investigation ofmembrane de-clogging techniques ... · investigation ofmembrane de-clogging...

Investigation of Membrane De-cloggingTechniques in the Submerged Membrane

Filtration Adsorption Hybrid System (SMFAHS)

Paul James Smith,' Saravanamuth Vigneswaran.!" Huu Hao Ngo,' Roger Ben-Aim/ Hung Nguyen,''University of Technology, Sydney, PO Box 123 Broadway NSW 2001 Australia

21nstitut Nat. Sciences Appliquees, Complexe Scientifique de Rangueil, 31077 Toulouse Cedex 4, France*Corresponding Author. Tel: +61-02-9514-2641; Fax: +61-02-9514-2633;

Email: s.vigneswararujauts.edu.au

Membrane clogging is a major obstacle to thesuccessful operation of the membrane separation process. Asubmerged hollow fibre membrane with powdered activatedcarbon (PAC) adsorption (adsorption-membrane hybridsystem) was used for the removal of organics from asynthetic wastewater representative of biologically treatedsewage e.fJluent. PAC usage successfully adsorbs themajority of the organics, and then the organic laden PAC isseparated by the membrane reducing the direct organicloading to the membrane. However, membrane cloggingstill occurs.

This study involved the development of anautomation system and Supervisory Control and DataAcquisition (SCADA) system for performing aninvestigation and evaluation of three automated de-clogging techniques.

The first de-clogging method involved the use ofperiodic relaxation, whereby permeate production for 12minutes was periodically stopped for 3 minutes and theshear forces created by the aeration system and the absenceof suction pressure during the relaxation period were usedto de-clog the membrane. The second de-clogging methodinvolved the use of a series of periodic back flushexperiments with varied frequencies and durations toforcepermeate in the opposite direction out through themembrane pores. The optimal results in terms of de-clogging the membrane were achieved using a 15 secondbackflush after 15 minutes of permeate production. Thethird de-clogging method involved the application of anunderstanding of results of the periodic back flush series ofexperiments to design an automation system with a newapproach to backflushing where an upper limit of atransmembrane pressure (TMP) increase each cycle wasused to initiate the backflush. The transmembrane pressurerepresents the pressure measured across the membrane andit is a vital parameter indicating the degree offouling of themembrane.

A periodic backflush wasfound to be significantlymore effective in terms of increasing the total quantity ofwastewater treated than was achieved using periodicrelaxation and was investigated in detail during the study.

FluidlParticle Separation Journal, Vol. 16, No.2, 2004

For the periodic backflush, an optimal frequency andduration was determined for treatment of wastewater with afixed foulant concentration. The new approach tobackflushing using more advances in the control systemincorporating the TMP increase each cycle resulted in a40% reduction in the number of backwashes required andwas capable of self-optimising operating parameters underan unsteadyfoulant concentration of wastewater.

Fluid/Particle Separation Journal, Vol.16, No.2, 2004,165 -173

KEYWORDS

Adsorption, hollow fibre membrane, powdered activatedcarbon (PAC), Programmable Logic Controller (PLC),Supervisory Control and Data Acquisition (SCADA),Submerged Membrane Filtration Adsorption HybridSystem (SMF AHS), transmembrane pressure (TMP)

INTRODUCTION

As a consequence of increasingly stringentstandards for wastewater disposal and reuse, various newmembrane technologies have emerged. Membraneprocesses such as reverse osmosis and nano-filtration canremove most of the pollutants, including dissolvedorganics, but their operational costs are high because ofhigh energy requirements and membrane clogging.Microfiltration and ultra-filtration require low energy andthus are cost effective options. However, they still cannotremove dissolved organic matter due to their relativelylarger pore sizes. In a number of water reuse applications,mainly in the late 1990s, ultrafiltration or micro filtrationwere combined with biologic processes in order to removethe dissolved organic matter (Vigneswaran et aI., 2003).When coupled with a biological process, microfiltration orultrafiltration enable removal of biodegradable dissolvedorganic matter. Microfiltration used in these hybrid systemscan be placed either in the external loop or in submergedconfiguration. Microfiltration used in the external loopoperates at high cross flow velocity which requires a highamount of energy. To reduce the energy requirement,

165

membrane clogging and to simplify the process, a reactortank with a submerged membrane is used.

The Submerged Membrane Filtration AdsorptionHybrid System (SMF AHS) is emerging as a highlypromising water and wastewater treatment technology(Vigneswaran et al., 1991; Seo et al., 1997; Snoeyink et al.,2000; Kim et aI., 2001; and Matsui et al., 2001 a, 200 1b)that combines addition of powdered activated carbon (PAC)in the reactor tank with the membrane system. In thisprocess, the pollutants (particularly the dissolved organicmatter) are first adsorbed onto the PAC, greatly reducingthe direct loading of dissolved organic pollutants onto themembrane to reduce clogging and significantly enhance theremoval efficiency of the process.

Although PAC usage minimises membraneclogging as the majority of the organic matter present isadsorbed onto the PAC particles rather than the membrane,clogging still occurs during long term operation. Thisresults in increased operating transmembrane pressure(TMP), decreased permeate flux or both and eventuallyleads to the membrane process being stopped for intensivechemical cleaning of the membrane. Simple and effectivede-clogging methods can further minimise the membraneclogging.

This research investigates the use of three de-clogging techniques in reducing long-term membraneclogging to maximize the treated water capacity of themembrane system. The first de-clogging method involvedthe use of periodic relaxation, whereby permeate productionwas periodically stopped and the shear forces created by theaeration system and the absence suction pressure during therelaxation period were used to de-clog the membrane. Thesecond de-clogging method involved the use of a series ofperiodic back flushes, whereby permeate was forced in theopposite direction out through the membrane pores. Thethird de-clogging method involved the application of theresults from the periodic back flush series of experiments todesign an automation system where an upper limit of aTMP increase each cycle was used to initiate the backflush.

THEORETICAL CONCEPTS

The results obtained for different cleaningconditions were compared in terms of productivity.

1. Periodic Relaxation

The membrane de-clogging technique of periodicrelaxation has the advantage of not using any permeate forbackflushing. However, it does not produce any permeateduring the period of relaxation. The theoretical productivityof the SMF AHS operating with periodic relaxation can becalculated as follows:

PFj itialxTp d 'PR (mL/min) = __ n_IIl T_O_u_cln....::ge-TProducing+ TRelaxing


where: P R = Theoretical productivity of MF AHS

operating with periodic relaxation without clogging

PFlnitial = Initial permeate flux of SMF AHS

TProducing= Period of permeate production cycle

TRelaxing= Period of relaxation cycle

The permeate produced when the SMF AHSoperates in periodic relaxation mode is calculated with acontinuous integral of the productivity. Practically, there isprogressive membrane clogging which results in areduction of permeate flux and an increase in TMP duringthe membrane operation. The cumulative flux can thus becalculated as follows:

I PFxT 'PPR(mL)= f ProdUCing.dt (2)

o TProducing+ TRelaxing

where: PP R = Cumulative permeate produced with

periodic relaxation modePF = Real-time permeate flux of SMF AHS

during operation (varies with clogging)


TRelaxing= Period of relaxation cycle

Another important measure of productivityincorporates losses due to both the periodic relaxation de-clogging technique and the clogging of the system overtime.

For the periodic relaxation operation mode, this iscalculated as follows:

(3)

where: PR R = Productivity (%) of the SMF AHS

operating with periodic relaxation (expressed as apercentage of the productivity of the system producingpermeate continuously with no relaxation period or fouling)

P R = Productivity of SMF AHS operating withperiodic relaxation (during a given cycle)

PFlnilial = Initial permeate flux of SMF AHS

before clogging

2. Periodic backflush

(1)

The membrane de-clogging technique of periodicbackflushing has the disadvantage of using permeate for thebackflush and not producing any permeate during thebackflush. The theoretical productivity of the SMF AHSoperating using a periodic backflush can be calculated asfollows:

PFI "lxTp d ' -BR] , 'a!xTB k h;~PBw(mL/min)= mIla ro uClng mil ac wasmrg (4)TProducing+ TBackwashirg

166

where: P BW = Theoretical productivity of SMF AHS

operating with periodic backt1ush without clogging

PFlnirial = Initial permeate flux of SMF AHS

BFlnitial = Initial backflush flux of SMF AHS


TBackwashing= Period of backflushing cycle

In a practical situation, the permeate flux, howeverdecreases with time as the membrane becomes clogged.Considering the decline in permeate flux with time, theproductivity during the membrane operation with theperiodic backflush can be calculated as follows:

I PFxT . -BFxT .PPBW(mL) = f ProduclOg Backwashing• dt (5)

o TProducing+ TBackwashing

where: PP BW = Cumulative permeate produced with

periodic backt1ushPF = Real-time permeate flux ofSMFAHS

during operation (varies with clogging)BF = Backflush water flux ofSMFAHS


TBackwashing= Period of backflushing cycle

For the periodic backflush operation mode, theproductivity incorporating losses due to both the periodicbackflush de-clogging technique and the clogging of thesystem over time is calculated as follows:

PRBW(%) = PBW xIOO(PFlnirial)

~--j PLe

Feed tank

Air compressor Air diffuser

where: PR BW = Productivity (%) of the SMFAHS

operating with a periodic backwash (expressed as apercentage of the productivity of the system producingpermeate continuously with no backwash or fouling)

P BW = Productivity of SMF AHS operating with

periodic backflush (during a given cycle)

PFlnitial = Initial permeate flux of SMF AHS

before clogging

Even during the operation of the SMF AHS witheither periodic backflush or periodic relaxation, themembrane becomes clogged and the permeate flux beginsto decline leading to a reduction in productivity from thetheoretical value. The reduction in productivity can beminimised by optimising the de-clogging conditions thatallow minimal membrane clogging. The membraneclogging is indicated by the continuous TMP dropmonitored by the system.

EXPERIMENTAL

(6)

The schematic diagram of the SMF AHSexperimental set-up is shown in Figure 1.

Wastewater was pumped into the reactor tankusing a feed pump and reed switches to control the leveland volume in the tank. A predetermined amount of PACwas initially added into the tank to adsorb the dissolvedorganic substances, which was subsequently separated bythe membrane filtration imposed by the suction pump. Nofurther addition of PAC was made during the experiment. Apressure gauge was used to measure the transmembranepressure of the hybrid system. An air diffuser was used tomaintain the PAC in suspension.

Permeate tank

Sludge drain

FP: Feed Pump,SP: Suction pump,BP: Backflush Pump,PS: Pressure Sensor,~ : Solenoid Valve~..- _ _ _ _ _ _ _ ...•..._ .._ :

Figure 1. Experimental set-up of the SMFAHS.

Fluid/Particle Separation Journal, Vol. 16, No.2, 2004 167

This research involved the development of anautomation system and a graphical operator interface whichenabled backflush optimization, productivity maximizationand accurate recording of project data for the SMFAHS.The automation system was developed through theprogramming and commissioning of a ProgrammableController. The operator interface was based on aSupervisorary Control and Data Acquisition (SCADA)system that was programmed to communicate directly withthe Programmable Controller.

Using the automation system, periodicbackflushing and periodic relaxation methods wereemployed to de-clog the membrane system. The relativemerits of these systems were studied in terms of achievedproductivity. Periodic backflushing involved reversing thedirection of permeate flow at twice the rate used during themembrane filtration to force permeate out through themembrane and remove foulants from the membrane surfaceand pores. Periodic relaxation involved periodicallystopping permeate production, allowing the shear forces of

the aeration diffuser to remove foulants from the membranesurface.

Initially, periodic relaxation was investigated. Thesecond series of experiments involved varying thebackflush frequencies and durations. For all experiments,the ratio of backflushing frequency to backf1ushing durationwas maintained at 60: I. Finally, a more advanced backf1ushwas trialed with the use of knowledge of the TMP dropincrease attained from the periodic backflush study as aninitiator for the backflush cycle.

Table 1 shows the constituents of the syntheticwastewater used in this study. The synthetic wastewaterconsists of 12 organic and inorganic with similar propertiesto biologically treated wastewater.

Table 2 shows the physical and chemicalproperties of the hollow fiber membrane used in this study.

Table 3 shows the characteristics of the PowderActivated Carbon (PAC) used in this study.

Table 1. Constituents of the synthetic wastewater used.

COMPOUNDS Weight (mg/L) COMPOUNDS Weight (mz/L)Beef extract 1.8 Sodium laurvle sulphate 0.94Peptone 2.7 Acacia gum powder 4.7Humic acid 4.2 Arabic acid (po Ivsaccharide) 5Tannic acid 4.2 (NH4)zS04 7.1Sodium lignin sulfonate 2.4 KzHP04 7NH4HC03 19.8 MgS043HzO 0.71

Table 2. Physical and chemical properties of the membrane.

Properties Hollow fibre membraneTotal surface area (rrr'), (320 fibres with 12cm length) 0.05Pore size (urn) 0.1Material PolyethyleneInner diameter (rnm) 0.27Outer diameter (mm) 0.41

Table 3. Characteristics of the PAC.

Specification PAC-WBIodine number (mg!g min) 900Ash content (%) 6 max.Moisture content (%) 5 max.Bulk density (kg/rrr' 290-390Surface area (mz!g) 882Nominal size 80% min finer than 75 micronType Wood basedMean pore diameter (A) 30.61Micronore volume (cc/z) 0.34Mean diameter (urn) 19.71Product code MD3545WB powderAdsorption capacity g'TOcJg'PAC\ 0.00567*

FluidlParticle Separation Journal, Vol. 16, No.2, 2004 168

RESUL TS AND DISCUSSION

1. Time based periodic backflushing

The effect of backtlush frequency and backtlushduration was studied. Table 4 shows the parameters used ineach of the experiments.

Although the periodic backflush reduced the TMPincrease rate and thus the membrane clogging, it could notbe completely eliminated. Figure 2 shows the results of thefive experiments outlined in Table 4.

Experiment Number Backtlush Duration (sec)

Table 4. Backflushing parameters used in the experiments.

Backflush Flux (mL/min)I2345

5153060120

45 ~

403530

l25=.c. 20:eI-

151050

0

Backflush Frequency (min)5153060120

8080808080

'" 5Min5Sec Pressure Drop at 40mUm (kPa)

.• 15Min15Sec Pressure Drop at 40mUm (kPa)

• 30Min30Sec Pressure Drop at 40mLlm (kPa)

c 1Hr1 MinPressure Drop at 40mLlm (kPa)

x 2Hr2Min Pressure Drop at 40mLlm (kPa)

100 200 300 400 500 600Filtration Time (min)

Figure 2. TMP profiles with operation time for different backflushing conditions.(Initial Permeate Flux = 40 mLlmin)

Results from the experiment indicate that abackflush duration of 2 minutes at a frequency of 2 hourswas insufficient to control the increase of TMP. Thebackflush duration of 5 seconds was also found to beinsufficient even though the membrane was backflushedevery 5 minutes. The TMP increase was minimal when themembrane was backflushed for 15 seconds every 15minutes of filtration time.

From a model developed for the relationshipbetween permeate flux and TMP, the resulting cumulativepermeate production and percentage productivity data withtime are shown in Figures 3 and 4.

Figure 3 shows the relationship between permeateproduction and time. The theoretical maximum productivity

Fluid/Particle Separation Journal, Vol. 16, No.2, 2004

curve shows the achievable amount of permeate productionwith losses only from the backflushing and not fromclogging. In real operating conditions, the membrane beginsto clog, resulting in a lower permeate flux and deviationfrom the theoretical productivity curve. The amount ofdeviation from the theoretical productivity curve isinversely related to the TMP of the membrane which iscaused by clogging. By selecting optimal backflushingparameters, this clogging can be minimised allowing thesystem to operate closer to the theoretical productivity. Aswell as the productivity improvements attainable usingoptimal backflushing parameters, the membrane can have alonger life.

Figure 4 shows the deviation of the productivityfor the various backflushing parameters.

169

If clogging was totally alleviated, the system couldoperate continuously at 95.1 % productivity (with a 4.9%reduction due to losses involved with backflushing).However, membrane clogging cannot be avoided and theproductivity diminishes. By choosing the optimal backflushconditions, the reduction in productivity can be minimised.

From the preliminary experimental investigation,the optimal backt1ushing condition was found to be a 15second backt1ush after every IS minutes of filtration. Thepressure drop during each cycle was approximately 1.5kPa.

Thus a detailed experiment was conducted using apredetermined value of TMP as the controlling parameter(i.e. as soon as the TMP dropped by UkPa during eachfiltration cycle, the backflush was started). After abackflush cycle, the permeate pump operates and the TMPbegins to stabilize. At this point, the control system logs aTMP reading. The TMP is then recorded and continuouslycompared to a value 1.5kPa below this recorded TMPreading. Once this value is reached, the backflush isinitiated.

25• 5minSsec Productivity (L)

0• 15min15sec Productivity (L)• 30min30sec Productivity (L) 0 •20 • XX 1Hr1 min Productivity (L) 0 X

d: ~ •x 2Hr2min Productivity (L) 0 •'tl XCIl o Theoretical Maximum Productivity (L) 0 tu 15 t:::l X'tle ~ tc..

~~.s

ra 10CIl ~E..Ii!CIlc..

:lQ

Q

Q

0 100 200 300 400 500 600Filtration Time (min)

Figure 3 Cumulative permeate production versus filtration time.

100

95 X

iI a *I lk J\• • * .\0 • .\

g 90 - • i .\0 • .\

.\~ 0 ~'> 85 ~ ~ti 0 •:::l 0'tl Ie - 5min5sec Productivity (%) 0c.. 80

.\15min15sec Productivity(%) 0a

75 • 30min30sec Productivity (%)Q

X 1Hr1min Productivity (%)

o 2Hr2min Productivity (%)70

0 100 200 300 400 500 600Filtration Time (min)

Figure 4. Productivity (%) versus Time.

FluidlParticle Separation Journal, Vol. 16, No.2, 2004 170

2. Comparison of the three methods

A series of 20 hour experiments were thenperformed with different modes of cleaning (the optimaltimed periodic backflush condition, the TMP based pressureinitiation backflush and periodic relaxation).

Over a 20-hour period, TMP based pressureinitiation backflush method resulted in a similar long-termpressure drop to the optimal timed periodic backflush, butrequired less frequent backflushes leading to a furtherimprovement in productivity. This also has the benefit ofautomatically finding the optimal backflushing parametersfor wastewater varying in influent concentration. Thebackflush was found to be much more successful thanrelaxation in terms of productivity and clogging.

Figure 5 shows the results of TMP for the optimalfixed time backflush, periodic relaxation and the TMPbased pressure initiation backflush with time.Although there can be no theoretical measure of themaximum productivity from using a TMP based pressureinitiation backflush, it resulted in a 40% reduction in thenumber of backflushes required and more permeateproduction. The success of these results is due to thebackflush occurring when required (TMP drop = 1.5 kPa)rather than at a fixed time when it may not be required.

Figure 6 shows the deviation of the productivityfor the both the timed backflush and the periodic relaxationmethods of cleaning.

Figure 5. TMP profiles versus time for different de-logging methods.

100

95

90

~ 85~.2: 80ti::l'Ce :: •Q.

65 I

60 l0

• 12minProduction3minReiaxation Productivity ('!o)

• 15minProduction15secBackwash Productivity ('!o)

. .• • • • • • • • • • • • • • • • • • •

2 4 6 8 10 12 14 16 18 20

Figure 6 Productivity (%) versus operation time.

Filtration Time (hr)

FluidlParticie Separation Journal, Vol. 16, No.2, 2004 171

As can be seen in Figure 6, the theoreticalproductivity for timed back flushing is 95.1% and forperiodic relaxation it is 80%. During operation, the actualproductivity for periodic relaxation decreased at a fasterrate than that of periodic backflushing. This indicates thatattempting to operate the system using periodic relaxationwith an increase in the theoretical productivity througheither increasing the permeate production period or

decreasing the relaxation period would be unsustainable asclogging would erode any productivity gains.

Figure 7 shows the cumulative permeateproduction verse TMP drop for the various de-cloggingconditions.

16 • •• • •• • • •14 • ••• •12 11 I. I. I.D" D" D"li D. D D D D

D" D ••I. I. I. D

D I. D D D Da. I. I. I. D:. 10 •D I. I.

Co • I.

e 8 •••C Da. 6 •:::i: • 12minProduction3minReiaxation Productivity (L)l- I.

4D 15minProduction15secBackwash Productivity (L)

2I. TMP based backwash initiation Productivity (L)

00 5 10 15 20 25 30 35 40

Permeate Produced (L)

Figure 7 Cumulative permeate production versus TMP Drop for different de-clogging conditions.

Severe membrane clogging is indicated by a TMPdrop of over 50 kPa and once reached requires themembrane to undergo an intensive physical and chemicalclean. These figures indicate that optimized cleaningmethods and parameters can effectively increase the totalquantity of wastewater treated before the membranerequires an intensive physical and chemical clean.

CONCLUSIONS

• Periodic backflushing was found to effectivelyreduce reversible clogging on the membrane surfaceallowing a high permeate flux rate to be maintained;• An optimal frequency for the periodic backflushingexists for this fixed foul ant concentration of wastewater;• The optimal frequency for the periodic backflushingcannot be fixed in case of unsteady foulant concentrationsof wastewater;• A periodic backflush was significantly more effectivein terms of increasing the total quantity of wastewatertreated than was achieved using periodic relaxation;• The use of the pressure based backflush systemenables the SMF AHS to operate effectively underunsteady concentrations of wastewater as a backflush isonly initiated when required rather than at fixed timeintervals for a periodic backflush system; and


• The productivity of the SMF AHS is comprised of aninitial theoretical limit of productivity that begins todecrease as clogging occurs.

ACKNOWLEDGEMENTS

This research was funded by the Australian ResearchCouncil discovery grant on the Submerged Membrane-Adsorption Hybrid System. The first author would like tothank the Institute for Water and Environmental ResourceManagement for his scholarship.

REFERENCES

Vigneswaran, S., Chaudhary, D. S., Ngo, H. H., Shim, W.G., Moon, H., "Application of PAC membrane hybridsystem for removal of organics from secondary sewageeffluent: experiments and modeling", Separation Scienceand Technology, vol: 38,10,2003, pp. 2183-2199

Vigneswaran, S., Vigneswaran, B., and Ben Aim, R.,"Application of micro filtration for water and wastewatertreatment", Environmental Sanitation Reviews, no. 31, June1991

Seo, G. T., Ohgaki, S., and Suzuki, Y., "Sorptioncharacteristics of biological powdered activated carbon in

172

BPAC-MF (Biological activated carbon-microfiltration)system for refractory organic removal", Water Science andTechnology, vol: 35, 7,1997, pp. 163-170

Snoeyink, V. L., Campos, c., and Marinas, B. J., "Designand performance of powdered activatedcarbon/ultrafiltration systems", Water Science andTechnology, vol: 42, 12,2000, pp. 1-10

Kim, H. S., Katayama, H., Takizawa, S., and Ohgaki, S.,"Removal of coliphage QP and organic matter fromsynthetic secondary effluent by powdered activated carbon-


microfiltration (PAC-MF) process", Proceedings of IWASpecialised Conference on Membrane Technology, Israel,2001, pp. 211-219

Matsui, Y., Colas, F., and Yuasa, A., "Removal of syntheticorganic chemical by PAC-UF systems - 2: ModelApplication", Water Research, vol: 35, 2, 2001a, pp. 464-470

Matsui, Y., Yuasa, A., and Ariga, K. "Removal of syntheticorganic chemical by PAC-UF systems - I: Theory andmodelling", Water Research, vol: 35, 2, 2001b, pp. 455-463

173

investigation ofmembrane de-clogging techniques ... · investigation ofmembrane de-clogging...

Documents