TOC removal of 70–75%. It cannot be observed a clear
relationship between COD and turbidity removal and the filtration
rate (Figs. 3 and 4). Presumably, the variability of COD and turbid-
ity removal with the filtration rate may be due to the experimental
error associated with the measures. Petala et al. [22] worked on
reclamation of secondary effluents produced by an activated sludge
plant. They tested an advanced treatment consisting of sand fil-
tration, GAC adsorption and ozone disinfection. The reduction in
turbidity was around 60% after the GAC filtration stage, whereas
TOC removal efficiency achieved in this process was higher than
80%.
A 45 ␮m microfilter was used to measure the SDI15 of samples
of the GAC filtrate taken after the GAC filtration experiments had
concluded. This filter is an external unit and it is not placed in the
GAC filtration system. The SDI15 of the GAC filtrate could not be
measured due to the fact that the 0.45 ␮m microfilter was blocked
after few minutes of filtering. However, the observation of the filter
after the SDI15 test revealed a colourless surface and the absence of
particle deposits.

3.2. SEM analysis and IR spectra

SEM images of the SDI microfilter before and after plugging

in ‘GAC filtration’ experiments with the wastewater effluents are
shown in Fig. 5. Fig. 5a shows the SDI microfilter after filtration
of deionised water at 0.207 MPa to achieve microfilter compaction.
This image was taken to compare the original non-fouled SDI micro-
filter with Fig. 5b and c. No fouling layer was observed in Fig. 5a, as
expected. Fig. 5a provided a clear image of the microporous matrix
of the SDI microfilter. Fig. 5b and c shows the 0.45 ␮m SDI micro- Fig. 5. SEM images of SDI filter: (a) deionized water filter, (b) plugged filter and (c)
filter after the determination of the SDI15 of the GAC filtrate. The plugged filter enlarge.
difference between Fig. 5b and c is that Fig. 5c is an enlarged image
of Fig. 5b. These figures revealed that a fouling layer was formed on
the filter surface. This fouling layer had a gel appearance without
the presence of particle deposits or salt scaling. The enlarged image
of the plugged filter shows the gel layer in detail (Fig. 5c). The struc-
ture of the fouling observed on the SDI filters is presumably due to
the presence of organic and colloidal matter.
An IR analysis was performed to identify the organic functional
groups that were responsible of fouling in GAC filtration experi-
ments (Fig. 6). The IR spectra of the SDI fouling deposit revealed
the presence of various organic functional groups. The possi-
ble bonds were C–O (1100–1450 cm−1 ), C O (1550–2050 cm−1 ),
C–H (2800–3000 cm−1 ) and O–H (3100–3500 cm−1 ). Therefore the
plugging of SDI microfilters was a consequence of the deposition of
organic matter on the filter surface.

3.3. Ultrafiltration

In this set of experiments, turbidity and SDI were measured for

feed water and UF permeates. The SDI15 test performed with the
feed could not be finished due to microfilter plugging (SDI > 5). As it
is already explained in Section 3.1 when referring to Table 4; Table 5 Fig. 6. IR spectra of SDI fouling layer.
888 C. Benito-Alcázar et al. / Journal of Hazardous Materials 178 (2010) 883–889

only includes the important parameters that control RO membrane

fouling and that are significant to test if the UF pretreatment is suit-
able to produce a high quality RO feeding. Therefore, Table 5 shows
the values of turbidity and SDI15 of the feed and the UF permeate.
It can be observed that both UF membranes achieved a high tur-
bidity reduction (around 98%) producing a transparent permeate.
Moreover, the 150 kDa membrane as well as the 30 kDa membrane
showed an SDI15 around 4. This fact revealed UF as a proper pre-
treatment for this effluent to be fed in RO processes, because an
SDI lower than 5 is required for RO feeding according to membrane
manufacturers. However, an SDI lower than 3 is recommended to
minimize fouling and to ensure a successful RO membrane oper-
ation. Because of this, UF cannot be considered as a successful
pretreatment to produce high water quality for RO feeding. This
conclusion is not influenced by the value of other parameters (not
measured and not shown in Table 4 as they cannot modify this
result) because once one of the recommended parameters is out
of range, the pretreatment is not considered successful. Therefore,
NF and GAC filtration coupled with NF is investigated in the next
section of this work.
When comparing with other UF tests performed with secondary
effluents, the value of the SDI15 obtained is considerably higher
than the values found in the literature for other secondary effluents.
Qin et al. [23] studied the UF of a municipal wastewater secondary
effluent to test a pretreatment for RO reclamation. A hollow fiber UF
membrane with a pore size of 0.05 ␮m was tested. In addition, the
process was chemically enhanced with alum dosage. Product water
with an average SDI15 of 1.6 was produced. On the other hand, van
Hoof et al. [24] studied UF with secondary effluents of an IWWTP
as well as two MWWTPs. The SDI15 value of the UF filtrate was
1.8 on average. The discrepancy between the SDI15 values for the
UF permeates can be explained because the secondary effluents
of MWWTP and IWWTP are quite different and the IWWTP sec-
ondary effluents vary with the processes carried out in the factory.
As the initial tested effluent in the above mentioned works [23,24] Fig. 7. Flux vs. time of NF and GAC + NF for two different samples. (a) Sample 1 and
(b) Sample 2.
and the one considered in the present work are not the same, the
comparison of both of them must be considered with restrictions.
NF). Moreover, the results of the second sample showed the same
behaviour for both feedings regarding permeate flux (Fig. 7b) and
3.4. NF and GAC filtration coupled with NF permeate flux decline. In this case, the final flux decline for NF was
5.47% while GAC + NF showed lower value of the final permeate
The two samples of the pretreated secondary effluent used for flux decline (4.89%). A salt rejection around 70% and a permeate
this set of experiments were characterized. The COD and TOC values conductivity around 1 mS/cm were obtained for both samples.
of these samples are summarized in Table 6. Fig. 7 shows the perme- The characterization of the feed, the GAC filtrate and the NF and
ate flux of NF and NF with previous GAC filtration experiments for GAC + NF permeates is summarized in Table 6. It can be observed
Samples 1 and 2. There is small or no difference in terms of perme- that the reduction in COD and TOC was much higher for the
ate flux between NF and GAC coupled with NF presumably because first sample. Lower values of COD and TOC in the permeate were
GAC filtration does not remove the pollutants that are able to foul obtained compared to Sample 2, despite the feeding sample showed
the NF membrane. This means that, when GAC filtration is applied higher values. It can be observed that GAC, NF and GAC + NF elim-
to the petrochemical effluent of this work, it only removes certain inated around 90% of the measured parameters and that all the
pollutants. These pollutants are those that, in the case of the NF treatments removed more TOC than COD. Nevertheless, the reduc-
experiments alone, are swept away from membrane surface by the tion of these parameters for the second sample was lower than 85%.
tangential flow or go straight from the entrance of the NF module to NF removed more TOC, whereas GAC and GAC + NF removed more
the exit of the NF module due to tangential forces. The experiment COD. On the other hand, the SDI test resulted in an SDI15 of 0.5 for NF
with the first sample revealed similar results in both cases (Fig. 7a). permeate and an SDI15 of 0.7 for GAC + NF permeate. Turbidity val-
The rate of permeate flux decline was very similar during both tests ues are not shown in Table 6 because NF membranes retain all the
and at the end of the experiment (4.69% for GAC + NF and 4.83% for molecules retained by UF membranes and also smaller molecules

Table 6
Sample and treated water values of studied parameters for different treatments.

Sample Parameter Feeding sample GAC NF GAC + NF

1 COD (mg/L) 162 19.1 (88.2%) 15.4 (90.5%) 15.8 (90.2%)

1 TOC (mg/L) 34.48 3.27 (90.5%) 2.93 (91.5%) 1.66 (95.2)
2 COD (mg/L) 146 25.0 (82.9%) 33.8 (76.8%) 23.9 (83.6%)
2 TOC (mg/L) 24.00 6.56 (72.7%) 4.12 (82.8%) 5.00 (79.2%)
2 SDI15 >5 >5 0.5 0.7
 C. Benito-Alcázar et al. / Journal of Hazardous Materials 178 (2010) 883–889
(ions). Therefore, if in the case of UF experiments turbidity was neg-
ligible (lower than 0.1 NTU), in the case of NF experiments it will
be even lower.
The removal achieved in the experiments with GAC filtration
was similar to that attained in the previous tests (Table 4) and the
results observed in the literature [10,22]. Furthermore, COD and
TOC removal after NF was higher than that observed in another
study about nanofiltration of a petrochemical secondary effluent
[12]. Lora et al. [12] tested two different NF membranes (including
NF270 of Filmtec® ) with a secondary effluent of a petrochemical
plant. In that case, the COD and TOC removal achieved was around
60%.
For all the samples and the experiments performed, flux decline,
COD and TOC removal and SDI tests were similar for NF and
GAC + NF. In this way, the results did not show a clear benefit on
using GAC filtration prior to NF for this effluent. Therefore, both pre-
treatments could be suitable to reduce the effluent fouling potential
up to RO feeding requirements.
4. Conclusions
Regarding the studied parameters, the GAC filtration rate did not
affect significantly the quality of the filtrate obtained for this pre-
treated effluent. GAC filtration achieved an important reduction of
turbidity, COD and TOC. However, the SDI filter plugging indicated
a fouling potential on RO membranes. Therefore, the GAC filtrate is
not suitable to be fed in a RO process.
SEM analysis and IR spectra revealed that the fouling on the SDI
filters formed a gel layer of organic substances.
By using UF membranes, the reduction of the SDI up to RO
feed requirements (SDI15 < 5) was attained. However, as the SDI
was higher than 3, fouling problems can appear on RO membranes
during the operation with this feeding.
The NF process achieved the minimum value for SDI of the treat-
ments studied (SDI15 < 1), assuring a better quality of the RO feeding
for a successful operation. Moreover, according to the water quality
requirements (Table 1) NF could be suitable to produce cooling and
fire-extinction water. On the other hand, regarding the flux char-
acteristics, as well as the quality of permeate, there is no benefit on
GAC filtration prior NF.
It is interesting for further investigation to perform SEM and
IR analysis to evaluate the characteristics of the possible organic
foulants present in the filtrate steams of UF, NF and GAC + NF pro-
cesses, despite the SDI values of those streams being lower than 4.
Acknowledgement
The authors of this work wish to gratefully acknowledge the
financial support of the Spanish Ministry of Science and Technology
(MCYT) through the project no. CTQ2005-03398.
889
References
[1] A.S. Alidi, A multiobjective optimization model for the waste management of
the petrochemical industry, Appl. Math. Model. 20 (1996) 925–933.
[2] O. Botalova, J. Schwarzbauer, T. Frauenrath, L. Dsikowitzky, Identification and
chemical characterization of specific organic constituents of petrochemical
effluents, Water Res. 43 (2009) 3797–3812.
[3] M.E. Mohamed, M. Al-Dousary, R.Y. Hamzah, G. Fuchs, Isolation and charac-
terization of indigenous thermophilic bacteria active in natural attenuation of
bio-hazardous petrochemical pollutants, Int. Biodeter. Biodegrad. 58 (2006)
213–223.
[4] J.D. Van Hamme, A. Singh, O.P. Ward, Recent advances in petroleum microbi-
ology, Microbiol. Mol. Biol. Rev. 67 (2003) 503–549.
[5] D. Dean-Ross, J. Moody, C.E. Cerniglia, Utilization of mixtures of polycyclic aro-
matic hydrocarbons by bacteria isolated from contaminated sediment, FEMS
Microbiol. Ecol. 41 (2002) 1–7.
[6] M. Richardson, Environmental Xenobiotics, Taylor and Francis Ltd., 1996.
[7] J. Illueca-Muñoz, J.A. Mendoza-Roca, A. Iborra-Clar, A. Bes-Piá, V. Fajardo-
Montañana, F.J. Martínez-Francisco, I. Bernácer-Bonora, Study of different
alternatives of tertiary treatments for wastewater reclamation to opti-
mize the water quality for irrigation reuse, Desalination 222 (2008) 222–
229.
[8] J.J. Qin, M.N. Wai, M.H. Oo, K.A. Kekre, H. Seah, Feasibility study for reclamation
of a secondary treated sewage effluent mainly from industrial sources using a
dual membrane process, Sep. Purif. Technol. 50 (2006) 380–387.
[9] H.H. Chen, H.H. Yeh, S. Shiau, The membrane application on the wastewa-
ter reclamation and reuse from the effluent of industrial WWTP in northern
Taiwan, Desalination 185 (2005) 227–239.
[10] S.L. Kim, J. Paul Chen, Y.P. Ting, Study on feed pre-treatment for mem-
brane filtration secondary effluent, Sep. Purif. Technol. 29 (2002) 171–
179.
[11] Filmtec Case Histories, http://www.dow.com/liquidseps/lit/casehist.htm.
[12] J. Lora, D. Sanz, J.M. Gozálvez, A. Santafé, J.M. Arnal, M.C. León, S. Navalón,
Ultrafiltration and Nanofiltration Pre-Treatment of Reverse Osmosis to Reuse
Wastewater in a Petrochemical Industry, October 21–26, IDA World Congress
on Desalination and Water Reuse, Maspalomas, Gran Canaria, Spain, 2007,
IDAWC REF:(MP07-232).
[13] Hydranautics Technical Manual, http://www.membranes.com/.
[14] Filmtec Technical Manual, http://www.dow.com/liquidseps/lit/tech inf.htm.
[15] A. Mosset, V. Bonnelye, M. Petry, M.A. Sanz, The sensitivity of SDI analysis: from
RO feed water to raw water, Desalination 222 (2008) 17–23.
[16] M. Into, A.S. Jönsson, G. Lengdén, Reuse of industrial wastewater following
treatment with reverse osmosis, J. Membr. Sci. 242 (2004) 21–25.
[17] J.M. Wong, Testing and implementation of an advanced wastewater reclama-
tion and recycling system in a major petrochemical plant, Water. Sci. Technol.
42 (2000) 23–27.
[18] R. Arviv, D. Mourato, J. Coburn, ZeeWeed membranes for water reuse and
desalination applications, Membr. Technol. (2002) 6–12.
[19] Metcalf & Eddy, Ingeniería de aguas residuales: tratamiento, vertido y reuti-
lización, third ed., McGraw-Hill, 1995.
[20] APHA, Standard Methods for the Examination of Water and Wastewater, twen-
tieth ed., 1998.
[21] ASTM D4189-07 Standard Test Method for Silt Density Index (SDI) of
Water.
[22] M. Petala, V. Tsiridis, P. Samaras, A. Zouboulis, G.P. Sakellaropoulos, Wastewater
reclamation by advanced treatment of secondary effluents, Desalination 195
(2006) 109–118.
[23] J.J. Qin, M.H. Oo, H. Lee, R. Kolkman, Dead-end ultrafiltration for pretreatment of
RO in reclamation of municipal wastewater effluent, J. Membr. Sci. 243 (2004)
107–113.
[24] S.C.J.M. van Hoof, A. Hashim, A.J. Kordes, The effect of ultrafiltration as pre-
treatment to reverse osmosis in wastewater reuse and seawater desalination
applications, Desalination 124 (1999) 231–242.