Alternative approach to improving operating flux of MBR

In order to improve the flux of membrane bioreactor (MBR) system, many studies have been conducted on various mechanical, chemical and process conditions. Scouring air is normally used to introduce shear effect on membrane surface to minimize fouling. Coagulant can also be added to improve the sludge formation so that fouling is reduced (Yong et al. 2006). In addition, ozone can be used to improve MBR performance (Lee et al. 2010). MBR operating conditions such as mixed liquor suspended solids (MLSS), hydraulic retention time (HRT), sludge retention time (SRT), food to microorganism ration (F/M), return activated sludge (RAS), etc. can also be optimized for flux improvement. Since fouling of membranes is partially caused by colloidal particles which can penetrate into membrane pores, it is believed that removal of those particles from mixed liquor could improve the operating flux of MBR systems. Hydrocyclone can be used to segregate different sizes of particles from the mixed liquor and thus it is selected for the separation of the dense particles from the RAS stream to be kept in the MBR system. Lighter particles are disposed as waste activated sludge (WAS) from the overflow of the hydrocyclone. This approach was investigated at a MBR system which is treating domestic sewage in Singapore.

Objective of this study is to evaluate the performance of the MBR when a hydrocyclone is introduced in activated sludge system.

Figure 1 shows the MBR system (1,000 m³/d) with hydrocyclone which was installed to treat a partial stream of the RAS. After the installation of the hydrocyclone, dense particle fraction of the mixed liquor was returned to the anoxic/oxic tank as RAS. The lighter particle fraction of the RAS overflow from the hydrocyclone was removed as WAS. Respective flow rates of feed, underflow and overflow of hydrocyclone are 1.8 m³/h, 0.3 m³/h and 1.5 m³/h.

pH, total suspended solids (TSS), biochemical oxygen demand (BOD) and chemical oxygen demand (COD) of MBR feed (influent) were in the range of 6.0–7.5, 100–600 mg/L, 100–450 mg/L and 400–1,100 mg/L, respectively. MLSS in aeration and membrane tanks were maintained at 2,500 mg/L and 4,000 mg/L, respectively. HRT of the anoxic tank and the oxic tank was 4.46 hours. And, SRT was targeted to be approximately 5 days.

Figure 2 shows the MLSS level in the oxic tank before and after installation of hydrocyclone. Generally, MLSS was around 2,000–3,000 mg/L most of the time. However, MLSS increased sometimes due to carry over of influent from the side stream. It was as high as 8,000 mg/L during Dec 2015 and Nov 2016. On the other hand, accidental or unscheduled WAS during system maintenance may have caused low MLSS. During Aug 2016–Mar 2017, MLSS fluctuated significantly due to frequent carry over and system maintenance.

Figure 3 shows the operating range of time to filter (TTF) of the sludge in the membrane tank for the period of Jan 2015–Mar 2017. It was observed that the TTF remained around 40 sec before the installation of hydrocyclone. Then, TTF decreased to the level of 20–25 sec during initial 7 months after the hydrocyclone was in place to improve the sludge quality. However, TTF increased again to the range of 30–40 sec after a spike to more than 60 sec around end of Aug 2016. High TTF appeared after the incident with the MLSS spike of approximately 6,000 mg/L around end of Jul. Performance of the hydrocyclone may have been influenced by the fluctuation of the MLSS concentration. The impact could be significant when fluctuation remains there for an extended period.

Figure 4 illustrates the sludge volume index (SVI) trend of the oxic tank sludge during Jan 2015–Mar 2017. Generally, SVI trend follows the MLSS trend as shown in Figure 2. However, it should be noted that fluctuation of SVI (100–1,000 mL/g) was wider especially when MLSS fluctuated (1,000–6,000 mg/L) for extended period from Aug 2016 toward the end of this investigation.

Monthly average of colloidal total organic carbon (cTOC) and soluble COD (COD-S) are illustrated in Figure 5. During the operating period with the hydrocyclone, cTOC was more stable in the range of 6–13 mg/L. Although COD-S was relatively high at 300–400 mg/L during Jan–Mar 2017, cTOC was stable at around 10 mg/L as shown in Figure 5.

When monthly average of cTOC and permeate COD (COD-P) are reviewed as shown in Figure 6, COP-P is found to be lower than 15 mg/L from Oct 2016 toward the end of this investigation. However, when the 5x ratio of cTOC to COD-P is studied as shown in Figure 7, the trend of final 6 months seems to move toward higher level when it is compared to the trend during earlier period of hydrocyclone installation. As shown in Figure. 3, TTF values are also relatively higher during Oct 2016–Mar 2017 compared to those during earlier period of installation.

Trends of trans-membrane pressure (TMP) before and after hydrocyclone installation for Train-1 and Train-2 are shown in Figures 8 and 9, respectively. As shown in Figure 8, TMP gradually increased from 12 kPa to 17 kPa between May to August 2015 and sharply increased towards 32 kPa within 2 weeks during August. After recovery cleaning (RC) was conducted, TMP recovered to 11 kPa and increased again to 18 kPa in 4 months period before hydrocyclone was installed for sludge improvement. After the installation of hydrocyclone, it was observed that TMP remained stable at around 16–18 kPa for almost 8 months. Then, TMP gradually increased again to 21 kPa the following 6 months and sharply increased to 35 kPa during March 2017. It should be noted that sludge characteristics and permeate quality started to deteriorate from August 2016 as discussed earlier. TMP trend of Train-2 in Figure 9 shows a similar pattern to that of Train-1 shown in Figure 8. From the TMP trend, it should be noted that recovery cleaning was not required for both trains for more than one year operation of MBR system with the hydrocyclone. This result suggested that the lighter particle fraction of mixed liquor could have negative impact on membrane fouling. Thus, it should be segregated and wasted from the MBR system.

In addition to the installation of the hydrocyclone to improve the sludge characteristics, it is necessary to pay close attention to minimize significant fluctuation of solid loading and organic loading to the activated sludge system. The operation can be extended much longer without requiring recovery cleaning.

Operation of the MBR system to treat domestic wastewater was improved by the installation of a hydrocyclone for removing lighter particle fraction of mixed liquor as a WAS stream from the system, providing a sink for the solids enhancing fouling. It was observed that increase of TMP can be suppressed at around 16–18 kPa for about 8 months while the MBR system was operating at net flux of 25 L/m²-hr(LMH) and sludge was conditioned by the hydrocyclone. Then, TMP gradually increased toward 21 kPa in the next 6 months before increasing sharply to 35 kPa after more than one year operation without any need for recovery cleaning. Typically, recovery cleaning was required every 3–4 months prior to introducing the hydrocyclone in the MBR system. Introduction of hydrocyclone results in reduction of recovery cleaning frequency and chemical cost in MBR system. Although sludge characteristics could be improved by hydrocyclone, significant fluctuation of sludge concentration for an extended period may have a negative impact on efficiency of the hydrocyclone and thus performance of the MBR system.