Fouling of membrane bioreactors during treatment of produced water

5 downloads 0 Views 123KB Size Report
Fouling of membrane bioreactors during treatment of produced water. A. Brookes. 1,a# .... Malvern Mastersizer 2000 (Malvern Instrument, UK). Viscosity was ...
1

Fouling of membrane bioreactors during treatment of produced water A. Brookes1,a#, B. Jefferson1,b, P. Le Clech1,c, S. J. Judd1,d* 1

School of Water Sciences ,Cranfield University, Cranfield, Bedfordshire, MK43 0AL, UK. a Tel. +44 (0) 1234 750111; Fax +44 (0) 1234 751671; email: [email protected] b Tel. +44 (0) 1234 750111; Fax +44 (0) 1234 751671; email: [email protected] c Tel. +44 (0) 1234 750111; Fax + 44 (0) 1234 751671; email: [email protected] d Tel. +44 (0) 1234 754842; Fax +44 (0) 1234 751671; email: [email protected]

Abstract The potential of membrane bioreactor technology for treatment of oil and gas field produced water has been assessed. A pilot scale immersed membrane bioreactor with tubular membranes treating a synthetic analogue wastewater was operated under steady state conditions. Membrane fouling is examined by both critical flux determination and longer term trials . Liquid phase analysis has included high performance size exclusion chromatography (HPSEC) to derive molecular weight grade efficiencies between each stage of the process. Gas phase monitoring of key components in the feedwater including BTEX compounds (Benzene, toluene, ethylbenzene and xylene) has been undertaken. Overall the link between the bio transformation of organics from feed to soluble microbial products has been identified in relation to membrane fouling to provide a framework for optimising MBR operation.

1. Introduction Produced water may be defined as the wastewater that is brought to the surface during production of oil and gas. This water tends to be of varying salinity and contain dispersed and dissolved organics such as BTEX compounds (Benzene, toluene, ethylbenzene and xylene) and polycyclic aromatic hydrocarbons. Current UK legislation requires water discharged from offshore platforms to contain no more than 40mg.L-1 of oil. A voluntary reduction to 30mg.L-1 is expected to be introduced and over a number of years legislation is likely to become tighter. In the long term, there is likely to be a requirement for the removal of the dissolved oil component and reduction of aromatic compounds, something that cannot be achieved by conventional intensive physical separation technologies such as hydrocyclones and centrifuges. Membrane bioreactors (MBRs) offer intensive, low- footprint, low-sludge production biological treatment of waters with high organic loads, and produce water of high quality commensurate with the coupling of a permselective membrane, of pore size typically in the 0.1 to 1 µm range, with a conventional bioreactor [1]. As such it is a candidate technology for the removal of dissolved components from oil- laden waters. This paper investigates the propensity to fouling of high strength industrial wastewater by critical flux determination using the flux step method and long term operation. Analysis of components from feedwater through bioreactor to permeate, are made in an attempt to understand the fate of organics through the process. 2. Materials and methods 2.1 Pilot scale rig A submerged membrane bioreactor, 40L in volume, was operated under steady-state ___ *Corresponding author # Presenting author

2 conditions. The vertically mounted tubular membrane used was supplied by Milleniumpore (Sunderland, UK), and had a total area of 0.2 m2 , rated pore size of 0.1µm and in to out flow. Flow velocity was 0.2 m.s-1 . A single module has dimensions of 60mm diameter and 0.9m length was made up of twelve 8mm bore lumen. A synthetic produced water analogue was employed with a target COD of 1575 ± 100 mg.L-1 , comprising: methanol; ethylene glycol; BTEX compounds (Benzene, toluene, ethylbenzene and xylene); polycyclic aromatic hydrocarbons (acenaphthene, biphenyl, phenanthrene); naphthalene; phenol; carboxylic acids (acetic acid, proprionic acid, valeric acid) and sodium chloride (1%). Nutrients in the form of urea and orthophosphate required to support micro-organism growth were also dosed into the feedwater. All critical flux tests and long-term fouling trials were carried out following membrane cleaning with 0.5% Ultrasil 75 at 50o C for 24 hours. Tests on real sewage relate to an influent of settled sewage with mean COD of 200 mg.L-1 and synthetic sewage to a peptone and meat extract based recipe with COD of 460 mg.L-1 . Permeate was continuously removed with a peristaltic pump (Watson-Marlow Ltd, model 505 S, Falmouth, UK) connected to the membrane module. 2.2 Experimental analysis Biomass was collected and analysed for solids content, size, viscosity, dewaterability and organic content in terms of EPS (extracted extracellular polymers) and SMP (soluble microbial products). EPS in the context of this paper refers to the organic material bound to the biomass recovered by a process of heating/solvent extraction following the procedures of Zhang [2]. SMP refers to the soluble/colloidal fraction with supernatant collected after centrifuging the biomass. The MLSS and MLVSS content were determined using the Standard Methods (APHA et al., 1998). The particle size distribution was measured using Malvern Mastersizer 2000 (Malvern Instrument, UK). Viscosity was measured using a Brookfield DV-E viscometer (Brookfield Viscometers Limited, Harlow, UK) at given shear rates. Sludge dewaterability was measured using the capillary suction time (CST) test (Triton CST filterability tester, model 200, Triton Electronics Ltd, Essex, UK). Characterisation of the molecular size of a given solution was undertaken by high performance size exclusion chromatography, or HPSEC (Shimadzu LC-10 AD, coupled with a Biosep S 3000 SEC column supplied by Phenomenex). The dissolved organic carbon content was measured by Shimadzu TOC-5000A analyser. Filtration tests were carried out on specific components using a standard filter cell operated at constant pressure of 1 bar using a 0.2 µm pore size disc membranes. Off- gas monitoring was provided by pumped atmospheric samples from the headspace of the reactor onto GC sample tubes containing Chromosorb resin. Sampling was carried out for 4 hour runs in duplicate for each system, at 20 mL.min-1 giving a total sample volume of 4.8 L. Samples were desorbed and analysed by National Physical Laboratory, Middlesex, UK. These tests were carried out on 3L porous pot rig at steady state, with 24 hour HRT, 30 day SRT and MLSS concentration of 4g.L-1 . 3

Results and discussion

3.1 Fouling Results from short term fouling experiments by critical flux determination (fluxstepping) for an MBR fed with produced water are compared with those found for synthetic and real sewage sources from a previous study using the same pilot scale rig [3] in Figures 1-

3 3. A step height of 2 L.m-2 .h-1 for synthetic sewage and 3 L.m-2 .h-1 for produced water and real sewage were used with step durations of 15 minutes. The systems were kept at steady state with an MLSS concentration of 6 g.L-1 for produced water bioreactor and 3 g.l-1 for real and synthetic sewage fed bioreactors. The critical flux value is generally determined visually as the last flux step at which the transmembrane pressure (TMP) remains constant [4]. According to this definition, and from Fig 1, the value of the critical flux for produced water can be defined as 6 L.m-2 .h-1 , with a corresponding TMP of 12 mbar, since for higher fluxes the TMP no longer appears stable. However, closer examination of the initial flux steps reveals that the TMP is never absolutely constant at any point during the test. The lowest flux measured of 3 L.m-2 .h-1 yields a net TMP increase of 4.5 mbar, giving a fouling rate (dP/dt) of 0.025 mbar.min-1 . Fouling rates accelerated at increasing fluxes (Table 1), with maximum dP/dt reaching 43.6 mbar.min-1 for the final flux step (18 L.m-2 .h-1 ). The corresponding average TMP (Pave) during this step was 719.7, though approximately halfway through the step the pressure transducer reached its maximum value (Figure 1). The instantaneous TMP increase (∆P0 ) (Table 1) shows ∆P0 to be stable at lower fluxes, ranging from 1.3 to 2.2 mbar for the first three steps and increasing with increasing flux to reach 8.3 mbar at 18 L.m-2 .h-1 . The comparable fouling rate values at the same flux, for real and synthetic sewage fed bioreactors (Figures 2 and 3) were significantly lower than those measured for the produced water analogue, as indicated in Table 1.

1000

25

20

300 20

800

400

15

200

10

Flux (LMH)

10

TMP (mbar)

600

Flux (LMH)

TMP (mbar)

15

100

5 200

5

0

0 0

0.5

1

1.5 Time (hr)

2

2.5

0

3

0 0

1

2

3

4

5

Time (h)

Fig. 1. Critical flux determination, produced water 50

Fig. 2. Critical flux determination, synthetic sewage

25

200

20

160

30

15

120

20

10

10

5

40

0

0

0 0

1

2

3 Time (h)

4

5

Fig. 3. Critical flux determination, real sewage

TMP (mbar)

40

Flux (LMH)

TMP (mbar)

municipal 9 LMH synthetic municipal 7 LMH produced water 8 LMH

80

0

2

4

6

8

10

12

Time (d)

Fig. 4. TMP transients for longer term trials

Long-term experiments have been carried out for produced water fed MBR operating at 8 L.m-2 .h-1 . The data from these tests have been compared to both synthetic and real sewage matrices at 7 L.m-2 .h-1 and 9 L.m-2 .h-1 respectively (Figure 4).

4 Table 1. TMP behaviour during test fed with municipal wastewater and produced water J L.m-2 .h -1 dP/dt mbar.min -1 Pave mbar ∆P0 mbar 3 6 9 12 15 18 22

MW 2 2.3 1.4 1.7 1.7 1.3 2.4

PW 1.3 1.7 2.2 4.2 4.5 8.3 -

MW 0.023 0.059 0.095 0.097 0.119 0.140 0.328

PW 0.3 0.9 2.1 10 24.2 43.6 -

MW 2.2 4.5 7.0 10.1 13.3 16.5 21.7

PW 3.2 12.2 34.9 125.3 381.8 719.7 -

K L.m-2 .h -1 .bar-1 MW 1471 1306 1293 1187 1141 1105 944

PW 952.4 491.8 257.9 95.8 39.3 25 -

Trends from produced water show TMP to have increased fairly slowly for the first three days (from 4 to 25.2 mbar, corresponding to dP/dt = 0.005 mbar.min-1 ) and then rapidly up to 147 mbar at day 4. In the case of the analogue sewage, TMP increased only very slowly for the first four days (from 11 to 17 mbar, corresponding to dP/dt = 0.001 mbar.min-1 ) and then exponentially up to 176 mbar at day 8. For the real sewage operating at 9 L.m-2 .h-1 , the TMP seemed stable for up to 10 days. However, the TMP increased from 14 to 22 mbar within the 9 first days of filtration (dP/dt = 7.10-4 mbar.min-1 ). After day 10, the fouling rate accelerated to produce a TMP value of 54 mbar on day 12. Clearly, produced water has a significant fouling propensity certainly greater than that of both synthetic and real sewage. These trends agree with those proposed by Cho and Fane [5], where fouling is considered to be a two step process; a slow rise in TMP from gradual fouling followed by a sudden rise in TMP and rapid fouling resulting from uneven accumulation of trace foulants on the membrane surface increasing local flux equal to the effective critical flux. EPS level has been identified as being primarily responsible for fouling in MBRs [6] representing up to 90% of the total filtration resistance [7]. During this trial, EPS protein and carbohydrate were measured at 73 and 30 mg per g of MLSS for an MBR was fed with synthetic sewage, and slightly lower (60 and 17 mg per g of MLSS for protein and carbohydrate respectively) for real sewage; the significantly lower proteinaceous EPS level measured for real sewage may account for the lower fouling rate recorded for this matrix. However EPS values found for produced water fed bioreactor are significantly lower (20.3 and 13.6 mg per g of MLSS for protein and carbohydrate respectively), suggesting that other biomass quality determinants besides the measured EPS level may impact upon fouling. 3.2 Characterisation of organic matter HPSEC analysis may be used to indicate the apparent molecular weight (AMW) of compounds in the sample solution, and has been extensively used for characterising natural organic matter. MBR permeate has been compared with filtrate from the filter cell test, the biomass and the feed analogue (Figures 5-8). HPSEC traces for the feed (Figure 5) and permeate (Figure 7) appear similar but differ in absorption intensity and thus, presumably, constituent concentration. It is evident, from Figure 8, that the bioreactor membrane removes substantial quantities of the material(s) eluting at 5.5 minutes. Comparison of HPSEC profiles of SMP and extracted EPS for bioreactors fed with produced water and municipal and synthetic municipal sewage reveal marked differences (Figures 9-10). The principal peak at 5.5 minutes of the produced water analogue trace is absent in the municipal wastewater biomass HPSEC trace.

5 7000

180000 160000

6000

140000 UV absorbance

UV absorbance

5000 4000 3000

120000 100000 80000 60000

2000 40000 1000

20000

0

0 0

2

4

6

8

10

12

14

16

0

2

4

6

8

Elution time (mins)

10

12

14

16

14

16

Elution time (mins)

Fig. 5. HPSEC profile of produced water feed

Fig. 6. HPSEC profile of biomass Permeate

16000

Biomass after filtration test

70000 14000 60000 50000

10000

UV absorbance

UV absorbance

12000

8000 6000

40000 30000

4000

20000

2000

10000

0

0 0

2

4

6

8

10

12

14

16

0

2

4

6

Elution time (mins)

Fig. 7. HPSEC profile of permeate, produced water feed Produced water

10

12

Municipal

Fig. 8. HPSEC profile comparison permeate and biomass filtrate, produced water feed

Synthetic municipal

Produced water

160000

1400000

140000

1200000

120000

Municipal

Synthetic municipal

1000000 UV absorbance

UV absorbance

8 Elution time (mins)

100000 80000 60000

800000 600000 400000

40000

200000

20000 0

0 0

2

4

6

8

10

12

14

16

0

2

Elution time (mins)

4

6

8

10

12

14

16

Elution time (mins)

Fig. 9. HPSEC profiles of SMP from bioreactor fed with produced water and municipal sources

Fig. 10. HPSEC profiles of EPS from bioreactor fed with produced water and municipal sources

3.3 Fate of organic contaminants Analysis of the permeate revealed COD removal to be consistently greater than 95% throughout the trials, although approximately 50% of the feed is methano l. When using an aerobic process to treat wastewater containing volatile organic compounds (VOCs) loss of these contaminants through stripping must be considered [8]. Results from the atmospheric analysis (Table 2) indicate there to be no significant loss to the atmosphere of VOCs in this study. Liquid phase analysis of the permeate is needed in order to complete a species-specific mass balance. Table 2 Off gas concentration IN Liquid Flow conc. L.hr-1 µg.L-1 Benzene 35000 0.125 Toulene 19000 0.125 Ethylbenzene 4000 0.125 Xylene 10000 0.125

Mass flow µg.hr-1 4375 2375 500 1250

OUT Off gas conc. µg.L-1 0.467 0.351 0.076 0.074

Flow L.hr-1 72 72 72 72

Mass flow µg.hr-1 33.6 25.3 5.5 5.3

% lost in off gas 0.77 1.06 1.09 0.43

Off gas conc. ppmv 0.14 0.09 0.02 0.02

6 4

Conclusions

A study of fouling and biomass speciation for MBR treatment of produced water and sewage anologues and real sewage has revealed marked differences both in hydraulic behaviour and foulant species. When fed with the produced water analogue the mean permeability was found to decrease by 50-65% for each 3 L.m-2 .h-1 flux step between 3 and 15 LMH, finally decreasing to 25 L.m-2 .h-1 bar-1 at an imposed flux of 18 L.m-2 .h-1 . At the same flux the municipal sewage permeability was around 1100 L.m-2 .h-1 bar-1 . On the other hand, both the proteinaceous and carbohydrate EPS levels were substantially higher in the municipal biomass, suggesting that this parameter cannot in isolation be considered to be an indicator of fouling propensity for a given biomass concentration. HPSEC analysis reveals that UV-absorbing (and hence ostensibly aromatic) organic chemical species in the biomass differ substantially for MBRs fed with different feedwaters. Acknowledgements The authors would like to thank UK Engineering and Physical Sciences Research Council for their financial support. The authors would like to thank the following industrial sponsors for their financial and technical support: British Petroleum, Shell Global Solutions and Hamworthy KSE. References [1] Stephenson, T., Jefferson, B., Judd, S. and Brindle, K. (2000) Membrane bioreactor for wastewater treatment, IWA publishing, London, UK. [2] Zhang, X., Bishop, P. L. and Kinkle, B. K. (1999) Comparison of extraction methods in quantifying extracellular polymers in biofilms. Wat. Sci. Technol., 39 (7), 211-218. [3] Le-Clech, P., Chang, I-S., Smith, S., Jefferson, B. and Judd, S. (2002) (in press) Critical flux determination by the flux-step method in submerged membrane bioreactor. J. Membr., Sci. [4] Chen V., Fane A.G., Madaeni S. and Wenten I.G. (1997) Particle deposition during membrane filtration of colloids: transition between concentration polarization and cake formation. Journal of Membrane Science 125 (1), 109-122. [5] Cho, B. D. and Fane, A. G. (2002) Fouling transients in nominally sub critical flux operation of a membrane bioreactor. J. Membrane Sci., 209, 391-403. [6] Chang I.S. and Lee C.H. (1998) Membrane filtration characteristics in membranecoupled activated sludge system - the effect of physiological states of activated sludge on membrane fouling. Desalination 120 (3), 221-233. [7] Shin, H. S., An, H., Kang, S. T., Choi, K. H., and Jun, K. S. (1999). Fouling characteristics in pilot scale submerged membrane bioreactor. Proc., 1st WEFTEC ’99, New Orleans. [8] Scholz, W. and Fuchs, W. (2000) Treatment of oil contaminated wastewater in a membrane bioreactor. Water Research 34 (14), 3621-3629.