Presented at the International Conference on Integrated Concepts on Water Recycling, Wollongong, NSW, Australia,14–17 February 2005.

0011-9164/06/$– See front matter © 2006 Elsevier B.V. All rights reserved

Desalination 188 (2006) 113–121

Fouling autopsy of hollow-fibre MF membranes inwastewater reclamation

Long D. Nghiem, Andrea I. Schäfer*

Environmental Engineering, University of Wollongong, NSW, 2522, AustraliaTel. +61 (2) 4221 3385; Fax +61 (2) 4221 4738; email: schaefer@uow.edu.au

Received 29 January 2005; accepted 29 April 2005

Abstract

Fouling in membrane filtration processes is problematic but inevitable as it occurs with the retention ofcontaminants that accumulate on the membrane surface. The causes of fouling are often specific, depending upon feedwater constituents, the membrane, and the operation regime. Therefore, it is desirable that a thorough investigation isperformed on fouled membrane elements of the affected plant. This technique is known as “membrane autopsy”, whichidentifies the cause of poor membrane performance, and hence, gives the opportunity to rectify or mitigate the problemand improve future plant design. The cause of membrane fouling at a small water recycling plant using a hollow-fibremicrofiltration system was investigated. A membrane autopsy protocol was developed for water recycling applicationsthat consists of four major steps: (1) tensile testing to investigate the membrane mechanical integrity, (2) direct visualinspection, (3) membrane surface analysis using field-emission environmental scanning electron microscopy (as wellas atomic force microscopy, although it is not used in this case) techniques, and (4) foulant constituent analysis. Resultsobtained from this study indicate that the membrane was fouled by a mixture of colloids and organic matters, enhancedby the presence of multivalent cations. Possible measures to mitigate fouling in this particular case have also beensuggested.

Keywords: Microfiltration; Water recycling; Wastewater treatment; Fouling; Membrane autopsy

1. Introduction

One of the major problematic and inevitableissues encountered in almost any membranefiltration plant is fouling, which is caused by a

*Corresponding author.

number of foulants including inorganic scales,microorganisms, particulates and organic matter[1]. Fouling often results in a severe productivityloss, premature module replacement and some-times variation in treated water quality (mem-brane retention). Therefore, it is desirable that a

L.D. Nghiem, A.I. Schäfer / Desalination 188 (2006) 113–121114

thorough autopsy investigation is performed onfouled membrane elements of the affected plant.However, membrane autopsy techniques are oftensophisticated and expensive and the literature onmembrane autopsy is relatively limited [2]. Todate, most studies focus mainly on spiral-woundmodules and nanofiltration and reverse osmosismembranes [3–7].

Hollow-fibre microfiltration (MF) membraneshave been successfully used in many waterrecycling schemes worldwide. Examples of waterrecycling plants using hollow-fibre MF mem-brane filtration in Australia are the GerringongGerroa Sewerage Scheme, Water Reclamationand Management Scheme at Sydney OlympicPark, the Illawarra Wastewater RecyclingStrategy project, Luggage Point, and Rouse Hill.Summaries of these schemes can be found in apublication by Radcliff [8]. Although widely usedand recognised as a well-proven technology forwater recycling applications, sporadic prematurefailure of the membranes does happen in somecases. This is mostly due to inadequate pre-treatment, inappropriate cleaning and operatingprotocols, and to a much lesser extent it may beattributed to unsuitable membrane materials andmodule design. Driven by the need to understandbetter the performance of hollow-fibre MF mem-branes in a water recycling context, this studyaims to:C investigate the fouling processes specific to a

water recycling plant using hollow-fibre MFmembranes,

C identify major foulants and fouling mechan-isms relevant to water recycling applications,and

C provide suggestions to improve the perfor-mance of the hollow-fibre MF membrane fil-tration process.

2. Description of the Taronga Zoo waterrecycling plant

Due to the contamination of Sydney harbour,

Taronga Zoo launched a program to improve thequality of the run-off in partnership with Clean upAustralia, Sydney Water Corporation and theDepartment of Health. The existing primary watertreatment system was upgraded to deal with theadditional contaminants from the run-off and thefirst flush run-off and wastewater from animalcage wash downs and moats was directed throughthis treatment plant. The treatment involves MFas well as ultraviolet treatment for disinfection.Photographs of the treatment plant are shown inFig. 1 and a schematic of the core treatmentprocesses in Fig. 2.

Wastewater mainly consists of stormwater,hose down run-offs, and moat fillings. Afterflowing through a screen and a grit removalchamber, the water is biologically treated with anANI-Kruger unit. Following intermittent biologi-cal treatment, the effluent is filtered using ahollow-fibre MF system before being stored in aholding tank.

The treated wastewater is reused by the zoo towash down animal enclosures, irrigating the zoo’sparks and gardens, refill moats, and flush publictoilets (see Fig. 3). Feed water quality was notmonitored extensively prior to this study, henceno analysis of likely foulants in the feedwater canbe provided. However, as stormwater, exhibit androad hose-downs make up most of the feed water,suspended solid content and turbidity are ex-pected to vary significantly. Current on-goingconstruction at the site is possibly further con-tributing to such variation and peak loads.Limited water analysis indicates that suspendedsolids can be as high as 585 mg/L (see Table 1).

3. Materials and methods3.1. Origin of the membranes

A hollow-fibre MF membrane element takenfrom the water recycling plant was used in thisstudy. The fibres were taken from one replacedmodule (all six modules are operated in parallel)

L.D. Nghiem, A.I. Schäfer / Desalination 188 (2006) 113–121 115

Fig. 1. Taronga Zoo water reclamation plant (A) overlooking Sydney Harbour and (B) micro-filtration unit.

Fig. 2. Schematic diagram of the wastewater treatmentprocess at the Taronga Zoo. Various intermediate storagereservoirs are not included in the schematic.

Table 1Influent quality to the wastewater reclamation plant(adapted from [9])

Minimum Maximum Average

Dry weather flow, m3/d

100 650 250

BOD, mg/L 2 185 29Suspended solids, mg/L

1 585 80

Total nitrogen, mg/L

2.5 27.1 7.2

Total phosphorus mg/L

0.2 4.5 0.9

Faecal coliforms cfu/100 mL

103 109 106

and it is anticipated that current constructionactivities in the zoo adversely affect the moduleperformance due to high suspended solids in therun-off.

3.2. Mechanical strength

Mechanical properties of the membrane fibreswere examined using an Instron 4302 tensile

L.D. Nghiem, A.I. Schäfer / Desalination 188 (2006) 113–121116

Fig. 3. Usage of recycled water in (A) animal moats and(B) toilet flushing in the zoo’s public toilets.

testing instrument shown in Fig. 4. Load cell andloading rate were 10 N and 50 mm/min, respect-ively. A sample length of 200 mm was selectedfor the study.

Fig. 4. Photograph of an Instron 4302 tensile testinginstrument.

3.3. Visual analysisThe fouled membrane was carefully inspected

using a light microscope (Leica DMR ResearchMicroscope), equipped with a digital camera.Direct visual observation of the membrane sur-face and membrane element was also conducted.High-resolution and high-magnification imagesof the membrane surface were taken using anHitachi 4500 II field environment electron micro-scope (FESEM).

The FESEM was equipped with an Oxford Isisenergy depressive X-ray analyzer (EDX) thatallows semi-quantitative analysis of the elementalcomposition of the foulant layer to be performed.The excitation energy was set at 20 kV. Mem-brane samples were coated with carbon prior toFESEM analysis. A virgin (clean membrane)

L.D. Nghiem, A.I. Schäfer / Desalination 188 (2006) 113–121 117

sample was also inspected for comparisonpurposes.

3.4. Foulant extraction

Fifteen fibres with a length of 35 cm wereobtained from each membrane element for foulantextraction. They were cut into small pieces,soaked in 50 mL of 0.8 M HNO3 or 0.1 M NaOHand then ultrasonicated for 180 min in a typicallaboratory sonication bath (Sonicare, SA, Aus-tralia). The extracted solution was neutralisedusing HNO3 when a caustic extracting solutionwas used. All reagents used in this study were ofanalytical grade. Standard inorganic elements ofthe extracted solution were analysed using aVarian ICP-OES with a simultaneous wavelengthscanning and total organic content of the solutionwas analysed using a Shimadzu TOC-VCSHanalyser.

4. Results and discussion4.1. Mechanical strength

Tensile strength is a relatively new parameterinvestigated in autopsy studies. It presents themechanical strength of the membrane fibre, andhence is directly related to material properties ofthe membrane. The impact of fouling on tensilestrength is to date very little understood. It can beanticipated that modification of the membranepolymer will have an impact on the membranetensile strength, and it is also a general indicatorof the “tiredness” of the material after extensivecleaning and backwashing. The fouled mem-branes may become more brittle and hence moresusceptible to integrity problems or the polymermay become stretched and the actual pore sizechange. Elongation at the break point of fivevirgin and five fouled membrane samples ispresented in Fig. 5.

In this case, it appears that the fouled mem-brane has become more brittle, possibly because

Fig. 5. Elongation at break point of the fouled membranefibres compared to virgin membrane fibres.

the membrane has been allowed to dry prior totesting. This may also be attributed to normalwear and tear during the filtration process andbackflush operation. A significant variation inelongation at break point can also be observed forthe fouled membrane. However, it is interestingto note that both fouled and virgin membraneshave approximately the same young modulus,yield point and elastic profile (data not shown).No clear evidence of material degradation couldbe observed for the membranes investigated.

4.2. Visual observation

A picture of the virgin as well as the fouledmembranes is shown in Fig. 6. Visual investi-gation reveals a strong discolouration of thefouled membranes with a reddish fouling layer.The fouled membrane also demonstrates a severeexpansion of the fibres, presumably due to themechanical stress of filtration and backflushing.Some fragile and brittle characteristics upontouching, in particular near the end of themodules where the fibres were attached to theplotting resin, can also be observed. This isconsistent with the results obtained from mechan-ical testing described above.

The fouling layer morphology was furtherexamined in detail and an electron microscopicpicture of the fouling layer is shown in Fig. 7. An

L.D. Nghiem, A.I. Schäfer / Desalination 188 (2006) 113–121118

Fig. 6. Light microscopic pictures of a virgin membrane (left) and fouled membrane (right). Bar = 150 µm.

Fig. 7. Electron microscopic pictures of a virgin membrane (left) and fouled membrane (right) taken by the FESEM.

electron microscopic picture of the clean mem-brane surface is also presented for comparisonpurposes. Schäfer et al. [1] methodically sum-marised four distinctive common fouling categor-ies namely scaling/inorganic, organic, colloidal/particulate, and biofouling. There is no evidenceof bacteria or fungi on the fouling layer andbiofouling is unlikely the main cause in thisinstance. Fig. 7 reveals a firm cake layer on themembrane surface. This cake layer is probablymade up of organic matter or colloidal particlesor, most likely, the combination of both. Whilescaling is uncommon in MF processes and noevidence of crystallisation can be observed on the

fouling layer, it is possible that multivalentcations, for example Ca2+, can act to cement col-loidal particles, organic matter, and the membranesurface one another as previously shown [10].This phenomenon has also been exquisitelyexamined by Li and Elimelech by measuring theattractive force between these entities with andwithout the presence of multivalent cations [11].

It is noteworthy that the reddish colour of themembrane sample may reflect the presence of ionoxide. Clays together with silica colloids aretypical particulate matter encountered in exhibitand road hose-down run-offs, which make upmost of the influent in this case, in particular

L.D. Nghiem, A.I. Schäfer / Desalination 188 (2006) 113–121 119

Fig. 8. Energy dispersive X-ray spectrometry of the virgin membrane surface (top) and the foulant layer (bottom).Excitation energy was set at 20 kV (note difference in y-axis scale).

during the construction phase. This hypothesis canbe confirmed by examining the elementary com-position of the fouling layer and foulant charac-terisation as shown in the following section.

4.3. Major foulant identification and foulantcomposition

The presence of metallic elements such as ion,aluminium, calcium, and magnesium is clearlyevident in the EDX spectrometry graph of thefouling layer, as shown in Fig. 8. A significantamount of silica can also be observed. Further-more, Fig. 8 reveals a large peak of oxygen and

trace amount of other non-metallic elements suchas chlorine, sulphur, and phosphate. Such a signi-ficant amount of oxygen may be attributed toorganic matter present in the fouling layer,although it is prudent to note that it can also befrom the membrane polymer. While a small peakof oxygen can be observed in the EDX spec-trometry graph of the virgin membrane, nochlorine, sulphur, or phosphate can be found (seeFig. 8).

Foulants were extracted from the fouling layerin accordance to the extraction protocol describedabove. Both nitric acid and sodium hydroxidewere used as extracting solutions. Acidic solution

L.D. Nghiem, A.I. Schäfer / Desalination 188 (2006) 113–121120

Table 2Foulant quantity and composition extracted from themembrane by sonicating the membrane in 0.8 M HNO3 or0.1 M NaOH solution for 180 min

Concentration in the extractedsolution (mg/L)

Extracting solution

Foulant 0.8 M HNO3 0.1 M NaOHAl 12.9 2.4Ca 34.3 3.5Fe 97.2 2.0K 1.5 1.4Mg 4.2 0.6Mn 3.0 0.1Na 18.7 NAa

TOC 32.4 22.6

aNot available, as sodium hydroxyl was used as theextracting solution.

is typically used to extract inorganic foulantswhile caustic solution is commonly used toremove organic foulants [7]. Sonication was alsoapplied to ensure that complete extraction couldbe achieved. Results of the extraction study arepresented in Table 2.

Despite the fact that sonication was applied,caustic solution failed to extract foulants from thefouling layer (see Table 2). This is because thefouling layer, in this case, consists predominantlyof inorganics and precipitation of metal hydroxylsprevents the extraction process from occurring.The amount of organic matter extracted by thecaustic solution is also less than that by the nitricacid solution. As previously discussed, metalswith high valency can aggregate or complex withorganic matter and hinder them from beingextracted. This is consistent with visual obser-vation that a reddish colour of the membranepersists after the extraction process with thecaustic solution. In contrast, it appears that acidicsolution can completely extract the foulants in

this case. The amount of extracted metallicelements is in excellent agreement with EDXanalysis. As can be seen in Table 2, ion, calcium,and aluminium are predominant in the foulinglayer. A significant amount of organic mattermeasured by total organic carbon can also beobserved.

4.4. Implications and possible remediationstrategies

Mechanical testing indicates that no materialdegradation of the membrane due to exposure tooxidants or any other chemical agents hasoccurred, although the fouled membrane fibresappear to be slightly more brittle. Visual obser-vation by light microscopy and FESEM com-bined with the results from foulant compositionanalysis consistently reveal that the membranewas fouled with a mixture of colloidal particlesand organic matter. Complexation and otherinteractions between multivalent cations, organicmatter, and colloidal particles have most likelyworsened the fouling process caused by thesuspended solids load itself. It is advisable toenhance pre-treatment prior to the MF unit,possibly by applying coagulants, to maintain alow and stable turbidity level in the feed. Achemical cleaning protocol and air back-flushingfrequency may also need to be modified for anoptimal fouling mitigation approach. Monitoringof the feedwater composition over a period oftime will provide further insight into the foulingprocess.

5. Conclusions

This study demonstrates that the membraneautopsy protocol presented here can be rigorouslyused to identify the cause of membrane fouling inhollow-fibre MF applications. Necessary mea-sures to mitigate fouling can then be suggestedfor further examination. In this particular in-

L.D. Nghiem, A.I. Schäfer / Desalination 188 (2006) 113–121 121

stance, it appears that a combination of colloidalparticles and organic matter enhanced by multi-valent cations is the main cause of membranefouling. It is expected that membrane fouling willstrongly depend on feed water constituent com-position and chemistry and also on the operationcondition of the water recycling plants. In allcases, this membrane autopsy protocol will serveas a unique and valuable tool to support waterrecycling applications and trouble-shooting.Further work is in progress to investigate the in-depth fouling mechanisms and enhance autopsyapproaches.

Acknowledgements

Long Nghiem would like to thank the Aus-tralian Institute of Nuclear Science and Engineer-ing (AINSE) for a PhD top-up scholarship.Joanne George is acknowledged for extensivetechnical and analytical support in the Environ-mental Engineering Laboratory at the Universityof Wollongong as well as Greg Tillman andDr Konstantinov in Materials Engineering andISEM for the support with microscopy and ICPanalysis, respectively.

This project was also partly supported by theInternational Science Linkages programme estab-lished under the Australian Government’s inno-vation statement, Backing Australia’s Ability,funded by the Commonwealth Department ofEducation, Science and Training for the projectOzAquarec: Integrated Concepts for Reuse ofUpgraded Wastewater in Australia (CG030025).

Daryl Edwards from Taronga Zoo is thankedfor provision of a fouled membrane module,water samples and extensive technical infor-mation on the scheme.

References

[1] A. Schäfer, N. Andritsos, A.J. Karabelas, E.M.V.Hoek, R. Scheider and M. Nyström, Fouling innanofiltration, in: Nanofiltration — Principles andApplications, A. Schäfer, D. Waite and A. Fane, eds.,Elsevier, 2004, pp. 169–239.

[2] D.L. Nghiem and A. Schäfer, in prep.[3] F.H. Butt, F. Rahman and U. Baduruthamal, Charac-

terization of foulants by autopsy of RO desalinationmembranes. Desalination, 114 (1997) 51–64.

[4] T. Darton, U. Annunziata, F. del Vigo Pisano andS. Gallego, Membrane autopsy helps to providesolutions to operational problems. Desalination, 167(2004) 239–245.

[5] L.Y. Dudley and E.G. Darton, Membrane autopsy —a case study. Desalination, 105 (1996) 135–141.

[6] A.M. Farooque, A. Hassan, and A. Al-Amoudi,Autopsy and characterisation of NF Membranes afterlong-term operation in a NF–SWRO pilot plant, IDAWorld Congress on Desalination and Water Reuse.San Diego, CA, 1999.

[7] E.-m. Gwon, M.-j. Yu, H.-k. Oh and Y.-h. Ylee,Fouling characteristics of NF and RO operated forremoval of dissolved matter from groundwater.Water Res., 37 (2003) 2989–2997.

[8] J.C. Radcliffe, Water recycling in Australia. Aus-tralian Academy of Technological Science andEngineering, 2004.

[9] Environment Technology Case Studies Directory:Wastewater and Stormwater Treatment Process for aZoo. http://www.eidn.com.au/taronga.html. AccessedJan. 2005.

[10] N. Oschmann, D.L. Nghiem and A. Schäfer, Foulingmechanisms of submerged ultrafiltration membranesin greywater recycling. Desalination, 179 (2005)215–223.

[11] Q. Li and M. Elimelech, Organic fouling andchemical cleaning of nanofiltration membranes:measurements and mechanisms. Environ. Sci.Technol., 38 (2004) 4683–4693.