Q IWA Publishing 2008 Water Science & TechnologyWST | 57.

2 | 2008

291

Design and start-up of a high rate anaerobic membrane

bioreactor for the treatment of a low pH, high strength,
dissolved organic waste water
P. J. Van Zyl, M. C. Wentzel, G. A. Ekama and K. J. Riedel

ABSTRACT
A Submerged Membrane Anaerobic Reactor (SMAR) is being developed for the treatment of
waste water originating in Sasols coal to fuel synthesis process. The laboratory-scale SMAR uses
A4-size submerged flat panel ultrafiltration membranes to induce a 100% solids-liquid separation.
Biogas gets extracted from the headspace above the anaerobic mixed liquor and reintroduced
through a coarse bubble diffuser below the membranes. This induces a gas scour on the
membranes that avoids biomass immobilization and membrane fouling. The substrate is a high
strength (18 gCOD/) petrochemical effluent consisting mostly of C2 to C6 short chain fatty acids
with a low pH. Because of this, the pH of the reactor has to be controlled to a pH of 7.1. Organic
Loading Rates of up to 25 kgCOD/m3reactor

volume/d

P. J. Van Zyl
M. C. Wentzel
G. A. Ekama
Water Research Group, Department of Civil
Engineering, University of Cape Town,
Rondebosch, Cape Town 7701,
RSA
E-mail: vzypie007@mail.uct.ac.za
K. J. Riedel
Sasol Technology R&D Environmental Science &
Technology, Sasolburg,
RSA
E-mail: KarlHeinz.Riedel@sasol.com

has been observed with effluent COD normally

,500 mgCOD/ and FSA ,50 mgN/ with no particulates .0.45 mm at hydraulic retention times
of 17 hours. 98% of the COD is converted to methane and the remainder to biomass. Mixed
Liquor (MLSS) concentrations .30 gTSS/ can be maintained without deterioration of membrane
fluxes, even though the Diluted Sludge Volume Index (DSVI) indicates that the sludge cannot be
settled. No noteworthy deterioration in membrane performance has been observed over the 320
day operational period.
Key words

| anaerobic digestion, biogas recycle, MBR, membrane scour, SCFA

INTRODUCTION
Bulking, oxygen requirements and a high sludge production

historically hampered the implementation of anaerobic

make the aerobic treatment of Sasols Fischer-Tropsch Acid

systems: e.g., slow growth rates resulting in long start-up

Water (FTAW) a high strength dissolved petrochemical

times, long retention times and poor solid liquid separation

effluent problematic and costly. The FTAW stream has a

(SLS) particularly for dissolved organics. Attempts have

COD load equivalent to that of 5.4 million people. This

been made to overcome these SLS problems by either

effluent is currently being treated in the second largest

including large footprint secondary settling tanks (SST) or

activated sludge plant in the world at Secunda, South Africa

by immobilising/granulating the biomass, e.g. in the UASB

(Phillips & Du Toit 2001). The composition of this stream is

and EGSB systems (Britz et al. 1999). However it was

mostly short chain fatty acids (SCFA), which makes it

proven

amenable to anaerobic treatment. Advantages of anaerobic

petrochemical effluent is not prone to granulation (Britz

treatment will include; (1) high COD removal rates

& Nel 1986) and fixed bed systems have long start-up times

resulting in smaller reactor volumes, (2) no oxygen require-

and de-sludgeing problems.

that

anaerobic

biomass

cultured

on

this

ments, (3) . 90% lower sludge production, and (4) biogas

The combination of anaerobic digestion and membrane

containing energy rich methane. A few problems have

technology has, up to this point, not been very successful

doi: 10.2166/wst.2008.083

292

P. J. Van Zyl et al. | Submerged membrane anaerobic reactor

Water Science & TechnologyWST | 57.2 | 2008

since fouling of the membranes is a major problem (Yoon

The synthetic feed has a composition based on the

et al. 2004). This project attempts minimize membrane

Fischer-Tropsch Acid Water (FTAW) produced at the Sasol

fouling by of using flat panel Kubotaw membranes, with

2 & 3 plants in Secunda, South Africa. This stream

biogas recycled for scour (and mixing) to overcome

comprises mostly C2ZC6 SCFAs and some methanol and

membrane fouling. Membranes provide a means to separate

ethanol. FTAW is a high strength soluble organic effluent

the Hydraulic (HRT) and Solids Retention Time (Rs).

with very low alkalinity and no nutrients (Phillips & Du

This will allow for high mixed liquor concentrations

Toit 2001). Synthetic FTAW is made up to 18 gCOD/ and

(.12 gTSS/), smaller reactor volumes and a superior

buffered with NaHCO3 and the reactor pH is controlled

effluent quality with no particulates . 0.45 mm (Ramphao

with a NaOH dosing pump, maintaining the reactor pH at

et al. 2004). This paper reports on the design and start-up of

7.1, which is in the optimal range for acetoclastic

a laboratory scale submerged membrane anaerobic reactor

methanogenesis (Britz & Nel 1986).

(SMAR) with biogas scour for the treatment of Sasols

FTAW. Primary focal topics will include (1) the response
of the flat panel membranes to the anaerobic biomass,
(2) Organic loading rates and effluent qualities that were

SYSTEM RESPONSES

achieved, (3) nutrient requirements and (4) operational

System start-up

mixed liquor concentrations.

To promote membrane scour a substantial solids concentration is required in the reactor, which was obtained from a
primary sewage sludge digester. To acclimatize this anaero-

EXPERIMENTAL METHODS

bic biomass to FTAW, initially a mix of 70% glucose and

The laboratory-scale submerged membrane bio-reactor

liquid volume is 23 , and is maintained at 378C with a
200 W heating element. The biogas headspace above the
liquid volume is 35 , intentionally large in the event of
foaming problems. The biogas in the headspace is recycled
via a compressor and reintroduced through a coarse bubble
diffuser situated below the membranes. The gas recirculation ensures that the reactor is completely mixed and also
scours the submerged membranes. The excess biogas
escapes the system through a pressure release valve,
which ensures that the reactor is operated at a small
positive pressure (50 mmH2O). This is done to (1) avoid air
(oxygen) leaks into the system and (2) to induce an
additional Trans Membrane Pressure (TMP) over the
membranes. Three A4 size flat panel Kubota
with a surface area of 3 0.117m

membranes

are used. These

membranes have a nominal pore diameter of 0.45 mm

30% acetic acid was introduced, first at a low concentration

and then gradually increasing up to 18 gCOD/. Reaching
18 gCOD/, the glucose and acetic acid was gradually
replaced with FTAW. Thereafter the Organic Loading Rate
(OLR) [kgCOD/m3reactor volume/d] was increased by increasing the flow through the reactor, with the governing factor
being an effluent COD ,500 mgCOD/.
Originally, alkalinity was dosed via the dosing pump,
and only NaHCO3 was used. However, a large amount of
HCO2
3 was required to maintain the pH at 7.2, viz.
21.4 kgNaHCO3/m3

reactor volume/day.

This pH control

scheme was revised. Currently, pH control is done by (1)

adding alkalinity to the feed -in NaHCO3 form to increase the
buffer capacity (800 mgCaCO3/) and pH (3.66) of the
synthetic feed medium to that of actual FTAW and (2) to
control the reactor pH with a 10% (w/w) NaOH solution via
the dosing pump.

and are designed to operate at a TMP between 10

1,000 mmH2O. These membranes have a gas scour requirement of 0.75 m3gas/h/m2membrane and typical fluxes are in

Organic removal rates and effluent quality

the 20 27 effluent/m2membrane/h range at TMPs between

During the first 180 days of operation, insufficient nitrogen,

100 150 mm H2O and 20 gTSS/ sludge concentration

phosphorus and micro-nutrients were fed which unnecess-

(Ramphao et al. 2004).

arily prolonged the start-up time. After this was remedied,

293

Figure 1

P. J. Van Zyl et al. | Submerged membrane anaerobic reactor

Water Science & TechnologyWST | 57.2 | 2008

Organic loading rate and effluent quality vs. time.

by increasing the nutrient dosages, a 1000% increase in the

incrementally increasing the influent flow rate, the TMP

OLR was observed in 120 days (Figure 1).

showed no increase.

From day 210 (15 Apr, Figure 1), full strength FTAW
3

Normally higher fluxes require higher TMP, but the

has been fed and OLRs of up to 25 kgCOD/m /d have been

membrane fluxes are only 1 10% of the 20 30 /m2/h. This

observed. Effluent CODs , 500 mgCOD/ with no particu-

over-sizing of the membrane surface area was intentional

lates . 0.45mm is observed under normal conditions (HRT

because (1) there was no prior knowledge as to how the

^ 17 hours). The system seems robust and recovers from

membranes would respond to the anaerobic biomass and

shock loadings (130%) within 2 to 3 HRTs.

(2) the upper limit of the OLR was also not known.

Trans membrane pressure and fluxes

Nutrient requirements

The SMAR has been in operation for 320 days. Parameters

Since FTAW is chemically created water, it has no natural

measured daily include TMP over and effluent flux through

nutrients and very little alkalinity (, 50 mgCaCO3/). From

the membranes (Figure 2). As the OLR increased by

Figure 3, it can be seen that nutrients have a significant

Figure 2

TMP and membrane flux vs. time.

P. J. Van Zyl et al. | Submerged membrane anaerobic reactor

294

Figure 3

Water Science & TechnologyWST | 57.2 | 2008

MLSS & sludge age vs. time.

effect on OLR (compare day 1 180 with low nutrients to

the upper limit of the Mixed Liquor Suspended Solids

day 181 to 320 with sufficient nutrients). Most prominent

(MLSS) and the factors that govern this limit, is not yet

macro nutrients include N and P and micro nutrients

defined. During start-up of the SMAR, the MLSS was kept

include Ca, Co, Cu, Fe, I, K, Mg, Mn, Mo, S and Zn. The

as high as possible to decrease the start-up period as much

membranes trap all the particulate COD and endogenous

as possible. No increase in TMP was observed for MLSS

biomass inside the SMAR, where it can be hydrolyzed

concentrations up to 36 gTSS/. From Figure 3 it can be

almost to completion. This induces an abnormally high

noted that the MLSS has gradually increased over time and

nutrient recycle within the biomass. It was found that the

from day 170 to 320 it was controlled at 30 gTSS/, which

system can operate with a COD:N:P ratio of 2000:10:1

yielded a sludge age of 92 days. The MLSS is typically 91%

without nutrient deficiency. This is more than three times

VSS. The long sludge age is due to the very low biomass

higher than the typical COD:N:P ratio of 650:10:1 used for

yield of the acetoclastic methanogens, the most abundant

anaerobic systems treating acidic substrates. An operational

trophic group in the system. The sludge settleability in terms

effluent ammonia (FSA) , 50 mgN/ can be maintained.

of the Dilute Sludge Volume Index (DSVI) of the sludge

mass was around ^3,000 ml/g. At the MLSS concentration
of , 30 gTSS/, it does not settle at all.

Mixed liquor: operational concentrations and

characteristics
In aerobic membrane reactors, the upper limit of the mixed

Mass balance

liquor concentration (up to18 gTSS/) is governed by the

Due to the low biomass yields, on average only 1.36% of the

alpha values of oxygen transfer (Ramphao et al. 2004). In

COD fed to the system (over 70 days) was converted to

anaerobic reactors, where oxygen transfer is not required,

biomass, 98.6% was converted to biogas. The biogas

Table 1

Average COD mass balance over 70 days

Hydraulic retention

COD to

Mass

time (HRT)

Sludge age (Rs)

Influent COD

Effluent COD

COD load

Methane
in biogas

COD to methane

biomass

balance

[hours]

[days]

[mgCOD/l]

[mgCOD/l]

[gCOD/d]

[%]

[gCOD/d]

[gCOD/d]

[%]

31.5

175

19,101

612

334.86

56.81

342.61

4.56

103.7

295

P. J. Van Zyl et al. | Submerged membrane anaerobic reactor

Water Science & TechnologyWST | 57.2 | 2008

produced was between 55 63% methane and up to

of alkalinity dosing will also be explored in detail at a

400 /day (STP) was produced. Table 1 gives an average

later stage of the project. The system alkalinity require-

COD mass balance over the system for the last 70 days (day

ment is currently 2.5 kgNaHCO3/m3reactor

250 to 320).

and ^ 0.33 kgNaOH/m3reactor volume/day

25 kgCOD/m3reactor volume/d.

The OLR increase strategy is to increase the daily

volume/day

at an OLR of

COD load on the system until the effluent COD is

More than 98% of the COD is converted to methane.

. 500 mgCOD/, this OLR is then maintained until the

This energy rich biogas can be used for digester

system recovers and the effluent COD , 500 mgCOD/.

heating, electricity generation or even recycled for fuel

Since the system is still in the OLR increase phase, the

production. It is estimated that if the biogas produced

average COD load and biogas production is relatively low

from the anaerobic digestion of the entire FTAW stream

compared to later values (Figure 1). However, over the 70

is converted to electricity, 2.02 kWh/kgCODRemoved can

day period a mass balance of 103.7% could be maintained.

be produced at an efficiency of 0.33; this will be , 7

times more than is required to operate the system.
The excess electricity generated would be sufficient

CONCLUSIONS

to power a large town (.10,000 average American

house holds).

The Submerged Membrane Anaerobic Reactor (SMAR)

developed for the treatment of Fischer-Tropsch Acid Water
has been in operation for 320 days and the following
conclusions can be drawn.
FTAW can be treated in a SMAR; OLRs of up to

ACKNOWLEGMENTS
This research was supported by the National Research

25 kgCOD/m /d are possible with effluent quality of

Foundation, Sasols Environmental Science & Technology

COD ,500 mgCOD/, FSA , 50 mgN/ and no parti-

Department and the University of Cape Town and is

culates . 0.45mm, under normal operating conditions.

published with their permission.

The recirculation of biogas for reactor mixing and

membrane scour appears to be successful. After 320
days of operation no deterioration in membrane performance has been observed. Although the membranes
have been operated at only 10% of their operational flux
capacity, the flux-TMP response corresponds well with
tap water fluxes at the same TMP.
MLSS concentrations .30 gTSS/ were maintained
without deterioration of membrane fluxes, even though
the DSVI (3000 ml/g) indicates that the sludge cannot
be settled by traditional methods. The process produced
significantly less biomass (2 3%) than traditionally used
aerobic systems, and has no oxygen requirements.
The total alkalinity of the system needs to be controlled
within the 1500 3500 mgCaCO3/ range, to avoid
complications like high Ripley Ratios or Na poisoning.
This is done by adding buffer (HCO2
3 ) to the influent and
by NaOH dosing used for pH control. The optimisation

REFERENCES
Britz, T. J. & Nel, L. H. 1986 The influence of different substrate pH
values on the performance of a down-flow anaerobic fixed bed
reactor treating petrochemical effluent. Biotech. Let. 8(4),
293 298.
Britz T. J., Trnovec W., Van Schalkwyk C. & Roos P. 1999
Enhanced granulation in up-flow anaerobic sludge bed
digesters (UASB) by process induction and microbial
stimulation. WRC Report. No 677/1/99.
Phillips, A. & Du Toit, P. J. 2001 Water Reuse and Re-cycling at
Sasol. Water and Environmental Technology R & D,
Sasolburg, RSA.
Ramphao M., Ekama G. A., Lakay M. T., Mafungwa H. & Wentzel
M. 2004 The performance and kinetics of biological nitrogen
and phosphorus removal with ultra-filtration membranes for
solid liquid separation. UCT Msc Thesis, 200 pages.
Yoon, T. I., Lee, H. S. & Kim, C. G. 2004 Comparison of pilot scale
performances between membrane bioreactor and hybrid
conventional wastewater treatment systems. J. Membr. Sci.
242, 5 12.