Linking biofilm growth to fouling and aeration performance of fine-pore diffuser in activated sludge

Water Research 90 (2016) 317e328

Contents lists available at ScienceDirect

Water Research journal homepage: www.elsevier.com/locate/watres

Linking biofilm growth to fouling and aeration performance of fine-pore diffuser in activated sludge Manel Garrido-Baserba a, *, Pitiporn Asvapathanagul a, b, Graham W. McCarthy a, Thomas E. Gocke a, Betty H. Olson a, c, Hee-Deung Park d, Ahmed Al-Omari e, Sudhir Murthy e, Charles B. Bott f, Bernhard Wett g, Joshua D. Smeraldi h, Andrew R. Shaw i, Diego Rosso a, c, ** a

Department of Civil & Environmental Engineering, University of California, Irvine, CA 92697-2175, USA Department of Civil Engineering and Construction Engineering Management, California State University, Long Beach, CA 90840, USA c Water-Energy Nexus Center, University of California, Irvine, CA 92697-2175, USA d School of Civil, Environmental and Architectural Engineering, Korea University, Anam-Dong, Seongbuk-Gu, Seoul 136-713, South Korea e DC Water, 5000 Overlook Ave SW, Washington, DC 20032, USA f Hampton Roads Sanitation District, Virginia Beach, VA 23471-0911, USA g ARAconsult, Unterbergerstraße 1, A-6020 Innsbruck, Austria h United States Environmental Protection Agency, 1200 Pennsylvania Ave NW, Washington, DC 20460, USA i Black & Veatch, 8400 Ward Pkwy, Kansas City, MO 64114, USA b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 6 July 2015 Received in revised form 18 November 2015 Accepted 9 December 2015 Available online 18 December 2015

Aeration is commonly identified as the largest contributor to process energy needs in the treatment of wastewater and therefore garners significant focus in reducing energy use. Fine-pore diffusers are the most common aeration system in municipal wastewater treatment. These diffusers are subject to fouling and scaling, resulting in loss in transfer efficiency as biofilms form and change material properties producing larger bubbles, hindering mass transfer and contributing to increased plant energy costs. This research establishes a direct correlation and apparent mechanistic link between biofilm DNA concentration and reduced aeration efficiency caused by biofilm fouling. Although the connection between biofilm growth and fouling has been implicit in discussions of diffuser fouling for many years, this research provides measured quantitative connection between the extent of biofouling and reduced diffuser efficiency. This was clearly established by studying systematically the deterioration of aeration diffusers efficiency during a 1.5 year period, concurrently with the microbiological study of the biofilm fouling in order to understand the major factors contributing to diffuser fouling. The six different diffuser technologies analyzed in this paper included four different materials which were ethylene-propylenediene monomer (EPDM), polyurethane, silicone and ceramic. While all diffusers foul eventually, some novel materials exhibited fouling resistance. The material type played a major role in determining the biofilm characteristics (i.e., growth rate, composition, and microbial density) which directly affected the rate and intensity at what the diffusers were fouled, whereas diffuser geometry exerted little influence. Overall, a high correlation between the increase in biofilm DNA and the decrease in aF was evident (CV < 14.0 ± 2.0%). By linking bacterial growth with aeration efficiency, the research was able to show quantitatively the causal connection between bacterial fouling and energy wastage during aeration. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Aeration Fouling Fine-pore diffuser Biofilm Activated sludge Efficiency

1. Introduction * Corresponding author. ** Corresponding author. Department of Civil & Environmental Engineering, University of California, Irvine, CA 92697-2175, USA. E-mail addresses: [email protected] (M. Garrido-Baserba), [email protected] (D. Rosso). http://dx.doi.org/10.1016/j.watres.2015.12.011 0043-1354/© 2015 Elsevier Ltd. All rights reserved.

Aeration is commonly identified as the prime contributor to energy use in wastewater treatment or water reclamation (Reardon, 1995; WEF, 2009). Since the 1970s, fine-pore diffusers have become the most common technology employed in the

318

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

aeration systems of biological processes for municipal water resource reclamation facilities (WRRF) (Aberley et al., 1974). Such diffusers supply air in the form of fine bubbles (with diameter ~5 mm), to transfer oxygen into the water to meet the oxygen demand of the microbes in the biological process treating the incoming wastewater (Ashley et al., 1992; Chern et al., 2001; Metcalf & Eddy, 2014). Fine-pore diffused aeration is the most commonly used aeration system due to its high aeration efficiency (AE, kgO2 kWh
1), a result of elevated oxygen transfer efficiency (OTE, %) or its correction (SOTE, %), to standard conditions, (defined as the product of the a factor and the OTE at standard conditions, i.e. 20  C, 1.013$105 Pa, 0 mgTDS l
1, 0 mgDO l
1). Fine-pore diffusers have lower a factors (dimensionless ratio of SOTE in process to clean water) when compared to coarse bubbles or surface aerators (Stenstrom et al., 1984; USEPA, 1989) and so in order to compare the efficiency of oxygen transfer in process conditions for processes operating at different elevations, temperature, dissolved oxygen (DO) and salinity, the product aSOTE (%) is used. Overall, fine-pore diffused aeration systems are usually more energy efficient than other systems for domestic wastewater treatment provided if they are operated and maintained properly. In some cases a system may have a very low a factor (e.g. industrial waste with a high surfactant concentration), that makes the fine bubble system less efficient overall when compared to a coarse bubble or surface aeration system. However, fine-pore diffusers once installed also tend to lose their initial performance due to biofouling (organic) and scaling (inorganic) phenomena on and within the porous surface of the diffusers (Fig. 1) (Gillot et al., 2005; Hertle, 2015; Kim and Boyle, 1993; Rosso and Stenstrom, 2006; USEPA, 1989; Wagner and Popel, 1998). This fouling (organic and inorganic) has the doublenegative effect of greatly diminishing the OTE and increasing the backpressure on the blower, which significantly deteriorates aeration efficiency, and thus the system performance and economic sustainability (Henze, 2008; Kim and Boyle, 1993; Rosso and Stenstrom, 2006). To account for fouling with time in operation, a fouling factor F is introduced:

F¼

aFSOTE aSOTEðtÞ aF aðtÞ ¼ ¼ ¼ aSOTE aSOTEð0Þ a að0Þ

(1)

where:

aFSOTE ¼ standard oxygen transfer efficiency in process water after time t (%) ¼ aSOTE(t) aSOTE ¼ standard oxygen transfer efficiency in process water at initial time (%) ¼ aSOTE(0) a(t) ¼ alpha factor after time t (%) a(0) ¼ alpha factor at initial time (%) ¼ a Thus, over time in operation the a factor becomes an agglomerate parameter (aF):

aF ¼ aðtÞ ¼ að0Þ$F

Fig. 1. Illustration of biofilm development on the surface of fine-pore diffusers, with micrographs showing details of the diffuser material and biological and inorganic coating.

(2)

Transfer efficiency decreases with the time in operation (Kaliman et al., 2008; Rosso et al., 2008). Whilst hydrostatic water head remains constant over the lifespan of the aeration system, the dynamic wet pressure (or DWP in Pa, i.e. the head loss of the diffuser element) increases inexorably with time in operation as the diffusers foul. In addition they may experience an increase in rigidity and hardness (Kaliman et al., 2008; Kim and Boyle, 1993; Rosso et al., 2008; Wagner and von Hoessle, 2004a). Since the increase in DWP may place the aeration system operating point near the blower surge zone, fouling can become the limiting factor of

aeration processes (Henze, 2008). The coating of diffuser surfaces or orifices by a biofilm, through biofouling, is a composite of bacterial flocs, protozoa, soluble microbial products and extracellular polymeric substances, particulate and soluble substrate from the influent wastewater. In addition the precipitation of inorganic salts causes scaling of carbonates, sulfates, silica, etc which add to the change in diffuser material properties. In aeration practice, the a factor is defined to quantify the effects of wastewater contaminants on the oxygen transfer efficiency (Stenstrom and Gilbert, 1981). The aF factor decreases with

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

time, but in reality the a factor does not change, but it is F that is decreasing. In fact, a combination of diffuser fouling (the combination of biofouling and scaling) and diffuser ageing results in a deterioration of the diffuser performance. F is calculated both from the ratio of a or aSOTE over time, hence in its calculation (eq. (1)) there is no distinction between the fouling effects and the diffuser ageing effects. In this research project we studied the dynamics of fouling on fine-pore diffusers in activated sludge. The scope of this project included the biological analysis of the fouling biofilms and an efficiency analysis of the diffusers, but we did not separately report the effects of diffuser ageing. To date, the deterioration in diffuser efficiency with prolonged operation has been related to fouling characteristics (Kim and Boyle, 1993) and to time (Frey and Thonhauser, 2004; Kaliman et al., 2008; Rosso and Stenstrom, 2005; Wagner and von Hoessle, 2004a). However, the nature and dynamics of the microbial communities involved in aeration diffuser fouling remains largely unexplored. In particular, the influence of operating conditions and diffuser material composition on biofilm development and community structure is poorly understood. Although aeration has been under intensive study for decades, there has been limited focus on factors that affect fouling rates. Rosso and Stenstrom
duit (2008) investigated the link be(2005) and Gillot and He tween aF and SRT; in the 1990s. Schoenenberger and Wren (1996) showed that aF is a function of oxygen uptake; several researchers have shown an apparent decrease in aF at very high MLSS concentrations (e.g. Racault et al., 2010; Schoenenberger et al., 2003). Groves et al. (1992) compiled a summary of measured aF for 21 different systems. Some research has been carried out on membrane fouling in membrane bioreactor (MBR) systems (Menniti and Morgenroth, 2010; Trussell et al., 2007, 2006), but the results are not transferable to the membranes in aeration diffuser systems. However, there is currently there is no consensus on the exact correlation between fouling (as distinct from a or aF), diffuser membranes properties (i.e. material type, surface roughness, etc) and operating conditions (Dissolved Oxygen, influent characteristics, Solid Retention Time etc) in the activated sludge process, which is crucial for the development of effective tools and strategies for the monitoring and control of biofilm fouling in diffusers membranes. Since microbial adhesion is a prerequisite for membrane fouling, prevention of microbial adhesion and colonization on the membrane surfaces will have a major impact in preventing membrane efficiency decrease (Zhao et al., 2004). Numerous research attempts have been undertaken to chemically modify diffuser materials aiming at producing diffuser surfaces that are less susceptible to fouling (Huang et al., 2008; Zhao et al., 2004). Since diffusers have a porous structure, it is difficult to coat the inside surfaces of the pores uniformly. In addition, membrane coatings have poor durability, lacking resistance to cleaning agents and are prone to leaching (Hamza et al., 1997). Commercially available membranes may be coated or polycoated to enhance their fouling resistance, but there is still a lack of independent verification of the extent of fouling resistance. In comparison to other environments, the diffuser provides a unique habitat for biofilm formation: high availability of oxygen, a wide range of substrate, and a (often biocompatible) surface for adhesion. Convective flows transport particulates (including bacterial cells) from the mixed liquor to the diffuser surface, whereas fluid/air shear over the diffuser module may cause bacterial detachment from the diffuser. In addition to the substrate from the incoming wastewater available to the fouling biofilm, the soluble microbiological products (SMP) produced by the planktonic biomass serves as substrate for the sessile microbial community. The environment may be different from one treatment plant to

319

another, and it has been observed that biofilm formation on a diffuser surface is affected by operating-process variables (i.e., availability of soluble substrate, dissolved oxygen level, iron concentration in the mixed liquor, organic loading rate) (Kim and Boyle, 1993; Kim, 1993; Rieth et al., 1990). Therefore, fouling and the rate of aeration efficiency decrease depends on the diffuser and wastewater characteristics, as well as on the process conditions. However, to date there is no way to predict reliably fouling. The overall goal of this study was to systematically characterize and compare the biofilm dynamics and aeration performance of finepore diffusers in activated sludge. The six diffusers selected were operated continuously for 15 months in a pilot reactor. Molecular techniques were used to quantify and examine the microbial structure in the biofilms retrieved from the diffuser surface at different points in time. The first part of this research compared the increasing biofilm growth and the decreasing aeration performance, providing a link between the fouling phenomena and the decrease in oxygen transfer efficiency, measured as depression of alpha factors, which ultimately can be used to provide information on deterioration of aeration energy efficiency. 2. Material and methods 2.1. Pilot installation A testing tank was constructed with dimensions of 1.5  1.5  4 m (L  W  D) and was used for both clean and process water tests (Fig. 2). Clean water and off-gas tests were performed on 6 different fine-pore diffusers. The tank was continuously fed with activated sludge from the first aerobic zone of the activated sludge process, which is the zone expected to face the worst fouling conditions due to the highest localized oxygen demand. The tank was fed from the bottom and continuously discharged from the top to avoid foam collection. A set of six diffusers was installed inside the tank (Table 1): 4 membrane tubes (2  EPDM, polyurethane, and silicone) and two discs (EPDM and ceramic). Clean water tests preceded process water studies as performance benchmarks for the new diffusers. Process water tests were carried out to establish the performance baseline for new diffusers in wastewater (i.e., before fouling begins). The testing tank was installed at the local water reclamation plant operating in biological nutrient removal (BNR) mode, and part of the activated sludge was continuously fed with in the research tank and back to the main plant (Table 2). 2.2. Molecular biology analysis 2.2.1. Distribution of nucleic acid associated with membrane The membrane was removed sterilely from the holding container, placed on a cling wrap surface and a cutting template placed over the membrane. Care was taken to place the template over an area where the biofilm cover appeared uniform. A blade dipped in alcohol and flamed prior to each cut was used to excise a 1 cm  4 cm piece of membrane which was then divided into 4 equal sections. Three of the sections labeled by membrane type, or sampling location for each of the three replicates (a, b or c) were placed into a 2 mL microcentrifuge tube. The remaining segment was weighed and then dried at 104  C for 1 h. After drying, the filter was removed and placed in a vacuum sealed desiccator until reweighed according to Standard Methods (APHA) to determine dry weight. 2.2.2. Nucleic acid extraction The extraction method is a modification of Asvapathanagul et al. (2012) which is based on the protocol for mixed liquor samples by

320

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

Fig. 2. Schematic of on-site column testing at IRWD.

Yu and Mohn (1999). Three bead beatings (cell lysis) were employed for all samples and then each was resuspended in 30 mL of DNAse/RNAse free water. DNAse Treatment and reverse transcriptase processes were then carried out according to the manufacturer instructions (Lifetechonologies, NY). The same RNA concentrations were used prior to the reverse transcriptase process. Extracted DNA was stored at
80  C until analysis. 2.2.3. Expression of nucleic acid content The concentration of nucleic acid can be expressed as concentration (ng/ml) recovered from the extraction per ml, or as ng of nucleic acid per g based on dry weight biofilm of each membrane sample, or lastly as ng of nucleic acid per cm2 of membrane surface. 2.2.4. Concentration of DNA The concentration of DNA was determined on a Nanodrop 10000 (Thermo Scientific, Wilmington, DE) according to manufacturer instructions. A single reading per triplicate was taken, but periodically throughout the readings the instrument was checked for drift. If the reading was negative, the Nanodrop was zeroed again and the reading redone. 2.2.5. qPCR for total bacteria Each environmental nucleic acid extract was amplified in triplicate using qPCR. The 1055f (50 -ATGGCTGTCGTCAGCT-30 ) (Wendt,

1977) and 1392r (50 -ACGGGCGGTGTGTAC-30 ) (Ferris et al., 1996) primer sets were used to amplify a 353 base pair fragment of the total bacterial 16S rRNA gene and gene copies were quantified using qPCR with a dual labeled fluorescent 16STaq1115 (FAM-CAACGAGCGCAACCC-BHQ1) probe (Harms et al., 2003). The total bacteria qPCR protocol was modified from (Harms et al., 2003) and consisted of 2 min holding at 95  C, 15s denaturing at 95  C, and 45s annealing at 50  C per cycle for 30 cycles. The master mix was composed of 3.5 mM MgCl2, 1 Buffer, 200 nM dNTP, 0.75U AmpliTaq DNA polymerase, 200 nM of each forward primer and reverse primer, 150 nM probe, 100 ng ml
1 of BSA and brought to a final volume of 20 ml with sterile HPLC water, to which 5 ml of each DNA sample was added. The qPCR was run on an Eppendorf RealPlex EP (Eppendorf, Hamburg, Germany).

2.2.6. Laser scanning confocal microscopy (CLSM) and image analysis The diffuser membranes were carefully removed and cut into pieces of 5 cm  5 cm to analyze EPS carbohydrate using concanavalin A (ConA) conjugated with tretramethylrhodamine isothiocynate (TRITC) (Invitrogen Co., Carlsbad, CA, USA). Ceramic disc material had to be excluded from the analysis. Excess electrolyte solution was carefully drawn off from biofilm-covered membrane pieces by gently touching the edge of the specimens with a tissue paper (Kim-wipes). Then, 100 ml of ConA dissolved in 10 mM

Table 1 Summary of diffusers installed at IRWD and their characteristics (d ¼ diameter; L ¼ length).

**

Geometry

Material

Size (d  L)

Air flow range (m3 h
1)

Continuous operation flow (m3 h
1)

Pore size (mm)

Tube Tube Tube Tube Disc Disc

Polyurethane (PU) EPDM Silicone (SI) EPDM EPDM Ceramic

75 mm  1 m 75 mm  1 m 50 mm  1 m 50 mm  1 m 225 mm (disc) 225 mm (disc)

8.5e15.3 8.5e15.3 2.8e5.6 2.8e5.6 2.8e5.6 2.8e5.6

5,6 5,6 2,8 2,8 2,8 2,8

1.16 1.28 0.5 0.81 0.67 e**

Not measured.

± ± ± ± ±

0.02 0.02 0.01 0.02 0.02

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328 Table 2 Summary of feed sludge characteristics and existing conditions at IRWD during the test period. Sludge feed characteristics and process Parameters Mean cell retention time (d) BOD, Sec. Influent (mg/l) COD, Sec. Influent (mg/l) MLSS, Sec. (mg/l) Ammonia, Sec. Influent (mg/l) Nitrate, Sec. Effluent (mg/l) Avg. TSS, Sec. Effluent (mg/l) Phosphorus (mg/l) WWTP Average Flow rate (MGD) Influent composition Diffusers type

8 ± 2.7 150 ± 28 295 ± 39 2500 ± 400 22 ± 7 10 ± 2 30 Not measured 16 (60 500m3/day) 90% domestic EPDM discs

phosphate buffer (pH 7.5) was added to cover the biofilm samples and incubated in the dark at room temperature for 20 min. Microscopic observation and image acquisition were performed using LSM 510 Meta Two-Photon CLSM (Zeiss, Oberkochen, Germany). Fifteen images per diffuser (five per each section: top, side and bottom) were taken along the horizontal diameter to avoid bias due to the uneven biofilm distribution on the filter membrane. The images were compiled and processed using ImageJ software (downloaded from http://imagej.nih.gov/ij) to render a 3D composite image to show the spatial distribution along horizontal (coverage) and vertical (thickness) distribution on the membrane as indicated by the fluorescence intensity. To establish if a significant difference existed between the biofilm thickness on the control and treatment filters, a paired two-sample mean T-test using a 95% confidence level was performed. The mean difference (MD) and the p value were reported for the significant differences. 2.3. Oxygen transfer efficiency measurements The volumetric mass transfer coefficients (kLa, time
1), oxygen transfer rates in standard conditions (massO2 time
1), and oxygen transfer efficiency in standard conditions for clean water (SOTE, %) were calculated according to the ASCE Clean Water Standard (ASCE, 2007) using the ASCE DO Parameter Estimation Program (DO_PAR 1.08; downloaded from: http://fields.seas.ucla.edu/research/dopar/ ). The oxygen transfer efficiency in process water (aSOTE, %) was measured using the Process Water Protocol (ASCE, 1997).

321

the average for all diffusers for the studied period was 0.75 ± 0.05. It can be seen how decreased over time while the biofilm growth ensued. The highest values, meaning less fouling characteristics, were exhibited by the 50 mm tubes, being SI and 50 mm EPDM tubes the least fouled diffusers (0.63 and 0.69, respectively), whereas PU and 75 mm EPDM presented the highest fouling values (0.49 and 0.52, respectively). The observed decreasing trend seems to find a stabilization value around 50% of the initial value at t ¼ 24mo. Mention that as the testing tank was not equipped with DO control daily periods of anoxia may have occurred which may influence and increase fouling rates. The aforementioned overall aeration efficiency decrease to a stabilization value around 50% of the initial value at t ¼ 24mo is according to Rosso and Stenstrom (2006) who analyzed datasets from 94 field measurements and showed a clear pattern of performance decline with time in operation. The efficiency declined rapidly during the first 24 months of operation, and it may be suggested that any diffusers that remain indefinitely in operation may end up performing at less than 50% of its original efficiency. From this observation, however, a conclusion in absolute terms should not be drawn, for the is a measure of relative efficiency (for used vs. new diffusers), hence it is a measure of consistency and not necessarily of absolute efficiency. In fact, diffusers that in general tend to operate at high air flux thereby releasing larger bubbles exhibit stable and high values, indicating that their efficiency (expressed as aFSOTE) maintains a low but consistent value over time. 3.2. Molecular biology In all figures the DNA increase is described by power fits. This research found that the best-fit regressions for both DNA concentration and were power-based regression models, represented by smooth lines along the manuscript. Power regressions functions are properly not limited functions which do not allow the independent variable to be zero. The zero time and corresponding number of bacteria had to be eliminated from the data set for these plots. In the case of DNA concentration the regression model (eq. (3)) was an excellent fit which accurately capture the interspecific form for transient growth to the steady state (or almost plateau conditions) observed experimentally (Fig 3), with an average coefficient of determination (r2) of 0.95 ± 0.02.

3. Results ln (y) ¼ B*ln (x þ 1) þ A

(3)

3.1. Oxygen transfer efficiency No disruptive episodes in the influent water or in the process conditions were registered during the project period and consequently the biofilm developed naturally and unperturbed. For all diffusers tested in this project, and the five sampling events throughout the campaign, the flow-weighted average is shown in Fig. 3. DO was recorded before and after each off-gas. The average DO in the pilot plant during the studied period was 3.5 ± 1.5 mg l
1. It should be noted that the testing tank was not equipped with DO control, and therefore daily periods of anoxia may have occurred, thus influencing fouling rates. However, our elevated daily average was maintained to prevent such anoxic periods. The average DO values during the off-gas testing (three flowrates) for each diffuser were as follows: 75 mm EPDM (3.4 ± 0.5); 75 mm PU (3.4 ± 0.6); 50 mm EPDM (1.4 ± 0.3); 50 mm SI (1.48 ± 0.5); Ceramic disc (0.8 ± 0.3) and EPDM disc (1.8 ± 0.4). The efficiency parameters steadily decreased during the time in operation (Fig. 3). All membranes expressed fouling characteristics,

values also obtained an accurate representation (r2:0.84 ± 0.12) by fitting the same regression model, which reinforce the correlation between both parameters. During these experiments a plateau was not observed, yet the decreased pattern of alpha values appear to bottom at a plateau value, suggesting a maturation of the biofilm or at least of its effects. By extrapolating to an asymptote, the rate of fouling decline seems to stabilize at ~0.5 after 24e30 months. Similarly, the biofilm growth rate also seems to stabilize at t ¼ 24mo. The values indicate that once biofouling begins, it increases in all diffusers as if a relatively stable base/plateau of microbial species establishes on the diffuser surface. It must to be noted that although different diffusers materials were tested, the relative microbial abundance on the diffusers remained within the same order of magnitude, and that the observed maximum DNA concentration was on discs in comparison to tubes after t ¼ 15 m. Ceramic and EPDM discs presented the highest total DNA concentration at the end of the testing period (Fig. 3), amounting to 2.10$104 ngDNA cm
2diffuser and 1.94$104 ngDNA cm
2diffuser,

322

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

Fig. 3. Diffuser fouling and biofilm increase (expressed as DNA concentration) for the six fine-pore diffusers over the project period. The fouling values are flow-weighted averages for each diffuser at each time of measurement. Also, for each diffuser, each DNA point represent the average DNA concentration value obtained from nine replicate samples for each testing period.

respectively, whereas the tubes average was 1.4$104 ± 0.4 ngDNA cm
2 diffuser, being PU the diffuser presenting the lowest DNA concentration 0.8 ± 0.3 ngDNA cm
2 diffuser. Throughout the project, the diffusers experienced changes in color, texture, and for the membranes, in dimension (i.e., exhibiting permanent elongation). Such changes were paralleled by the DNA sequencing results, which correspond to variations of the microbial species (data not shown). Once the diffuser had been colonized, the total bacterial abundance on the diffuser surface reached a relatively stable plateau or pseudo-steady state. Fig. 4 shows the main microbiological indicators used for all membrane diffusers after 15 months. The extracellular polymeric substances (EPS) analysis was performed exclusively for biofilm samples from membranes, and ceramic disc was excluded due to the incompatibility between the ceramic medium and the staining process for confocal microscopy. The PU membranes presented the lowest values in all biological indicators in comparison to other membrane materials. The PU bacterial abundance (1.22 109 cells/cm2) and total DNA concentration (8.76 ± 1.75 103 ng/cm2) represent 74.4% and 74.6% respectively of the average concentration for all membranes (4.82 109 and 34.2 ± 6.59 103). Similarly the EPS value for PU is 34.7% lower than the average value of all membranes. The 50 mm EPDM tube ranked highest in bacterial count and EPS concentration, confirming visual inspection and previous research (Hansen et al., 2004; Rosso et al., 2008; Wagner and von Hoessle, 2004a, 2004b). The three EPDM membranes behaved differently, possibly due to the different chemical composition put into the rubber molds. However, the maximum value for each

biological indicator was always associated with one of the three EPDM membranes. In order to provide a better understanding, two replicate diffusers (same manufacturer, material, and model) where installed and tested in two different operational conditions (low- and highrate processes). Fig. 5 shows that biofilms harvested in a high rate process have in almost all cases higher DNA concentrations than in low rate processes (until eventually a plateau is reached). It is also observed how the same diffusers type exhibited consistently higher DNA concentration (including its standard error) in the high-rate process when compared to the low-rate, both after 5 months and after 12 months. DNA concentration after 12 months was proportionally higher than the DNA concentration after 5 months, and the difference between high- and low- rate processes was more evident after 5 months than after 12 months. Energy-dispersive X-ray spectroscopy (EDX) was used to compare the elemental composition of diffuser surfaces between new and 15-month diffusers (except ceramic). The compositional analysis demonstrates that approximately 85e90% of the attached material to the diffuser surface was of organic nature, whereas as 10e15% was inorganic (see Supplementary information). The results showed the relevance of the biological component in the foulant composition. High concentrations of C, N and O indicated a high content of organic matter (up to 90%) presumably from microorganism and microbial-based materials (i.e. EPS). Known inorganic species may be present in the fouling but a low concentrations in comparison to the organic matter. The highest inorganic concentration were potentially compounds based on silicone

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

323

Fig. 4. Comparative of the microbiological indicators for all membranes diffusers at the end of the studied period (t ¼ 15months). The top plot shows the bacterial count (cells per unit area) measured with qPCR. The middle plot shows the DNA concentration (mass per unit area) measured with Nanodrop. The bottom plot shows the intensity of fluorescence for the extracellular polymeric substances, measured with confocal microscopy. The bars represent one standard deviation.

(2.27%), magnesium (1.48%), ferric (0.23%) and calcium (0.22%). Thus, the maximum expected presence of the most common compounds is silica (2.2%), magnesium carbonate (1.45%), calcium sulfate (0.5%), calcium carbonate (0.2%), calcium phosphate (0.08%) and iron phosphate (0.08%). 3.3. Linking aeration efficiency and molecular biology A key aspect of these results is that the values of consistently decrease with time for all diffusers (Figs. 6e7). The trend follows the aforementioned phenomena whereby an increase in DNA concentration corresponds to a parallel decrease in oxygen transfer efficiency (expressed in these figures as the flow weighted average of aF, or ). The trend assessment of total DNA mass per biofilm area for EPDM and ceramic discs is plotted in Fig. 6. A

punctual subsidence in the DNA increase at t ¼ 12mo for all diffusers was attributed to the seasonality of biological growth (the winter season for this location was 0mo < t < 3mo and 12mo < t < 15mo). Such a decrease, although noticeable in all membranes, had a more marked effect on discs diffusers. The increase in total DNA concentration of both disc diffusers at the fifth sampling event (t ¼ 15 mo) was considerably higher than the immediately preceding event, although it is described well by the trend line prediction. Throughout the sampling campaign, the two discs behaved with more similarity in terms of DNA concentration than that of any of the membrane tubes (Fig. 7). By plotting the top and bottom graphs in Fig. 3 against each other, Fig. 8 can be produced. It can be stated that the initial performance in process water (i.e., aSOTE) is eroded by biofilm growth, quantified by the DNA concentration in the biofilm colonizing the

324

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

cm
2

Fig. 5. Comparison of the mass DNA per cm2 of diffuser area for four diffusers types after 5 and 12 months in operation at different loading rates (HR ¼ high rate process; LR ¼ low rate process).

diffuser surface. The average coefficient of variance regarding the correlation between biofilm DNA and the decrease in diffuser performance (expressed by aF) for the six diffusers was 14 ± 2% and it never exceeded the 36%. The coefficient of variance in SI and ceramic never exceed 17% whereas EPDM disc (with an average CV of 20 ± 4) exhibit the highest variance of 36%. Therefore, Fig. 8 indicates the decrease in is directly correlated to the increase of the biofilm growth, capturing and quantifying the parallel decrease of together with the increase on DNA concentration. On average an increase of 2-log in DNA concentration diminished by 20%, whereas a 3-log increase reduced the efficiency parameter close to 30%, and to approximately 40e45% when reaching what it seems a plateau around the 4- log concentration. Diffuser geometry showed slight influence in the correlation. 50 mm tubes presented slightly lower at 4-log concentration (~35%) in comparison to the average of the tubes (~43%) or discs (~48%).

4. Discussion Although similar in magnitude, the difference between discs and tubes may be understood by the diffuser shear forces. Diffuser tubes may experience higher surface shear exclusively in the top from the bubbles having a lesser sweeping effect at the side and bottom area while the discs release bubbles from all their surface. By comparing the mass of total solids for different locations of tube differs (i.e. top, side, and bottom), the effect of shear stress was evaluated. If shear stress was important in the accumulation of foulants in the tube diffusers, the mass of total solids at top and side locations would be less than that at bottom location. Statistical analyses comparing different groups based on magnitude of variance (i.e. ANOVA) have demonstrated that the mass of total solids at top and side locations were significantly lower than that at bottom location (P < 0.05). Thus, considering lower mass of total solids at top location, it can be deduced that higher shear stress caused lower accumulation of foulants, especially during early period of fouling. Thus, it can be inferred that the top location of tube diffusers had similar environment to the surface of disc diffusers in terms of shear stress, which will explain lower DNA concentration in disc at the beginning of the study. Besides, Kwok et al., 1998; €sche et al., 2002 observed that once a stable and robust miWa crobial consortium thrive to attach in highly shear stress conditions it develops a thicker and denser microbial population, which would explain the higher DNA concentration at t:15mo (2$104 ngDNA

for discs in comparison to 1.4$104 ± 0.4 ngDNA cm-2 tubes). diffuser The experimental design took the complexity and dynamics of the system into consideration. The investigation has shown that, although the EPDM disc diffuser had higher initial transfer efficiency, it decreased over time. In contrast, the transfer efficiency of the ceramic started lower but the decrease over the whole period was lower. These results are consistent with practice and numerous sources in literature where the alpha factors for ceramic discs were the lowest (ASCE, 1989; Libra et al., 2005, 2002). This is mainly because the size distribution of bubbles released by ceramic discs spans always to the larger range. The higher the air flow applied to a porous ceramic diffuser, the larger the average size of bubbles. This is true also for polymeric membranes, however for ceramic diffusers as the air flow increases more pores discharge the air flow and the larger pores discharge proportionally even larger bubbles. Also, as fouling produces pore obstruction, punched membranes lose pore count while ceramic may release air from pores that initially were inactive (i.e., the smaller pores with high pressure drop) but with time become an option at the expense of increased pressure drop and blower energy. Both the full-scale low-rate and the side stream pilot installation had the same 225 mm EPDM disc diffusers. Nevertheless, the aF registered for the pilot was lower (22.8 ± 4.7%) than the values measured in the full-scale plant using the same off-gas analysis. Previously, Rosso et al. (2012) used the same layout for accelerated fouling studies and the side stream tank reproduced fouling with a speed approximately five times that of the full-scale aeration system. The reason for accelerated fouling is because the air flow per diffuser in the side stream tank is fixed while the full-scale system varies the air flow under feedback from the DO control system. Thus, the conditions experienced in the research tank may have triggered the biofilm growth which significantly decreases the aeration efficiency, thus strengthening the evidence of an existing link between microbiology and aeration efficiency. Fig. 6 shows the results for the EPDM tubes. EPDM is currently the most common elastomeric material employed in fine-pore aeration, however, EPDM is not the only component in the formulation for these diffusers as other excipients and additives can be found in the mold. Thus, the ill-defined composition and the degradation of these different materials may define the diffuser performance. Moreover, the membrane geometry, including orifice shape, size, and density, may play a role in what is evident in Fig. 6, and in comparing Fig. 5 and: the values for EPDM membranes, they span approximately from 0.75 to 0.60 for 75 mm tubes and 225 mm discs, but only approximately from 0.60 to 0.45 for 50 mm tubes. However, the highest DNA concentration was recorded on the surface of EPDM discs, not on the 50 mm tube, indicating that although there is an evident correlation between the DNA concentration increase and the erosion in , the diffuser itself may experience mechanical ageing that adds to the decline in caused by the biofilm growth. Previous studies reported the material degradation of diffusers with time in operation (Kaliman et al., 2008; Rosso et al., 2008; Wagner and von Hoessle, 2004a), indicating that all elastomers experience alteration of mechanical properties, albeit with site-specific behavior. Two emerging elastomeric materials are polyurethane (PU) rubber and silicone (SI). Silicone membranes, not containing plasticizers or additives and being inorganic in nature, have been proposed as the best means to avoid problems because these are less affected by biological fouling and thickness variation than other materials (Wagner and von Hoessle, 2004a). Although the biological indicators in Fig. 4 do not confirm this hypothesis as SI membrane presents similar values at t ¼ 15mo to all the EPDM membranes, Fig. 7 shows that the lowest , on average, was diffuser

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

325

Fig. 6. Correlation between total biofilm DNA concentration and the decrease in diffuser performance (expressed by aF) for the six membranes. For each plot the coefficient of variation never exceed the value reported.

recorded for 50 mm SI tubes. Thus, despite the higher DNA concentration throughout the project, the SI tube maintained its efficiency (in relative terms) which was more stable over time, when compared to other emerging materials like the PU membrane. When comparing these materials with EPDM tubes, of the same geometry, pore size, and manufactured by the same entity, one may observe that although the DNA concentration increased to a similar extent, the dynamics are different. Consistently, the same was true for both 75 mm tubes. It should be noted that the 50 mm EPDM tube experienced earlier growth of DNA concentration but stabilized to a plateau similar to the other tubes. While the two EPDM tubes have patterns declining with similar curvature and trend, indicating that the higher surface area of the 75 mm EPDM

tube may have played a role in consistently maintaining a higher than that of the 50 mm EPDM tube. The same cannot be stated when comparing the 50 mm EPDM and 50 mm SI tubes or the 75 mm EPDM and 75 mm PU tubes. In fact, by comparing the trends of the 50 mm EPDM and 50 mm SI tubes, with identical orifice size and pattern, it can be observed that at the beginning of operation their values were comparable. However 50 mm EPDM tube rapidly departed the trend whilst the SI tube maintained an almost flat pattern. Conversely, the comparison of 75 mm EPDM and PU tubes shows the PU's prompt departure from the 75 mm EPDM decline. As mentioned earlier, the 50 mm tube experienced the highest microbial activity impact based on the head loss and visual inspection. According to these results, and supported by the higher

326

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

Fig. 7. (top) Images of the 225 mm EPDM disc diffuser with the corresponding aFactor for each flow rate and the biofilm growth rate expressed in DNA concentration over the 15 months period. Note that the visible dark areas within the diffuser membrane corresponds to the DNA extraction sampling points for each testing episode. (bottom) Images of the 225 mm ceramic disc diffuser with the corresponding aFactor for each flow rate and the biofilm growth rate expressed in DNA concentration over the 15 months period.

EPS content and DNA concentration, bacteria may be using carbonrelated softeners within the membrane as substrate. This may be due to the different EPDM formulations (EPDM is the main but not the only component of the mold). As biofilm density increases organisms at the surface may encounter limited nutrients, triggering the mining of substrate such as softener in the diffusers membranes composition (i.e., EPDM membranes). Softener consists for the most part of organic material and can probably be used by some bacteria as carbon source after enzyme induction. Breakdown products may serve as substrate sources for other bacteria on or near the surface. Wagner and von Hoessle (2004a) showed softener content in EPDM membranes can be degraded by as much as 74% when compared to new EPDM membranes. From the comparison of the two replicate diffusers installed in two different operational conditions (Fig. 4) can be seen that same diffusers type in high rate process exhibited higher DNA

concentrations in the whole period. These findings are in accordance with previous studies reporting higher rate processes enhance biofilm growth (Henze, 2008; Wagner and von Hoessle, 2004a). DNA concentration after 12 months was proportionally higher than the DNA concentration after 5 months, indicating that in both processes the biofilm matured during the year of operation. Also, the difference between high- and low-rate processes was more evident after 5 months than after 12 months, indicating that the biofilm ripening phase may have followed different progression, yet reaching a similar plateau. This may suggest that the difference between the low rate and the high rate is related to the initial colonization, which develops into a distinctive microbial consortium for each diffuser type. Therefore, higher organic loading at the beginning of the aeration tank stimulates biofilm formation on diffusers surfaces, but these decrease throughout the tank creating less biofilm on diffusers at the end of the tank (ASCE, 1989).

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

327

Fig. 8. (left) Average DNA mass per unit volume of EPDM membranes (50 mm and 75 mm) with their corresponding alpha factor over the 15 months period; (right) Average DNA mass per unit volume for PU (75 mm Polyurethane) and SI (50 mm Silicone) with their corresponding alpha factor over the 15 months period.

This implies that lower softener content will be found on the membranes closest to the influent. The direct implications of softener reduction would imply that cleaning procedures to remove biofilm attachment might not be completely effective since hardening due to the loss of softener reduces irreversibly. Moreover, organic/inorganic substances (including bacteria) adhering to slits or holes may not be totally removed, and even if it were possible to remove all material deposited on and around slits or holes, these would presumably clog forthwith. According to tests on membranes cleaned with chemicals, clogging appeared again after few months in operation (Redmond et al., 1992). Some studies have used physical measurements such as biofilm mass and thickness in order to assess the biofilm quantity (De Beer et al., 2004). Nevertheless, these approaches cannot identify the preliminary factors initiating biofilms and or define the differences in biofilm community structures across surfaces. Furthermore, these measurements fail to describe the biofilm characteristics and the microbial distribution within the sessile community. The use of total bacterial cells and DNA concentration were reported to provide more detail in microbial abundances in activated sludge when compared to suspended solid concentrations (Huang et al., 2010). Microbial measurements can represent the system performance better than suspended solids because not all mass in suspended solids corresponds to microbial cells. However, some limitations still exists as the total DNA obtained from biofilms includes both intercellular and extracellular DNA for prokaryotic and eukaryotic cells and further analysis is required to quantify and identify these fractions. Moreover, the mere measurement of DNA does not distinguish between viable and non-viable cells, but certainly is less biased than the suspended solids as surrogate for biofilm. Monitoring over time of bacterial indicators may allow investigators and engineers to detect these sudden increases, which, through their association with aeration performance decline, are an alternative avenue for monitoring oxygen transfer efficiency decline. Although diffuser performance may be monitored through

off-gas measurements, the microbiological analyses give insight and quantify the mechanistic cause of biofilm formation. 5. Conclusions Although each diffuser type presents its own performance and specific fouling response, the overall results provide the first evidence on the mechanistic relation between oxygen transfer efficiency and fouling phenomena (a combination of biofouling, scaling, and material ageing). High correlation between biofilm DNA and the decrease in diffuser performance (expressed by aF) was demonstrated (<14 ± 2% of coefficient of variance). The captured correlation showed that an increase of 2-log in DNA concentration diminished fouling factor by 20%, whereas a 40e45% was observed when reaching what it seems a plateau around the 4log concentration. Our results show that dynamic off-gas testing can effectively be used for monitoring the aeration system and to check design assumptions under operating conditions. This information can be used to improve the design of new aeration systems, setting the fundamental basis for selecting the best membrane diffuser according to the biofilm growth and diffuser fouling dynamics. Further, the use of new microbial techniques that allow for a rapid definition of community structure, the nature of the biofilm, phylogenetic tree (such as pyrosequencing) can be coupled with oxygen transfer efficiency measurements enabling identification of the role of microbial groups that are disproportionally in the biofilm compared to their prevalence in the floccular phase. Therefore, knowing the key microbial players, their affinity to diffusers' properties, the specific genes enabling the colonization process could be used in design or treat diffuser surfaces. Such studies will provide the fundamental basis to link microbial population dynamics, aeration efficiency, and process energy cost, and allow process designers and operators to identify the conditions that promote or discourage the disproportional growth of microorganisms most detrimental to aeration efficiency (hence, to the process

328

M. Garrido-Baserba et al. / Water Research 90 (2016) 317e328

energy intensity and cost). Acknowledgments This research was funded by the Water Environment Research Foundation (contract number INFR2R12) and the United States Environmental Protection Agency, with the support of the Irvine Ranch Water District, Hampton Roads Sanitation District, DC Water, and Southern California Edison. The authors thank David M. Hayden of the Irvine Ranch Michelson Water Reclamation Plant for the invaluable help and Alice K. Robinson of BKT for the help during field work. We thank Taek-Seung Kim for the EDX analyses. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.watres.2015.12.011. References Aberley, R., Rattray, G., Dougas, P., 1974. Air diffusion unit. Water Pollut. Control Fed. 895e910. ASCE, (American Society of Civil Engineers), 1989. In: Proceedings of the ASCE 1989, in: Environmental Engineering Specialty Conference. New York, USA, Texas. ASCE, 1997. ASCE Standard: Standard Guidelines for In-Process Oxygen Transfer Testing (18-96). New York. ASCE, 2007. Measurement of Oxygen Transfer in Clean Water [WWW Document] (accessed 25.02.15.). http://www.asce.org/templates/publications-book-detail. aspx?id¼8108. Ashley, K.I., Mavinic, D.S., Hall, K.J., 1992. Bench-scale study of oxygen transfer in coarse bubble diffused aeration. Water Res. 26, 1289e1295. http://dx.doi.org/ 10.1016/0043-1354(92)90123-L. Asvapathanagul, P., Huang, Z., Gedalanga, P.B., Baylor, A., Olson, B.H., 2012. Interaction of operational and physicochemical factors leading to Gordonia amaraelike foaming in an incompletely nitrifying activated sludge plant. Appl. Environ. Microbiol. 78, 8165e8175. http://dx.doi.org/10.1128/AEM.00404-12. Chern, J.M., Chou, S.R., Shang, C.S., 2001. Effects of impurities on oxygen transfer rates in diffused aeration systems. Water Res. 35, 3041e3048. http://dx.doi.org/ 10.1016/S0043-1354(01)00031-8. De Beer, D., Stoodley, P., Roe, F., Lewandowski, Z., 2004. Effects of biofilm structure on oxygen distribution and mass transport. Biotechnol. Bioeng. 43, 1131e1138. Ferris, M.J., Muyzer, G., Ward, D.M., 1996. Denaturing gradient gel electrophoresis profiles of 16S rRNA-defined populations inhabiting a hot spring microbial mat community. Appl. Environ. Microbiol. 62, 340e346. Frey, W., Thonhauser, C., 2004. Clogging and cleaning of fine-pore membrane diffusers. Water Sci. Technol. 50, 69e77.
duit, A., 2008. Prediction of alpha factor values for fine pore aeration Gillot, S., He systems. Water Sci. Technol. 57, 1265e1269. http://dx.doi.org/10.2166/ wst.2008.222.
duit, A., 2005. Predicting oxygen transfer Gillot, S., Capela-Marsal, S., Roustan, M., He of fine bubble diffused aeration systems e model issued from dimensional analysis. Water Res. 39, 1379e1387. http://dx.doi.org/10.1016/ j.watres.2005.01.008. Groves, K.P., Daigger, G.T., Simpkin, T.J., Redmon, D.T., Ewing, L., 1992. Evaluation of oxygen transfer efficiency and alpha-factor on a variety of diffused aeration systems. Water Environ. Res. 64, 691e698. http://dx.doi.org/10.2175/ WER.64.5.5. Hamza, A., Pham, V.A., Matsuura, T., Santerre, J.P., 1997. Development of membranes with low surface energy to reduce the fouling in ultrafiltration applications. J. Memb. Sci. 131, 217e227. http://dx.doi.org/10.1016/S0376-7388(97)00050-1. Hansen, E.J., Estevez, M.A., Es-Said, O.S., 2004. On the shrinking and hardening of EPDM rubber membranes in water sanitation filtration tanks. Eng. Fail. Anal. 11, 361e367. http://dx.doi.org/10.1016/j.engfailanal.2003.06.003. Harms, G., Layton, A.C., Dionisi, H.M., Gregory, I.R., Garrett, V.M., Hawkins, S.A., Robinson, K.G., Sayler, G.S., 2003. Real-time PCR quantification of nitrifying bacteria in a municipal wastewater treatment plant. Environ. Sci. Technol. 37, 343e351. http://dx.doi.org/10.1021/es0257164. Henze, M., 2008. Biological Wastewater Treatment: Principles, Modelling and Design. Hertle, A., 2015. Fine bubble diffused air aeration systems. Water Encycl. Huang, L.N., De Wever, H., Diels, L., 2008. Diverse and distinct bacterial communities induced biofilm fouling in membrane bioreactors operated under different conditions. Environ. Sci. Technol. 42, 8360e8366. http://dx.doi.org/ 10.1021/es801283q. Huang, Z., Gedalanga, P.B., Asvapathanagul, P., Olson, B.H., 2010. Influence of physicochemical and operational parameters on Nitrobacter and Nitrospira communities in an aerobic activated sludge bioreactor. Water Res. 44, 4351e4358. http://dx.doi.org/10.1016/j.watres.2010.05.037. Kaliman, A., Rosso, D., Leu, S., Stenstrom, M., 2008. Fine-pore aeration diffusers:

accelerated membrane ageing studies. Water Res. Kim, Y., Boyle, W.C., 1993. Mechanisms of fouling in fine-pore diffuser aeration. J. Environ. Eng. 119, 1119e1138. Kim, Y., 1993. Effects of biofouling on fine pore diffusers. J. Korean Environ. Eng. Kwok, W., Picioreanu, C., Kwok, W.K., Picioreanu, C., Ong, S.L., Van Loosdrecht, M.C.M., Ng, W.J., Heijnen, J.J., 1998. Influence of biomass production and detachment forces on biofilm structures in a biofilm airlift suspension reactor. Biotechnol. Bioeng. 58, 400e407. Libra, J.A., Schuchardt, A., Sahlmann, C., Handschag, J., Wiesmann, U., Gnirss, R., 2002. Comparison of the efficiency of large-scale ceramic and membrane aeration systems with the dynamic off-gas method. Water Sci. Technol. 46, 317e324. Libra, J.A., Sahlmann, C., Schuchardt, A., Handschag, J., Wiesmann, U., Gnirss, R., 2005. Evaluation of ceramic and membrane diffusers under operating conditions with the dynamic offgas method. Water Environ. Res. 77, 447e454. http:// dx.doi.org/10.2175/106143005X67359. Menniti, A., Morgenroth, E., 2010. Mechanisms of SMP production in membrane bioreactors: choosing an appropriate mathematical model structure. Water Res. 44, 5240e5251. http://dx.doi.org/10.1016/j.watres.2010.06.040. Metcalf, Eddy, 2014. Wastewater Engineering: Treatment and Resource Recovery. Racault, Y., Stricker, A.-E., Husson, A., Gillot, S., 2010. Effect of mixed liquor suspended solids on the oxygen transfer rate in full scale membrane bioreactors. Proc. Water Environ. Fed. 2010, 5853e5865. Reardon, D.J., 1995. Turning down the power. Civ. Eng. 65 (8), 54. Redmond, D., Groves, K., Daigger, G., Simpkin, T., Ewing, L., 1992. Evaluation of oxygen transfer efficiency and alpha-factor on a variety of diffused aeration systems. Water Environ. Res. 64, 691e698. Rieth, M., Boyle, W., Ewing, L., 1990. Effects of selected design parameters on the fouling characteristics of ceramic diffusers. Res. J. Water Pollut. Control Fed. 877e886. Rosso, D., Stenstrom, M.K., 2005. Comparative economic analysis of the impacts of mean cell retention time and denitrification on aeration systems. Water Res. 39, 3773e3780. http://dx.doi.org/10.1016/j.watres.2005.07.002. Rosso, D., Stenstrom, M.K., 2006. Economic implications of fine-pore diffuser aging. Water Environ. Res. 78, 810e815. Rosso, D., Libra, J.A., Wiehe, W., Stenstrom, M.K., 2008. Membrane properties change in fine-pore aeration diffusers: Full-scale variations of transfer efficiency and headloss. Water Res. 42, 2640e2648. Rosso, D., Jiang, L.M., Hayden, D.M., Pitt, P., Hocking, C.S., Murthy, S., Stenstrom, M.K., 2012. Towards more accurate design and specification of aeration systems using on-site column testing. Water Sci. Technol. 66, 627e634. http://dx.doi.org/10.2166/wst.2012.187. Schoenenberger, J.D., Wren, M.A., 1996. Modeling of alpha and oxygen uptake gradients: in municipal activated sludge processes e impacts on efficient aeration design. In: Federation, W.E. (Ed.), Annual Conference and Exposition -Water Environment Federation. Management; Innovative and Alternative Technology, pp. 515e526. Schoenenberger, M., Shaw, J., Redmon, D., 2003. Digester aeration design at high solids concentrations. In: 37th Annnal Wisconsin Wastewater Operators Association Conference. Wisconsin Dells, WI. Stenstrom, M.K., Gilbert, R.G., 1981. Effects of alpha, beta and theta factor upon the design, specification and operation of aeration systems. Water Res. 15, 643e654. http://dx.doi.org/10.1016/0043-1354(81)90156-1. Stenstrom, M., Vazirinejad, H., Ng, A., 1984. Economic evaluation of upgrading aeration systems. Water Pollut. Control Fed. 20e26. Trussell, R.S., Merlo, R.P., Hermanowicz, S.W., Jenkins, D., 2006. The effect of organic loading on process performance and membrane fouling in a submerged membrane bioreactor treating municipal wastewater. Water Res. 40, 2675e2683. http://dx.doi.org/10.1016/j.watres.2006.04.020. Trussell, R.S., Merlo, R.P., Hermanowicz, S.W., Jenkins, D., 2007. Influence of mixed liquor properties and aeration intensity on membrane fouling in a submerged membrane bioreactor at high mixed liquor suspended solids concentrations. Water Res. 41, 947e958. http://dx.doi.org/10.1016/j.watres.2006.11.012. USEPA, 1989. Fine Pore (Fine Bubble) Aeration Systems. Wagner, M.R., Popel, H.J., 1998. Oxygen transfer and aeration efficiency e influence of diffuser submergence, diffuser density, and blower type. Water Sci. Technol. http://dx.doi.org/10.1016/S0273-1223(98) 00445e454. Wagner, M., von Hoessle, R., 2004a. Biological coating of EPDM-membranes of fine bubble diffusers. Water Sci. Technol. 50, 79e85. Wagner, M., von Hoessle, R., 2004b. Biological coating of EPDM-membranes of fine bubble diffusers. Water Sci. Technol. 50, 79e85. €sche, S., Horn, H., Hempel, D.C., 2002. Influence of growth conditions on biofilm Wa development and mass transfer at the bulk/biofilm interface. Water Res. 36, 4775e4784. http://dx.doi.org/10.1016/S0043-1354(02)00215-4. WEF, 2009. Energy Conservation in Water and Wastewater Treatment Facilities, WEF Manual of Practice [WWW Document]. WEF Man. Pract. McGraw-Hill, Inc., New York, NY (accessed 25.02.15.). http://www.amazon.com/EnergyConservation-Water-Wastewater-Facilities/dp/0071667946. Wendt, H., 1977. Studies on phototoxic and photoallergic effects of cream, ointment and fatty ointment of fluocortin butylester (author's transl). In: ArzneimittelForschung, pp. 2238e2240. Yu, Z., Mohn, W.W., 1999. Killing two birds with one stone: simultaneous extraction of DNA and RNA from activated sludge biomass. Can. J. Microbiol. Zhao, Q., Wang, S., Müller-Steinhagen, H., 2004. Tailored surface free energy of membrane diffusers to minimize microbial adhesion. Appl. Surf. Sci. 230, 371e378.