Performance of MF and RO processes for water recycling: Advanced
treatment after conventional wastewater treatment.
Jawad Al-Rifai a,b, Hadi Khabbaz c, William E. Price d
a Assistant Professor, Civil Engineering Dept., Engineering Faculty, Philadelphia University,
Amman, Jordan, [email protected] Associate Professor, Australian Geomechanics Society (AGS), School of Civil and
Environmental Engineering, University of Technology, Sydney, City Campus.
[email protected] Professor, Australian Institute for Innovative Materials (AIIM), University of Wollongong,
Wollongong. [email protected]
*Corresponding author. Tel.: 00962797960453
d Present address: Civil Engineering Dept., Engineering Faculty, Philadelphia University,
Jerash Road, Amman 19392, Jordan.
AbstractThis study attempts to reveal the fate of both organic and inorganic trace contaminants including
endocrine disrupting compounds (EDCs), pharmaceutically active compounds (PhACs) and
metals in dual membrane processes. For this purpose, water recycling plant operating with
microfiltration (MF) and reverse osmosis (RO) membranes was selected. The partition of trace
contaminants between RO permeates and brine was investigated. The RO membrane was found
to serves as a large reservoir for bulk organic matter as well as trace organic compounds (TOCs)
to the adsorption of contaminants on membranes and their likely release in the brine. There was
no clear relationship between the content of organic carbon and nitrogen in influencing the
distribution of both PHACS and EDCS in the RO streams. This might be due to the complexity of
RO feed in treatment plants and the difference between field experiments compared to bench
scale experiments where synthetic water was used by other researchers
The RO membrane concentrates the inorganic compounds (anions, cations and heavy metals) in
the brine by a factor of between three and five. Furthermore, the concentrations of TOCs reached
1
µg/L levels in the brine with concentration factor ranged between one and five for all detected
trace compounds. These results suggest that the tightness of the RO membrane is more significant
contributor to compound rejection when compared to compound properties (hydrophobicity,
water solubility and molecular weight).
Key words: Mass balance; Pharmaceutically active compounds; Endocrine
disrupting compounds; Brine; Reverse osmosis.
1. Introduction
Advanced wastewater treatment systems combine a microporous membrane process such as
ultrafiltration (UF) and microfiltration (MF), followed by RO membrane. This combination has
become the industry standard practice for the reclamation of municipal wastewater for industrial
and indirect portable reuse applications.
Many attempts have been made to estimate the performance of membrane separation in order to
predict the mass balance through membranes (Williams et al, 1999; Bowen et al, 2002). Most of
these models are based on one or more compounds in base water and require sophisticated
solution techniques. However, prediction of removal efficiencies for organic constituents is much
more challenging than calculations for inorganic compounds (ICs) since the water properties of
the compounds and interactions with membrane properties significantly affect the compound’s
mass transfer (Williams et al, 1999; Van der Bruggen and Vandecasteele, 2002). Siegrist & Joss
(2012) reviewed principles and capabilities for the most important membrane-based applications
for wastewater treatment. Bellona et al. (2004) have conducted a comprehensive survey in order
to identify factors affecting the rejection of organic compounds in NF or RO membranes. A
complete understanding of the solute and membrane characteristics that influence rejection could
lay the foundation for modeling the fate of specific compounds during a high-pressure membrane
application (Bellona et al, 2004).
2
Research investigating the viability of NF/RO membranes has reported the incomplete rejection
of organic micropollutants such as endocrine disrupting chemicals (EDCs), pharmaceutically
active compounds (PhACs) and others (Kimura et al, 2003; Schäfer et al, 2003; Kimura et al,
2004; Nghiem et al, 2004). Most of these studies examined the rejection of micropollutants from
a bench-scale flat sheet membrane unit or by using a dead end filtration module and high feed
water solute concentrations. In addition, most of these experiments utilized deionized water
spiked with one or more target solutes and a virgin membrane neglecting solution matrix effects
and fouling commonly observed in full scale applications. These factors may lead to
overestimation of membrane efficiency due to neglecting such factors. Furthermore, most
researchers focus on NF membrane and rejection mechanisms of micropollutants (e.g., PhACs
and EDCs) under influence of various factors such as natural organic matter and cations
(Comerton et al, 2009); organic fouling (Agenson and Urase, 2007); solute and solution
properties (Lَpez-Muٌoz et al, 2009); membrane fouling (Yangali-Quintanilla et al.).
The inorganic compounds (ICs) of wastewater are derived from either natural water or added
through domestic, commercial and industrial usage. The presence of ICs in water can cause
significant concern with respect to drinking water quality, aesthetics and industrial use. The
concerns over ICs range from their toxicity to their impact on process operation, product
operation and product quality in industrial processes.
A significant part of the anthropogenic production of ICs ends up in wastewater. Major industrial
sources include surface treatment processes with elements such as Cu, Zn, Ni, and Cr, as well as
industrial products that at the end of their life are discharged in wastes. Wastewater treatment
plants (WWTPs) are expected to control the discharge of ICs to the environment. Karvelas et al.
(2003) investigated the occurrence and the fate of eight heavy metals (i.e, Cd, Pb, Mn, Cu, Cr,
Zn, Fe, and Ni) during WWT processing in the city of Thessaloniki in Northern Greece. They
3
found that the heavy metals were detectable in the wastewater samples in a range of µg/L with
frequency of occurrence of about 90-100%.
This study attempts to reveal the mass balance of both inorganic and organic trace compounds in
dual membrane processes through the water recycling plant (WRP). The mass balances of the
bulk organic constituents, anions and cations and PHACS AND EDCS s through the RO
membrane were evaluated and the partition between RO permeates and brine in the RO was
investigated. Mass balance was also assessed in relation to Water properties of the compounds.
The quality of different streams of membranes were assessed using the general characteristics
(e.g., pH, chemical oxidation demand (COD), nitrogen and phosphorus) to evaluate the treatment
processes efficiencies.
2. Materials and Methods
2.1 Facility Overview
WRP investigated in study is located in Australia. The WRP produces 8.8 ML/day of water and is
capable of producing of 10.6 ML/day. The product water is delivered to a nearby refinery as
cooling tower make up, boiler feed water and other process uses. The WRP returns the rejects
(MF back flush and RO brine) back to the head of the WWTPs. The reject flows are 30% or
approximately 3.6 ML/day when the plant is producing 8.8 ML/day. The WRP is receiving
secondary effluent from municipal WWTP. The municipal wastewater passes through a grit
removal unit and a diffused air activated sludge process, while WRP comprises of automatic
backwashing 300µm screens, MF and RO membranes as shown in Figure 1.
2.1.1 Microfiltration
The MF membrane itself is protected from gross solids by the 300 µm screens (called Amiad
screens). Six MF racks are fitted with 66 filter modules installed in each rack. The MF uses 0.1
micron membranes, with a typical recovery of 97%. It is operated in cross-flow mode with 5-10%
recirculation flow to maximize membrane usage and flux. Backwashing is carried out at 20 4
minute intervals. Compressed air bubbling proceeds every backwashing cycle, shaking off
accumulated materials. A clean in place (CIP) process is carried out with NaOH / NaOCl and
citric solutions on a monthly basis.
2.1.2 Reverse Osmosis
RO membrane is the next stage taken up after MF. The RO treatment is a 3-stage process per
block with an array of 18, 8 and 5 pressure vessels in first, second and third stage, respectively.
The RO system consists of six RO blocks. Each RO block has thirty-one pressure vessels. Each
pressure vessel houses six RO membrane elements. The elements are formed as Spiral Wound.
The membranes are periodically cleaned by flushing water across the membrane and by a
chemical CIP process which involves the flushing of HCl and caustic solution through the RO
pressure vessels.
The overall recovery from the RO system is 85%. The RO membranes are progressively fouled
and therefore a CIP procedure is carried out approximately every 6 months initiated by the
operators. In order to prevent bio-fouling, chlorine is added to the MF filtrate in order to maintain
it at about 1-3 mg/L. The RO membranes do not have a high tolerance level to free chlorine.
Hence, ammonia is added to the MF filtrate to convert free chlorine to chloramine with a level of
2 mg/L. Finally, a post chlorination of the RO permeate is performed to maintain a chlorine
residual to meet the product water specification. Furthermore, additional chemical treatment is
used to adjust water pH and chlorination. The RO membrane used in the plant is BW30 365 FR
(fouling resistant RO) manufactured by DOW/ Filmtec in a flat sheet configuration. The BW30
365 is a polyamide thin film with 99.995% solute rejection.
2.2 Target Analytes
2.2.1 Organic Trace Compounds
These compounds were categorized in two main groups: EDCs (i.e. bisphenol A and
nonylphenol) and PhACs (e.g., salicylic acid, diclofenac and carbamazepine. Selected PhACs
5
represent a wide range of water properties of organic compounds were classified into acidic and
neutral compounds. Table 1 shows molecular weight, partitioning coefficient (log Kow),
dissociation constant (pKa) and solubility of the target PhACs and EDCs. These characteristics
are used in predicting their behavior under clinical conditions and are used in the environmental
assessment (Table 1).
2.2.2 Inorganic CompoundsMore than thirteen of ICs were included in this study. These compounds were Aluminum (Al),
Arsenic (As), Barium (Ba), Boron (Br), Cadmium (Cd), Calcium (Ca), Chromium (Cr), Cobalt
(Co), Copper (Cu), Fluoride (F), Iron (Fe), Magnesium (Mg), Manganese (Mn), Molybdenum
(Mo), Nickel (Ni), Lead (Pb), Potassium (K), Selenium (Se), Silicon (Si), Sodium (Na), Sulfur
(S) and Zinc (Zn).
2.3 Sample CampaignsSamples were taken from various inlets/outlets of the processes of WRP as shown in (Figure).
Manual grab-sample collection was accomplished by either pouring directly from a tap into a 2 L
amber glass bottle or by the use of a small bucket and pouring into the bottle. The glass bottles
placed on ice using an ice box to keep the samples cold and transferred overnight to the
laboratory. Seven sets of samples were taken through the whole year.
2.4 Analytical MethodsTotal organic contents (TOCs) were determined by TOC analysis (Method 5310B) (American
Public Health Association (APHA), 2005) using a Shimadzu TOC-VCSH (Total Organic Carbon
Analyzer) equipped with ASI-V auto-sampler (Alvarez-Salgado and Miller, 1998). The TOC was
determined by measurement of non-purgeable organic carbon (NPOC- the fraction of TOC not
removed by gas stripping). The nitrogen content was determined by measuring total nitrogen
(TN) using a Shimadzu Total Nitrogen Module (TNM1) coupled with the Shimadzu TOC–VCSH
using a chemiluminescence detector (Alvarez-Salgado and Miller, 1998). Ultraviolet absorption
6
(UV) was measured at wavelength of 245 nm according to Method 5910 using Shimadzu UV-
Visible Spectrophotometer (model UV 1700 Phara Spec) (APHA, 2005). Turbidity (T)
measurements were performed according to Method 2130 (American Public Health Association
(APHA), 2005) using a turbidity meter (HACH, model 2100N, HACH, S.A/N.V, USA).
Electrical conductivity (EC) measurements were performed according to Method 2510 (Lisitsin et
al, 2005) with a conductivity meter as a surrogate parameter. Measurement of pH was performed
according to Method 4500-H+ (American Public Health Association (APHA), 2005) with a pH
meter.
Trace inorganic compound (ICs) (i.e. cations, anions and heavy metals) were measured by
Inductively Coupled Plasma-Atomic Emission Spectroscopy (Method 3120) according to the
Standard Methods for the Examination of Water and Wastewater (American Public Health
Association (APHA), 2005). Elements and their MDL (mg/L): Al (0.005), As (0.005), Ba (0.005),
Br (0.010), Cd (0.001), Ca (0.2), Cr (0.002), Co (0.005), Cu (0.001), F (0.001), Fe (0.002), Mg
(0.2), Mn (0.001), Mo (0.01), Ni (0.002), Pb (0.005), K (0.5), Se (0.01), Si (0.05), Na (0.5), S (1)
and Zn (0.005).
Upon receipt in the laboratory, samples were filtered using three different filters, GF/D (2.7 µm)
and GF/F (0.7 µm) Whatman filters and 0.48 µm Nylon filter membrane (Alltech, Australia).
Filtered samples were kept in amber bottles overnight at 4oC. The next day, the samples were
allowed to reach room temperature and adjusted to pH 2-3 by addition of 4 M sulfuric acid to
enhance trapping of acidic compounds on the solid phase extraction (SPE) sorbent. MilliQ water
(1 L) was also spiked with a standard mixture of the investigated compounds to confirm recovery
of analytes. Samples were analyzed in batches consisting of 5-6 samples, spiked samples and a
blank.
For solid phase extraction (SPE), 60 mg Water Oasis HLB sorbent cartridges (Waters, Australia)
were used. The SPE was performed on a 24-fold extraction manifold (Supelco, Visiprep 24). The
SPE cartridges were conditioned sequentially with 5 mL methyl tetra-butyl ether (MTBE), 5 mL 7
methanol and 5 mL MilliQ water prior to use. Extraction of the 1 L sample was carried out under
vacuum at a flow rate of approximately 15 mL/min. After sample loading the cartridge was
washed with 3 mL (5% v/v) methanol in water. In order to eliminate the presence of water from
the eluant, a column of anhydrous sodium sulfate was prepared and fitted under the SPE column
before the elution procedure started. The SPE columns were eluted with 5 mL (10% v/v)
methanol in MTBE. The elution volume was then evaporated to dryness at 39oC under a stream
of nitrogen.
In order to determine PhACs and EDCs concentrations, a derivatization step was necessary. The
extract residues were dissolved in 300 µL of acetonitrile and then derivatized by adding 100 µL
of BSTFA (N,O-bis (trimethylsilyl) trifluoroacetamide) and TMCS (trimethylchlorosilane)
(99:1). The analytes were allowed to react for 1 h at 70oC. Finally, 100 µL of fluazifop standards
were added to each sample before injection as an instrument internal standard to confirm injection
of each sample onto the GC column.
2.1.1 Identification and Quantification of CompoundsA Shimadzu-GC 17A gas chromatograph was used for identification and quantification of
compounds, equipped with an auto-injector model AOC-20i, mass detector model QP5000,
Phenomenex Zebron ZB-5 column and Split/Splitless injector. The oven temperature program
was 100 oC; 30 oC /min.; 150 oC (4 min); 3 oC/min; 19 oC; 1 oC/min; 205 oC (5 min); 30 oC /min;
250 oC (3 min). The injection port was maintained at 270 oC and operated in splitless mode.
Helium was used as a carrier gas (flow rate 1 mL/min) and the interface temperature was held at
270 oC. For identification of each analytes, three compound specific ions were recorded in the
single ion monitoring mode (SIM).
Deuterated internal standards (acetaminophen-d4, carbamazepine-d10, gemfibrozil-d6, ibuprofen-
d3, 4-n-nonylphenol d6, phenytoin-d10 and salicylic acid-d6,) were used to increase accuracy of
the analytical procedure. Internal standards were added to the initial water sample and were
8
followed through the entire analytical steps. Quantification was carried out by calculation of the
response factor based on the area of the target analyte and deuterated standard.
2.1.2 Method ValidationExtraction recoveries of target compounds were determined by using influent and effluent of
WWTP spiked by a mixed analytes at a concentration of 50 ng/L and the calculated mount
compared with the spiked concentrations. The recoveries vary between 70 and 92%. Standard
calibration curves generated using linear regression analysis gave generally good fits to the data
(i.e. R2 > 0.97) over the established concentration range (5-50 ng/L, excluding where this
concentration range fell below the detection limits of a particular compound). A five-point
calibration was performed daily and the possible fluctuation in signal intensity was checked by
injection standard solution at two concentration levels after each 8–10 injections.
The reproducibility of the method was studied by analyzing five replicates of the recovery
samples to ensure correct quantification. Reproducibility calculated as errors with 95%
confidence interval, which was found ranging from 3.4–9.5% Method detection limits (MDL) and
method quantification limits (MQL) were determined from spiked water samples, as the
minimum detectable amount of analyte with a signal to noise ratio of 3 and 10, respectively.
Method quantification limits ranged from 1-50 ng/L.
3. Results and Discussion
3.1 MF process
Maximum removal efficiencies of the MF process for the turbidity, TOCs and TN measurements
were 77%, 61% and 30%, respectively, while there were no significant changes in the EC
measurements. Similarly, the water characteristics of the MF backwash are in consequence with
the above results (Figure). Thus, the MF process provides an essential pre-treatment for the RO
by removing particulate and colloidal material from the feed but the removal is limited to
particles larger than the membrane pore size (Van der Bruggen et al, 2003a). 9
The MF membrane with pore size of 0.1 µm, did not show any significant separation of mass
flow of PhACs &EDCs through the MF process (see Table ). For example, the concentrations of
gemfibrozil were 140 and 855 ng/L for the MF permeate, which indicates that MF membrane
system did not reject the any of the PhACs &EDCs.
3.2 RO Filtration process
The average, rejection efficiency of the RO for turbidity, conductivity, UV absorbance, TOCs and
TN were 80%, 90%, 87%, 91% and 82%, respectively (Figure 2). The RO shows a superior
efficiency in reducing values of all conventional parameters to less than 1 mg/L for TN and total
organic carbon, less than 1 µS/cm and 1 NTU for conductivity and turbidity, respectively. These
investigations indicated the great advantage of using RO membrane in producing a high quality of
recycled water. In general, the rejections results of the RO membrane represent greatly reduced
pollutants in permeate. Lopez-Ramirez et al. (2006) found that the drinking water standards were
widely exceeded by the reclaimed wastewater for the RO membrane with multi barriers approach.
Furthermore, indicator micro-organisms were absent from the RO permeate, which would allow
safe reuse of water, such as irrigation of raw vegetables. In this study, there was no clear
relationship between the content of organic carbon and nitrogen in influencing the distribution of
TOCs in the RO streams. This might be due to the complexity of RO feed in treatment plants and
the difference between field experiments compared to bench scale experiments where synthetic
water was used by other researchers. Katsoyiannis (2005) found that organic pollutants favored
adjective transport in the dissolved phase of the treated effluent.
Not surprisingly, the RO membrane rejected most of the cations and anions on an average of 97%
and 78%, respectively (Table 3).The rejection of various cations either monovalent or multivalent
(i.e. Ca2+, Mg2+, Na+, K+ and Si2+) was steady with the highest value of 98% for calcium and
magnesium. Furthermore, the rejection of anions was similar to cations except for boron. The
rejection of boron was very poor (29%) leaving a maximum concentration of 0.16 ppm in
10
permeate and 0.40 ppm in brine, which might be a problematic to PO processes. The presence of
ions such as Ca & Mg in RO feed can influence the rejection of PhACs and EDCs. Some
researchers have reported that the presence of calcium alone may positively influence rejection
(Plakas et al, 2006) while other have observed a decrease in compound rejection in the presence
of both natural organic matter and cations (Comerton et al, 2009) either through influence on the
membrane charges, the interaction of compounds and humic acids with each other or the
membrane surface (Cho et al, 2000; Nghiem et al, 2005).
Various acidic pharmaceutical compounds were found in the feed that was derived from the
WWTP (Al-Rifai et al, 2007) and ground water (Radjenović et al. 2008). Radjenović et al. (2008)
recorded concentration of hydrochlorothiazide (58.6–2548 ng/l), ketoprofen (<MQL–314 ng/l),
diclofenac (60.2–219.4 ng/l), propyphenazone (51.5–295.8 ng/l) and carbamazepine (8.7–
166.5 ng/l). Clofibric acid and diclofenac were less frequently detected in the raw wastewater and
in the feed as reported in Al-Rifai et al. (2007). Therefore, the performance of RO against these
compounds could not be evaluated in the current experiment. However, Heberer (2002) has
reported that diclofenac could efficiently be removed from surface water or municipal sewage
effluents using membrane filtration. Other acidic compounds were detected in all feed samples as
shown in Figure 3. These concentrations were up to 841 ng/l for gemfibrozil in feed, 190 ng/L for
diclofenac in permeate and 2 390 ng/L for gemfibrozil in brine. The rejection efficiencies for
acidic pharmaceuticals ranged between 30% and 100% (Figure 4a). Notably, the mean rejection
increased from 68% to 93% with increasing log Kow from 1.2 to 4.8 for salicylic acid, naproxen,
ketoprofen, ibuprofen and gemfibrozil. It has been reported that the rejection of solute by NF/RO
membranes is affected by a wide range of parameters such as feed pH, solute charge expressed
through pKa, molecular weight and geometry, polarity and hydrophobicity, as well as the
membrane surface charge (Joss et al, 2011; Van der Bruggen et al, 1998; Van der Bruggen et al,
1999; Kiso et al, 2000; Kiso et al, 2001; Ozaki and Li, 2002; Kimura et al, 2003; Kimura et al,
11
2004).The rejection efficiency of RO shows weakly positive correlation between the log Kow and
pKa, and acidic PhACs . Surprisingly, these results were not in accord with the findings of
Jjemba (2006) who could not correlate the solubility, log Kow and pKa with predicting the
behavior of PhACs concentrations in the environment.
The average concentrations of neutral PhACs found in the RO feed, permeate and brine were up
to 550, 350 and 1 380ng/L, respectively (Figure 3). Joss et al. (2011) show that most organic
micropollutants are degraded and retained to below the limit of detection (10 ng/L) in an MBR
plant followed by RO, except for small and polar compounds such as the anticorrosive
benzotriazole, as well as persistent pharmaceuticals propranolol, diclofenac and carbamazepine.
In general, rejection efficiencies of these compounds were from 63% to 100% with an average of
82% (Figure 4b). Similar results were observed by (Joss et al 2011).
Rejection efficiency of neutral pharmaceuticals showed a weak positive correlation with log Kow
as well as the pKa. Tolls (2001) indicated that log Kow may not be good indicator of the behavior
of pharmaceuticals in the environment. Similarly, Comerton et al,(2008) observed that the
tightness of the RO membrane is more significant contributor to compound rejection when
compared to compound properties (e.g., hydrophobicity, log Kow, water solubility and volume).
The rejection of nonylphenol and bisphenol A varied greatly especially for NP. Rejection varied
between 0 to 100% for nonylphenol and 53% to 100% for bisphenol A (Figure 4c). Due to wide
range of rejection efficiency variability and limitation of data, there was no potential to determine
any relationship between the rejection and molecular weight or molecular size in this case study.
On the other hand, Kimura et al. (2003) identified a linear relationship between molecular weight
of the non-charged compounds and rejection. However, in this study, such a clear relationship
was not observed.
12
4. Discussion
Wastewater effluent contains significant amounts of TOCs and ICs as a result of incomplete
removal by WWT processes. MF membrane was found to reject particulate and colloidal material
to provide an essential pre-treatment for the RO membrane. However, low pressure membrane
(MF and UF) with common pore size of well above several thousand daltons are ineffectively in
totally removing PhACs & EDCs from the MF feed. On the other side, RO membrane was found
to be capable of significant rejection of most target compounds, with an average of 78% to 97%
for metals, 68% to 100% for acidic pharmaceuticals and 63% to 100% for neutral
pharmaceuticals. There were large variations among the removal of EDCs by the RO membrane
systems. For example, nonylphenol rejection ranged between zero and 100%, and due to this
variation no further conclusion was drawn for ECDs removal by RO systems. Excellent overall
performance of both NF and RO was noted by Radjenović et al. (2008), with high rejection
percentages (>85%) for almost all of the PhACs investigated including acetaminophen,
gemfibrozil and mefenamic. Although water characteristics of compounds tested in this study
varied over a wide range, a relationship between any of these and rejection could not be
described. No strong relationship between the rejection of PhACs and EDCs, and their water
properties was observed for the WRP. These results suggest that the tightness of the RO
membrane is more significant contributor to compound rejection when compared to compound
properties (hydrophobicity, water solubility and molecular weight).
Rejection mechanisms by RO were investigated by many researchers on bench scales(Ozaki and
Li, 2002; Van der Bruggen et al, 2003b) . Rejections may be influenced by dipole moment of
compounds, hydrophobicity of compounds represented by Kow and molecular size.
Most studies on rejection mechanisms by RO membrane were conducted on pilot or bench scales
using virgin membranes, high concentrations and using either base or synthetic water. These pilot
scales were run under ideal conditions to investigate such mechanisms. Several theories were 13
proposed by many researches (e.g, dipole moment of compounds, hydrophobicity and molecular
size) but these theories were failed under actual plants, Snyder and co-workers (2007) suggested
that the compounds which breached RO under the full scale were not consistent, and no clear
relationship between molecular structure and membrane could be established. Breaching of the
RO could be the result of diffusion into and through the membrane, short-circuiting of the
membrane or supporting media failure.
Acknowledgments
The project was funded through the Australian Research Council (ARC) Linkage project. The
author is greatly indebted to Prof. Andrea Schäfer for her guidance and support in the first part of
this research. The author thanks the personnel of the water treatment plants for organizing sample
collection and the provision of the required results. The authors did not reveal the name of the
WRP has been directed by the administrating office.
References
14
FIGURE CAPTIONS
FIGURE 1: A SCHEMATIC DIAGRAM FOR THE WATER RECYCLING PLANT
FIGURE 2: WATER CHARACTERISTICS OF RO FEED, PERMEATE AND BRINE
FIGURE 3: CONCENTRATION OF TRACE ORGANIC COMPOUNDS (NG/L) IN RO FEEDS,
PERMEATE AND BRINE FOR THE DIFFERENT SAMPLING PERIODS
15
Figure 1: A Schematic diagram for the Water Recycling Plant
* Sample number (3) Feed for Amiad screen; (2) MF Feed; (3) MF permeate; (4) RO feed; (5) MF Backwash; (6) RO Permeate; (8) RO permeate (Post to Chemical Conditioning (Caustic & Chlorine)); (9) RO Concentrate; (10) Product water
16
Figure 2: Water characteristics of RO feed, permeate and brine (A: Turbidity; B: Conductivity; C: Organic Carbon; D: Absorbance (245λ); E: Nitrogen; F: pH)
17
Sampling
date
Sampling
date
BP
ANP
GFZIB
U
KP
F
NP
X
AC
M
PM
D
SC
A
CM
Z
0
200
400
600
800
1000
1200
1400
May 2005
BP
ANP
GFZIB
U
KP
F
NP
X
AC
M
PM
D
SC
A
CM
Z
0500
1000
1500
2000
2500
3000
3500
4000
Sep. 2005
18
Brine
Feed
Permeat
e
Brine
Feed
Permeat
e
BPANPGFZIBUKPFNPXACMPMDSCACMZPHT
0
500
1000
1500
November 2005
BPANP
GFZDICIBU
KPF
NAPAC
M
PMD
SCA
CM
Z
PHT
02004006008001000
1200
1400
1600
1800
2000
Permeate
Feed
Brine
December 2005
19
Brine
Permeate
Feed
BPA
GEMIB
U
KPF
NAPAC
M
SCA
0
200
400
600
800
1000
1200
1400
Permeate
Feed
Brine
Feb. 2006
BP
ANP
GF
Z
IBU
NA
P
AC
M
PM
D
SC
A
0
200
400
600
800
1000
Perm eate
Feed
Brine
April 2006
20
BP
ANP
GFZIB
U
KP
F
NA
P
AC
M
PM
D
SC
A
CM
Z
0
500
1000
1500
2000
2500
3000
Permeate
Feed
Brine
June 2006
Figure 3: Concentration of trace organic compounds (ng/L) in RO feeds, permeate and brine for
the different sampling periods
BPA: bisphenol A; NP: nonylphenol; CLB: Clofibric acid; GFZ: gemfibrozil; DCF: Diclofenac; IBU: Ibuprofen; KPE: Ketoprofen; NPX: Naproxen; ACM: Acetaminophen; PMD: Primidone; SCA: Salicylic acid; CMZ: Carbamazepine; PHT: Phenytoin.
21
5. Contents
TABLE 1: GENERAL CHARACTERISTICS OF TRACE ORGANIC COMPOUNDS (BARCELO,
2007)..........................................................................................................................................................2
TABLE 2: WATER CHARACTERISTICS OF MF FEED, PERMEATE AND BACKWASH OF MF
PROCESS.................................................................................................................................................3
TABLE 3: CONCENTRATION OF TRACE ORGANIC COMPOUNDS (NG/L) IN THE MF
PROCESS.................................................................................................................................................4
TABLE 4: METALS COMPOSITION OF THE RO, BRINE AND PERMEATE..........................5
22
Table 1: General characteristics of trace organic compounds (Barcelo, 2007)
Compound Mol. Weight log Kow pKa
Diclofenac 296.2 4.51 4.15
Ibuprofen 206.2 3.97 4.91
Ketoprofen 254.3 3.12 4.45
Naproxen 230.3 3.18 4.15
clofibric acid 214.5 2.57 N/A
gemfibrozil 250.3 4.77 4.7
salicylic acid 180.2 1.19 3.5
acetaminophen 151.2 0.46 8.9
carbamazepine 236.3 2.45 7
phenytoin 252.3 2.47 8.33
primidone 218.3 0.91 12.26
bisphenol A* 228.3 3.32 10.1
nonylphenol* 220 4.48 10.3
* (Nghiem and Schäfer, 2006)
23
Table 2: Concentration of PhACs and EDCs (ng/L) in the MF process
Compounds Abbrv. Amaid
screen feed
MF feed MF
permeate
MF backwash
Sample no. 1 2 3 5
Range Range Range Range
bisphenol A BPA 100 - 1650 102 - 1750 140 - 1790 ND - 450
nonylphenol NP ND - 130 ND - 119 ND - 120 ND - 97
clofibric acid CLB ND ND ND ND
gemfibrozil GFZ 165 - 899 154 - 876 140 - 855 ND - 550
diclofenac DCF ND - 220 ND - 156 ND - 123 ND - 97
ibuprofen IBU 170 - 850 ND - 723 163 - 680 ND - 150
ketoprofen KPF ND - 199 ND - 180 ND - 180 ND - 500
naproxen NPX 110 - 263 100 - 324 60 - 295 ND - 290
acetaminophen ACM 111 - 250 120 -250 100 - 158 19 - 617
primidone PMD ND - 420 ND - 590 ND - 360 ND - 250
salicylic acid SCA 134 - 307 150 - 620 120 - 283 ND - 500
carbamazepine CMZ ND - 568 ND - 585 ND - 555 ND -536
phenytoin PHT ND ND ND ND - 90
ND: Not detected
24
Table 3: Metals composition of the RO, brine and permeate
Composition Average concentrations(1)Conc. Factor
(CF)
Ave.
RejectionSTD
Feed Permeate Brine
Sample no. 4 6 9
CATIONS
calcium (mg/L) 35.0 ND(2) 148.0 4.3 > 0.98 0.01
magnesium (mg/L) 35.6 ND 156.0 4.5 >0.98 0.02
sodium (mg/L) 314.0 12.7 1320.0 4.3 0.96 0.01
potassium (mg/L) 23.0 1.1 103.8 4.7 0.95 0.01
silica (mg/L) 8.8 0.3 34.0 3.9 0.97 0.01
∑ of Cations (me/L) 416.4 13.6 1761.8 21.6 0.97 0.01
ANIONS
bicarbonate (mg/L) 162.0 10.8 646.0 4.0 0.93 0.02
barium (mg/L) 0.05 ND 0.05 2.4 >0.78 0.16
boron as B (mg/L) 0.23 0.16 0.40 1.8 0.29 0.08
chloride (mg/L) 518.0 16.5 1674.0 3.5 0.97 0.01
sulfate (mg/L) 130.8 ND 466.0 3.6 >0.99 0.00
fluoride (mg/L) 0.21 0.07 0.79 3.8 0.66 0.26
∑ of Anions 811.3 27.5 2787.2 17.2 0.97 0.01
25
Composition Average concentrations(1)Conc. Factor
(CF)
Ave.
RejectionSTD
HEAVY METALS
iron (mg/L) 0.08 0.01 0.28 3.6 0.90 0.04
manganese (mg/L) 0.11 ND 0.5 4.2 >0.98 3.43
(1) Average based on detected concentrations in 5 samples through the season (2):ND: not detected, so that the rejection was calculated based on the detection limit as a value
26