Effect of SRT, HRT and F/M ratio on membrane performance

   SRT, HRT and F/M ratio are interconnected. Longer SRT directly means longer HRT and lower F/M ratio for a system with a fixed reactor volume, fixed influent strength, and fixed MLSS. If reactor volume increases while MLSS is constant, both SRT and HRT increase and F/M ratio decreases. Since all three parameters are interconnected, SRT will be used here to discuss the effect of the three parameters on membrane performance.

   It has been overwhelmingly reported that membrane fouling tends to decrease as SRT increases (or F/M ratio decrease). It has been also known that SRT is a more influential factor than MLSS affecting membrane fouling in typical MBR. As summarized in Table 1, when TMP rising rates were compared at three different SRTs in the two different phases of experiment, it was apparent that membrane fouling was the fastest at the shortest SRT (8 days) despite the lowest MLSS (Grelier, 2006). In other pilot tests, TMP increasing rate was measured at three different SRTs, e.g. 10, 30, and 50 days (Van den Broeck, 2012). Though MLSS was nearly five times greater at 50 day SRT than at 10 day SRT (~14 g/L vs ~3 g/L), TMP rose 58 times slower (0.187 mbar/d vs 10.8 mbar/d). The high membrane fouling rate at 10 day SRT was attributed to the deflocculation of microbial floc. In other study (Trussell, 2006), it was found that there was a critical SRT below which membrane fouling dramatically accelerated. Under the condition employed, TMP rising rate increased abruptly when SRT was shorter than 4-5 days. 

   In certain cases, the lower membrane fouling tendency at high SRT can be attributed to lower polysaccharides concentrations. As shown in Fig. 1, polysaccharides concentrations in mixed liquor tends to be decline as SRT increases. In the same study, when filterability of the sludge from three different SRTs were measured using a stirred cell equipped with 0.22 micron filter at 0.5 bar, the filtrate volume collected for the first 3 minutes were inversely proportional to the polysaccharides concentration. The same trend was observed in a pilot study using PVDF hollow fiber membranes with 0.1-0.2 micron pore size as shown in Fig. 2a.

   Determination of the optimum SRT is very complicated since it involves not only the membrane performance, but also capital costs, operating costs, oxygen transfer efficiencies (OTE), redundancies in tank volumes for peak flow, etc. For example, as discussed here, oxygen demand increases as SRT increases to meet high endogenous respiration rate, but the rising OTE partially relive the aeration demand. Likewise, longer SRT increases capital costs to build larger aeration tanks, but it not only reduces operating costs for membrane scouring and cleaning by reducing membrane fouling, but also provide redundant sludge holding capacity when sludge hauling is interrupted. Currently, most municipal MBR are designed targeting SRT of 20 days or so, but target SRT can go down to 12-15 days in some cases. The minimum SRT targeted by GE is 12 days (Diamond, 2010).

   Fig. 3 shows TMP profiles obtained with four parallel bench scale MBR that run at various SRT (3, 5, 10, 20 days) using flat sheet immersed membranes. While 1/3, 1/5, 1/10, and 1/20 of the total mixed liquor was removed from the system everyday, average MLSS of the four MBR were 4.48, 7.40, 13.04, 21.90 g/L, respectively. To compensate the unequal mixed liquor removal, the flux of each system was maintained at 15.6, 16.1, 16.6, and 16.9 LMH, respectively. The figure clearly shows membrane fouling decreases as SRT increases (Ng, 2006). The only incomplete part of this experiment appears that the MLSS were not same for all SRT. Therefore, one may be able to argue the negative impact on membrane fouling at low SRT is caused by low MLSS.

   The most rigorous experiments were performed by  Trussell et al (2006). While MLSS was maintained identical at 8 g/L by adjusting sludge removal (or SRT), influent flow rate was controlled to obtain various F/M ratio. Membrane flux was maintained at 30 LMH throughout the experiment by differentiating the permeate recycle rates to the influent. The experiment performed with three full size GE ZW500c modules showed a clear positive correlation between F/M ratio and membrane fouling rate. In addition, as SRT (or MCRT) decreased, membrane fouling rate increased.

Table 1. Experimental condition, where reactors with same volume were used in Phase I and reactors with different volumes were used in Phase II (Grelier, 2006)


SRT.ht9Fig. 1. Filterability as a function of SRT. Filterability was measured by the volume of filtration collected for 3 minutes using a stirred cell equipped with 0.22 micron membrane filter at 0.5 bar with stirring (Grelier, 2006).

SRT.ht11    SRT.ht18
Fig. 2. Effect of SRT on membrane fouling rate measured by the time derivative of resistance increase.

SRT.ht36Fig. 3. Normalized membrane suction pressure (or TMP) profile at different operating mean sell residence time (MCRT). MCRT equals to SRT in MBR (Ng, 2006).

SRT.ht37Fig. 4. Membrane fouling rated represented by permeability loss rate as a function of F/M ratio (Trussell, 2006).

Fuzzy relation between SRT (or F/M ratio) and EPS/SMP

   It has been consistently observed that membrane fouls less at high SRT in various settings from lab- to full-scale MBR (Cicek, 2001; Ng, 2006; Al-Halbouni, 2007; Patsios, 2011; Van den Broeck, 2012). But, the attempt to correlate the observation with SMP and EPS has not been always successful as discussed here. No consistent relation between the quantity of SMP/EPS and membrane fouling has been found. Combining these two, it can be postulated that biopolymer quality plays more crucial role than its quantity in membrane fouling. 

   Fig.5 compares biopolymer contents of the two mixed liquors grown at two different organic loading rates, i.e. 0.17 g COD/g MLSS/day and 0.50 g COD/g MLSS/day, using two lab scale MBR. Though TMP rising rate was 5-20 times faster with high F/M ratio depending on applied flux, there were no dominant trends in biopolymer contents of the mixed liquor with respect to the eight different parameters as shown in the figure.

SRT.ht1Fig. 5. EPS characteristics and soluble TEP in the MBRs at different F=M ratios. (A) Total EPS; (B) Total protein; (C) Total polysaccharides; (D) Bound EPS; (E) Soluble EPS; (F) Soluble protein; (G) Soluble polysaccharides; (H) Soluble TEP (Wu et al., 2013)

Effect of SRT (or F/M ratio) on membrane fouling in anaerobic MBR

   Very similar SRT (or F/M) effect was observed on membrane fouling in anaerobic MBR (AnMBR) fed with synthetic feed with 500 mg/L COD (Liu, 2012). When two AnMBR equipped with immersed hollow fiber membranes were run at 3.8 g COD/g MLSS/d and 0.1 g COD/g MLSS/d, respectively, it was apparent that membranes were fouled much quickly at higher F/M ratio (or lower SRT) as shown in Fig. 6, where HAnMBR and LAnMBR indicate AnMBR with high and low organic loading rates, respectively. It was also found that cake layer formation was responsible for 99% of the permeability loss in both HAnMBR and LAnMBR when the filtration resistance was analyzed based on resistance in series model shown here.

SRT.ht2Fig. 6. TMP increase and flux decline in high loading AnMBR (HAnMBR) and low loading AnMBR (LAnMBR) (Liu, 2012).

Effect of F/M ratio in MBR on the subsequent RO membrane fouling

   It is interesting to see the biological condition not only affects the MF/UF membrane performance in MBR, but also affects the RO membrane performance when RO is used as a polishing step of MBR effluent. When two lab scale MBRs were operated at the F/M ratio of 0.17 g/g/d and 0.50 g/g/d, respectively, using synthetic feed, the subsequent RO membrane (UTC-70, Toray) with 0.0186m2 treating MBR effluent was fouled four times quicker at high F/M ratio in terms of the TMP required to obtain a constant flux (Wu et al., 2013). It was found that the concentrations of dissolved organic carbon (DOC), polysaccharides, protein, and transparent exopolymer particles (TEP) were greater in the effluent of high F/M MBR (Fig. 7a – Fig 7d). When cartridge filter was used to filter MBR permeate before RO, the fouling rate of RO decreased significantly (Fig. 7e and 7f)

   On the contrary, RO membranes were fouled less when SRT of MBR was shorter (or F/M ratio was higher) in other study with similar experimental setup (Field et al., 2010), but the exact cause of the discrepancy is not known.

SRT.ht3Fig. 7. Effect of MBR permeate quality on RO fouling rate. (A) DOC in the MBR permeate. (B) protein in the MBR permeate. (C) polysaccharides in the MBR permeate. (D) TEP in the MBR permeate. (E) averaged fouling rate of RO membrane fed with*ltered RO feed. (F) averaged fouling rate of RO membrane fed with unfiltered RO feed (Wu et al., 2013).


© Seong Hoon Yoon