Evaluating occurrence of contaminants of emerging concerns in MF/RO treatment of primary ef ﬂ uent for water reuse – Pilot study

This study experimented with the novel approach of using a micro ﬁ ltration (MF) and reverse osmosis (RO) treatment train to treat the ef ﬂ uent of a primary settling tank at the Inland Empire Utility Agency in Chino, CA. The pilot used polyvinylidene ﬂ uoride hollow- ﬁ ber MF modules as pretreatment for an RO skid, which used Hydranautics ESPA2 membranes in a two-stage con ﬁ guration with a feed capacity of 6 gallon per minute (gpm). In this pilot con ﬁ guration, researchers monitored the removal of 38 most prevalent contaminants of emerging concerns (CECs) through the MF/RO process. To investigate how operating the RO process at two ﬁ xed recovery rates of 55% and 80% would affect the performance of the MF/RO membranes, researchers applied different ﬂ uxes (8, 10, 12 and 14 gal/d/ft 2 (gfd)) and evaluated the removal of CECs in 1-stage and 2-stage RO con ﬁ gurations. The occurrence of CECs in the MF in ﬂ uent, MF ef ﬂ uent, RO permeate, and RO concentrate were analyzed and studied. In the ﬁ rst phase (1-stage the RO process), ﬂ ux of 14 gfd showed a better rejection value of inorganics (95.2%) when compared with those of other ﬂ uxes. Meanwhile, in the second phase (2-stage RO process), ﬂ ux of 12 gfd showed a better rejection of inorganics (93.7%) when compared with those of other ﬂ uxes. Although concentrations of CECs slightly decreased in the RO permeate as the ﬂ ux has increased, statistical analysis showed no signi ﬁ cant differences between different ﬂ uxes in terms of CEC rejection.


INTRODUCTION
The water industry is increasingly implementing recycled water projects to respond to current demands and challenges, such as water shortages, that the world faces today.
To develop future water supplies that remain sustainable in dry years, water managers and their communities will heavily rely on reclamation plants and their abilities to make wastewater a viable source of potable water.
Several options exist to beneficially reuse water. Indirect potable reuse (IPR) is one method of creating high-purity product water with reduced energy inputs and economic costs (Rodriguez et al. ). In this process, municipal wastewater is treated through a conventional treatment train, including aerobic biological treatment, and processed through membrane technology, then discharged directly into groundwater or surface water sources, which act as an environmental buffer (Leverenz et al. ).
Another method is direct potable reuse (DPR). This process entails full advanced treatment and can directly deliver water to a potable water treatment plant's supply without any environmental buffer. With that being said, regulations on implementing DPR are still in the premature stages of development.
Today, water managers are incorporating newly developed tertiary treatment processes to their IPR or DPR treatment trains to produce higher-quality water, especially as an opportunity for water reuse. However, whatever treatment method they select, these managers still face two distinct challenges that must be addressed: (1) contaminants of emerging concerns (CECs) in wastewater, especially in the concentrate stream that is disposed of to the environment and (2) the relatively high energy consumption per volume of product water of these advanced processes.
One of the key issues related to water reuse is the occurrence of CECs (Romeyn et al. ). Some contaminants, nants. At the least, CECs have been found to present potential risks to water supplies due to their physiochemical properties, such as poor degradability and high water solubility (Knepper et al. ). These properties allow CECs to pass through most common filtration steps including some membrane treatment processes.
Membrane processes are well used in water treatment, including IPR and DPR processes, given their ability to produce stable and excellent effluent quality (Reith &  As of yet, no comprehensive study has evaluated the occurrence of CECs in concentration streams produced by IPR and DPR-type treatment trains. Most research pertains to the water product produced by such systems since membranes are known to effectively remove most CECs; however, concerns remain about heightened exposure to CECs in the concentrate water that is disposed to the environment. The objectives of this study are to (a) evaluate the occurrence of CECs in MF/RO treatment of a WWTP's primary effluent and (b) demonstrate the effectiveness of MF/RO in treating primary effluent as a novel water recycling process.

Contaminants of emerging concerns
A CEC list was created according to an exhaustive study that identified the most common CECs in the literature. The top 38 common CECs were carefully selected and categorized by type. Table 1 presents a summary of the CECs examined in this study and their properties. The list consists of chemicals with high frequencies of occurrence and health risks.
primary treatment by preliminary screening and grit removal, primary clarification, secondary treatment by aeration basins and clarification, tertiary treatment by filtration and disinfection using chlorine, and finally dechlorination. The plant is designed to treat an annual average flowrate of 11.4 million gallon per day (mgd) (Inland Empire Utilities Agency ).

Raw wastewater
The pilot project used primary effluent wastewater from the CCWRF as feed to the MF and RO system. Table 2 shows the wastewater characteristics that the IEUA reported in 2012.

MF system
The MF pilot system was a fully automated membrane system designed and maintained by PALL Corporation. Its operational parameters were measured continuously at ten-minute intervals and automatically recorded. Total feed flow rate into the MF unit was 25 gpm. Average flux during the course of experiments was 13 gfd.
The pilot unit included a hot water heater and chemical pumps for automatic enhanced flux maintenance cleans that were carried out every 24 hours. The system was equipped with two new UNA-620A hollow-fiber MF modules, each

Mass balance
The mass balance was calculated following the method used involved are as follows: The discrepancy in the mass balance of CECs compounds can be calculated and presented as M discripancy . To estimate the mass of CECs that is lost due to the membranes' capabilities, the following equation is used: The mass balance discrepancy, in percentage, is calculated as follows:

Experimental procedure
The RO feed tank (500 gallons (1892.71 L)) was filled with MF permeate at a constant flow rate of 10 gpm. To start the RO process, the pressure of the feed water was gradually increased along with the pump speed. After the target pressure (i.e., 300 kPa) was achieved, the RO feed valve was gradually opened. At the same time, the concentrate valve and permeate valves were fully opened and initiated. By increasing and decreasing the pump rate and controlling the flow rate of the concentrate valve, the target flow rate in the permeate can be achieved. To achieve the target flux and recovery rate, permeate and concentrate flow rates were calculated and set in the RO unit by changing the set points of the feed valve, concentrate valve and the feed's pump rate.
The experiments in this study were performed in two phases. In the first phase, only 1-stage RO was operated using 4-inch elements with a fixed recovery rate of 55%.
This recovery rate was selected to evaluate whether recovery has any significant effect on the membranes' ability to reject CECs. The total membrane surface area used in this phase was 510 ft 2 (47.39 m 2 ). Four different fluxes of 8, 10, 12, and 14 gfd were selected and targeted under the constant recovery rate of 55%.
In the second phase of the study, were collected no sooner than three hours after the start of testing to allow the RO system to stabilize. RO samples were taken when permeate and concentrate conductivity were constant for at least for an hour with no feed temperature variations.
The sample volume was 8 L for organic compounds analysis and 2.5 L for inorganic compounds analysis.
Prepared amber glass (for organic compounds) and polynutrients and poly-metals (for inorganic compounds) bottles were used for sampling. Bottles contained sodium thiosulfate and ascorbic acid (for organic analysis) and phosphoric acid, sulfuric acid and nitric acid (for inorganic analysis) as preservatives. Samples were chilled to below 4 C on ice or frozen gel packs and delivered to the local, certified laboratory on the same day. All CEC and inorganic analyses were performed at this location.
Due to limited resources, the MF feed (i.e., primary effluent) and MF product (i.e., RO feed) were sampled and analyzed for CECs only once during the study. According to CCWRF, water chemistry of raw sewage to the plant does not have significant variations over extended periods of time. However, a municipal WWTP can experience daily variations in its feed water's water chemistry.

Analytical method
The collected samples were then shipped in the same day to the Weck Laboratories, Inc. in City of Industry, California.

RESULTS AND DISCUSSION
Occurrence and removal of inorganics in the MF CECs originate from industrial and domestic products such as pesticides, PCPs, preservatives, surfactants, flame retardants and perfluorochemicals. These contaminants are also excreted by humans in the form of human waste that contains pharmaceutical residues or steroidal hormones. CECs also surface as chemicals formed during wastewater and drinking water treatment, known as disinfection by-products (DBPs).

RO pilot operation data
As mentioned before, this study consisted of two phases: the first phase was an MF process followed by a 1-stage RO process using 4-inch ESPA2-4040 elements with a target recovery rate of 55%. The second phase was an MF process Removal of inorganics through the RO process The RO system's performance was evaluated in terms of the permeate's pollutant concentrations and the membrane rejection. The rejection of the RO membrane was calculated as follows: where C f , mg/L, is the feed concentrations and C p , mg/L, is the permeate concentration.     The original extraction and/or analysis of this sample yielded QC recoveries outside acceptance criteria. It was re-extracted/re-analyzed after the recommended maximum hold time.
c Low internal standard recovery possibly due to matrix interference. The result is suspect.  As can be seen in Figure 3, CECs with the lowest rejections were meprobamate, beta-blockers and BPs, which had 48.2%, 57.7% and 59.3% removal, respectively. CECs that were completely rejected were 1,4-dioxane and methadone.
Similarly, more than 98% of hormones and gemfibrozil were rejected. Other CECs with high rejection rates were iopromide with 99.5% rejection and atorvastatin with 95.6% rejection. High rejection rates also occurred for some antibiotics such as amoxicillin, azithromycin, ciprofloxacin, sulfamethoxazole and trimethoprim. In addition, high removal efficiencies of 93% were observed for compounds such as caffeine, cotinine and DEET.
As mentioned before, NDMA was poorly removed in RO because of its low MW.
The concentration of 1,4 dioxane in the RO feed was lower than the notification level of 1 μg/L. And while 1,4 dioxane was not observed in the RO permeate, its concentration was higher than the notification level in the concentrate stream.
As expected, the compounds rejected during the RO treatment were concentrated to different degrees in the RO concentrate stream. In this study, the highest concentration in the concentrate was of acetaminophen at 130 μg/L at a flux of 8 gfd, and the lowest concentration was of NDMA at 2.7 ng/L at a flux of 14 gfd.
The concentration of each compound in the concentrate stream was found for every test condition in phase one (i.e., with 55% recovery), and the results were compared against one another. The highest concentrations of CECs were found at a flux of 8 gfd and 55% recovery. CECs with the highest concentrations were as follows: 1,4-dioxane,  This study found that charged compounds could be rejected by more than 90%, regardless of other physicochemical properties. Although the charge of the CEC compounds was not analyzed in this study, CECs such as diclofenac, ibuprofen, sulfamethoxazole and triclosan, which are negatively charged, were rejected by more than 90% in both phases. In contrast, the rejection of noncharged compounds such as acetaminophen was found to be influenced mainly by their size. To assess the percent rejection of charged/ionic CECs, frequency distributions were plotted for observed RO rejection. Figure 4 shows the frequency of observed rejection for neutral and ionic/charged CECs.

Mass balance of CECs in the RO process
Tables 7 and 8 show the summary of the mass balance analysis using Equations (4) and (5). In an ideal situation with zero lab-analysis error, all M disc values would be zero.
With that being said, when calculating the mass of discrepancy using mass balance analysis via Equations (4) Tables 7 and 8 represents desorption, and a positive value of M disc represents adsorption.
Furthermore, positive and negative M disc values could be attributed to lab measurement errors. Tables 7 and 8 note varying laboratory procures such as 'The concentration indicated for this analyte is an estimated value above the calibration range.' Therefore, some level of error may have been introduced to the lab results. Measuring chemicals in the level of nanograms per liter can be a sensitive process that always comes with some uncertainties about quality control (i.e. result replicates).
Understanding the removal mechanism and relationships between controlling parameters in the RO system is key to optimizing CEC rejection. At the early stage of filtration in RO when the membrane is not ripe, the dominant mechanism for removal is the adsorption of nano-amounts of CECs into the membrane surface (Nghiem et al. a, b), until it reaches equilibrium. Preliminary removal could yield false results (Nghiem & Schäfer ). A cake develops on the surface of the membrane that decreases its pore size to below the nominal rating, thus improving removal (Nghiem et al. a, b; Xu et al. ), but later develops fouling.
In addition to the pore size decreasing, this improvement in removal could also be due to the enhanced adoption capacity of the solid phase (e.g. fouling biofilm).
The adsorption mechanism correlates with solutesolid hydrophobic interactions (Nghiem & Schäfer ; Nghiem et al. ). Hydrophobic interaction between the solid phase, particularly the RO membrane, and solutes is one of RO's important rejection mechanisms. A membrane's hydrophobicity is typically characterized by its contact angle, whereas hydrophobicity of solutes can be correlated and quantified using the logarithm of the octanol-water partition (log K ow ). Molecules with log K ow greater than 2 are referred to as hydrophobic. Octanol and water partition coefficient values are determined as logs, the ratio of the concentration in the octanol phase against the concentration in the aqueous phase at adjusted pH, such that the predominant form of the compound is unionized. Figure 5 shows the effect of Log K ow on the removal examined CECs for phases of the test.
Hydrophobic properties have an influence on the sorption mechanisms. For instance, strong hydrophobic compounds such as aromatic pesticides, non-phenylic pesticides and alkyl-phthalates were highly rejected even by the lowest desalting membrane (Kiso et al. ). However, the retention decreases as the membrane is saturated and its ability for sorption is reduced. As studied by Braeken et al.
(), hydrophobic molecules are rejected better than hydrophilic molecules after long-term operation.
In this study, hormones such as estrone and 17-β-estradiol, azithromycin and methadone, which have values of log K ow >2, adsorbed to the solid phase and potentially followed this pattern. See the mass balance calculation and the results in Tables 7 and 8.

CONCLUSION
The effect of CECs on the public health and the environment has urged water managers to more actively implement strategies that remove these compounds not only from drinking water but also from the wastewater treatment process. Primary treatment is currently unable to eliminate all substances; therefore, it is usually followed by secondary treatment. The results showed that 1-stage RO with a 55% recovery rate had a better removal rate of CECs when compared with 2-stage RO with a 80% recovery rate. As the concentration gradient of contaminants increased across the membrane at the higher recovery rate, the overall removal rate decreased for various compounds.
Azithromycin, hormones, carbamazepine, diazepam, gemfibrozil, atorvastatin, methadone and iopromide were removed the most effectively by RO in both phases. All these compounds have MW >200 g/mol and are also based on the log K ow . All those CECs also have hydrophobic characteristics; therefore, the RO process was able to remove them efficiently. In contrast, NDMA, propranolol, acetaminophen and meprobamate were the least effectively removed, given their low MW (less than 200 g/mol).

SUPPLEMENTARY DATA
The Supplementary Data for this paper are available online at http://dx.doi.org/10.2166/wrd.2019.004.