Effect of attached growth on membrane performance
The membrane performance of BF-MBR varies depending on experimental condition. If suspended solids are low, e.g. lower than 3-5 g/L, membrane can foul faster than in conventional MBR with higher MLSS without biocarrier. On the contrary, if MLSS of BF-MBR is higher than 3-5 g/L, most literature report that membrane fouling is either not affected or reduced comparing to that of conventional MBR with suspended culture.
Little has been known about the membrane performance in full-scale BF-MBR, but some data from lab-scale is known. A side-by-side comparison was performed with a lab scale BF-MBR and a conventional MBR at 10 g MLSS (Sombatsompop, 2007). In this study, only a small portion of microorganisms was attached on biocarrier and this portion was neglected in MLSS calculation. Filtration was performed intermittently at 10 minutes on and 2 minutes off. The membranes used was 0.1 micron polyethylene hollow fiber membranes with 0.42 m2 surface area (STNM424 model, Mitsubishi Rayon). It was observed that TMP rose much slower in BF-MBR as shown in Fig. 1. As a result, membrane cleaning interval extended around 5 folds. Particles were found substantially smaller in BF-MBR (~30 micron vs ~200 micron), but no significant differences were found in polysaccharides and protein contents in mixed liquor.
Fig. 1. TMP changes with time at MLSS of 10 g/L in suspended and attached reactors at 7 LMH, where the attached microorganisms in attached growth MBR was neglected when MLSS was measured (Sombatsompop, 2007)
Biocarrier volume fraction (ratio of biocarrier volume in reactor) affects the extent of membrane fouling mitigation. In one experiment as shown in Fig. 2a, the time required to reach a TMP of 30kPa increases as biocarrier volume fraction increases regardless of the air flow rate. Ironically the average particle sizes were found smaller with higher biocarrier volume fraction (75.3 micron for 5% and 49.6 micron for 20%) though the critical flux theory suggests smaller particle size can hamper membrane performance. The lower membrane fouling tendencies at higher biocarrier volume fraction was attributed to the lower extracellular polymeric substances (EPS) concentration (Fig. 2b).
Fig. 2. Effect of biocarrier volume fraction on: a) membrane fouling, b) extracellular polymeric substances (EPS) (Lee, 2006).
It may not be feasible to allow biocarriers to directly contact with membranes in full-scale BF-MBR due to the potential damage on membrane, but the data from a lab scale experiment shows that it is an effective method to mitigate membrane fouling. When polyurethane sponge cubes coated with activated carbon (13 mm) was added to mixed liquor with and without iron net that separates the cubes from membrane, a dramatic difference was observed in TMP rising rate depending on the existence of the iron net. As shown in Fig. 3, TMP rose much slower when cubes were allowed to contact with membranes directly (no iron net) although polysaccharides and protein levels were found similar in both reactors. It was apparent that the direct contact of sponge-type biocarriers with membrane reduced membrane fouling.
Fig. 3. TMP curve with and without direct contact of membranes by biocarriers (Lee, 2006).
Interestingly, if suspended solids levels are too low, e.g. < 3-5g/L, membrane fouling can accelerate with biocarriers. In one experiment performed with lab scale BF-MBR, where looped codes were fixed around a U-shaped hollow fiber bundle as a biocarrier, TMP rose faster as suspended solids decreased. Same trend was observed in the control reactor without biocarrier (suspended growth) as shown in Fig. 4. The differences in MLSS in the two groups of data were caused by the attached microbial growth on biocarrier that is not detected as MLSS. In the reactor with biocarrier, soluble polysaccharides and protein levels were found slightly lower and particles were somewhat smaller, which is in line with other experiments shown above.
The exact cause of accelerated membrane fouling at low MLSS is hard to be determined, but one potential simple explanation is that the higher F/M ratio at lower MLSS in attached growth, which also means lower SRT, caused higher membrane fouling rate (Lee, 2002). In the experiment shown in Fig. 4, F/M ratio increases as MLSS decreases since same amount of feed solution is fed to less suspended solids (or microorganisms). The increased F/M again causes decreased solids retention time (SRT) since more solids must be removed from the biological tank to keep the low MLSS. As a consequence, SRT tends to decrease as MLSS decreases under a given organic loading rate. As detailed here, membrane fouling is much better correlated with F/M ratio rather than MLSS and this might be a reason why membrane fouls faster when MLSS is low. Despite of this explanation, the possibility that low MLSS itself contributes to the accelerated membrane fouling still remains.
Similar observations were also made in an independent study (Wei, 2006b), where the positive impact on flux was observed only at high MLSS such as 8 g/L and at low biocarrier dose, e.g. 5 v/v%. At either low MLSS or high biocarrier dose, membrane fouling was accelerated. It was attributed to the collision effect of the floating biocarrier on microbial floc, which breaks floc into smaller particles.
In terms of organic removal efficiency, most literature consistently reveal there are little differences between BF-MBR and conventional MBR in typical application conditions (Lee, 2001; Basu, 2005; Johir, 2011). This is simply because MBR with suspended growth produces already high enough effluent quality and there is little room to improve it biologically.
More discussions on the membrane fouling in the presence of biocarrier is provided here.