From R&D to application: membrane bioreactor technology for water reclamation

Since 2003, PUB Singapore has been developing the membrane bioreactor (MBR) technology for water reclamation. The MBR technology has significant versatility and can be applied for production of industrial water or combined with reverse osmosis (RO) to produce NEWater. The NEWater is PUB's trade-marked high-grade reclaimed water that surpasses drinking water requirements, and a key pillar of Singapore's water sustainability.

The conventional treatment route for NEWater production begins at a water reclamation plant (WRP), where influent used water is treated using conventional activated sludge process (CASP) that comprises the main treatment units of primary sedimentation tank, bioreactor and secondary sedimentation tank (SST). The CASP-treated water, or secondary effluent, is then supplied as feedstock to a NEWater factory, where it is further purified with dual membrane processes comprising microfiltration (MF)/ultrafiltration (UF) and RO followed by ultraviolet disinfection to produce NEWater.

To date, PUB has more than a decade of experiences with the MBR technology. From pilot studies and demonstration testing, the MBR technology has now progressed to full-scale implementation. PUB continues to embark on R&D in developing innovative MBR systems for future application. This manuscript provides an overview of initiatives for developing the MBR technology for water reclamation at PUB from R&D to full-scale application.

The beginning of PUB's R&D journey of MBR technology could be traced back to the pilot studies conducted at the now-decommissioned Bedok Water Reclamation Plant (Bedok WRP). Three MBR pilot plants with capacity of 300 m³/d each were commissioned in 2003 to test for different types of submerged membranes.

The pilot studies encompassed a range of investigations that included analysis of total organic carbon (TOC) for overall monitoring of organics, Gas Chromatography-Electron Capture Detector (GC-ECD) for scanning of homogenated organics, High Performance Liquid Chromatography (HPLC) for scanning of non-volatile organics, and Gas chromatography–mass spectrometry (GC-MS) for scanning of semi-volatile organics and various types of liquid and sludge analytical and monitoring methods. The pilot studies conclusively showed that the MBR product water quality is equivalent to or better than the UF product water quality in the NEWater process (Tao et al. 2005).

The pilot studies were subsequently expanded to include RO testing. It was found that the MBR-RO process achieved more stable and greater removal of organics as compared to the NEWater process of CASP-MF-RO. The RO-TOC values were in the range of 24–33 ppb for the MBR-RO process, while these were in the range of 33–53 ppb for the NEWater process. Furthermore, it was found that higher RO flux could be attained in the MBR-RO process as compared to the NEWater process (Qin et al. 2006).

Scaling up from the pilot studies, an MBR demonstration plant with capacity of 5-MGD (=23,000 m³/d) was constructed at the Ulu Pandan WRP and began operations in 2006. The MBR demonstration plant was designed based on a retrofitting concept, where two existing aeration tanks in the North Works of Ulu Pandan WRP were modified into anoxic and oxic zones of the bioreactor. New membrane tanks and a MBR-dedicated process control building were constructed. The MBR demonstration plant is able to produce high quality filtrate (Table 1).

Filtrate water from the MBR demonstration plant is supplied as industrial water. Industrial water is mainly supplied to chemical plants and petroleum refineries to be used for cooling applications. For industrial water supply, nitrate is not a pertinent parameter. Although the specified MBR product water requirement was 20 mg/L for nitrate, typical operational value for nitrate achieved by the MBR demonstration plant was approximately 6 mg/L and well within the specification.

As with any water infrastructure project, cost is a very important consideration. Historically, production for industrial water was based on an elaborate physical-chemical treatment process at the Jurong Industrial Water Works (JIWW), and was land-intensive (Section 3).

Using MBR technology to produce industrial water is therefore advantageous, as an MBR plant will entail lower capital costs in Singapore's context compared to conventional treatment process because of smaller footprint (it is done by retrofitting an existing activated sludge plant) and hence lower land cost. There are also operational advantages, as there is no longer the need to operate JIWW to produce industrial water.

When the MBR pilot studies were carried out on the three different types of membrane system in 2003, the specific power consumption of the three pilot plants ranged from 1.3 to 1.7 kWh/m³. The baseline operation was based on the recommendations of system suppliers, and the initial designs were found to be conservative. Process optimisation was then performed in two phases. The first phase involved increasing permeate flux while keeping the air supply rate constant. With this measure, the specific power consumption could be reduced to approximately 1 kWh/m³. The second phase involved reducing air supply rate without causing any significant change in product quality and membrane fouling rate. By the end of the second phase of pilot testing, it was established that the specific power consumption could be reduced to approximately 0.8 kWh/m³. Further information on the MBR pilot studies can be found in Tao et al. (2005).

Moving to demonstration plant testing, the guarantee figure for specific power consumption was 0.7 kWh/m³. A step by step approach was then applied to reduce specific power consumption at the 5-MGD MBR demonstration plant.

The first step was through improved design and selection of equipment, and specific power consumption of 0.6 kWh/m³ was achieved during baseline operation with mixed liquor suspended solids (MLSS) concentration of around 10,000 mg/L. The second step was to reduce the MLSS level in the aeration tank to around 6,000 mg/L by shortening the solids retention time (SRT). This optimisation measure of operating at lower SRT to save energy was feasible due to the warm climate condition in Singapore with the temperature here about 30°C. Lower MLSS leads to lower sludge viscosity and better aeration efficiency. The third step was to optimise the MLSS recirculation by recirculating the mixed liquor (2Q) from the membrane tank directly to the inlet of the anoxic tank and thereby relieving pumping for more recirculation. The fourth step was optimisation of process aeration with DO set point of 1.5 mg/L. The fifth step was optimisation of membrane scouring rate.

In the course of process optimisation, it was ensured that any power reduction was not achieved at the expense of water quality and cleaning frequency. These efforts were able to drive down the specific power consumption of MBR from the initial 1.3 kWh/m³ recorded during pilot studies to the guaranteed figure of 0.7 kWh/m³ and finally to an optimised specific power consumption of less than 0.5 kWh/m³ (Figure 2).

Operations’ experiences from the MBR demonstration plant confirm the advantages of the MBR technology that include robustness to take shock loadings, superior filtrate quality, small footprint and automation in operation. With successful pilot studies and demonstration testing, the MBR technology progressed from R&D to full-scale application at Jurong WRP (section 3), and Changi WRP (section 4).

The Jurong WRP, located in the western part of Singapore, treats used water of both industrial and domestic origin in the proportion of approximately 40% to 60%. The Plant currently has a configuration comprising 3 Phases with overall treatment capacity of 45 MGD (=205,000 m³/d).

Since 2013, Phase 1 treatment facilities at the Jurong WRP have been converted from CASP to an MBR Plant. The Jurong MBR Plant has a treatment capacity of 15-MGD (=68,000 m³/d) to treat domestic used water and produce high quality filtrate water suitable for reuse as industrial water. Together with the 5-MGD MBR Demonstration Plant at Ulu Pandan WRP (section 2), the two MBR Plants replaced the Jurong Industrial Water Works (JIWW) in supplying an alternative source of water to industries for non-potable use and thereby conserving potable water resource that is otherwise required for such purpose.

By leveraging on the MBR technology to produce industrial water, significant savings in process and land requirements could be achieved. The JIWW occupied a land area of some 5.2 hectares (=52,000 m²). It received secondary CASP-treated effluent from the Ulu Pandan WRP as feedstock, which was subjected to an elaborate tertiary treatment process comprising fine screen, pre-chlorination with sodium hypochlorite and pH correction with caustic soda, chemical clarification with alum, sand filtration, aeration, and finally, post-chlorination with sodium hypochlorite. The MBR Plants, on the other hand, avoid the need for such elaborate treatment process and require additional area of only some 2,050 m² for the construction of membrane tanks that are spread out over 400 m² at the Ulu Pandan MBR Plant and 1,650 m² at the Jurong MBR Plant.

Currently, MBR 3.1 has been successfully commissioned, plant-proved and operating since May 2015. MBR 4.1 is on-track for completion within 2015, while MBR 1.1 and 1.2 are slated for completion in 2016. When the MBR-retrofits are completed, the total treatment capacity of Changi WRP will increase from the original 176 MGD (800,000 m³/d) to 202 MGD (920,000 m³/d).

By leveraging on the MBR-technology to retrofit the available bioreactor basins, Changi WRP is able to overcome the bottleneck of the existing SST and ensure adequate supply of feedstock for NEWater production.

Liquid chromatography-organic carbon detection (LC-OCD) analysis is a useful research method that is used in MBR studies (Rosenberger et al. 2006; Lyko et al. 2008). The LC-OCD analysis provides quantitative as well as qualitative result in the form of chromatogram on the classes of dissolved organic carbon (DOC) compounds present in water. The LC-OCD can characterise DOC based on molecular weight (MW), and is able to differentiate between biopolymers, humic substances (humics), building blocks, low molecular weight (LMW) neutrals and LMW organic acids.

The LC-OCD analysis was performed by researchers from the National University of Singapore. Analysis was carried out on three different water samples at Changi WRP: (a) Effluent from conventional treatment for the liquids Train 3, (b) Filtrate from MBR 3.1, and (c): Combined product water sampled from Train 3′s plant effluent wet well where MBR filtrate is mixed with effluent from conventional treatment.

Quantitatively (Table 4), sample (b) MBR filtrate has the lowest total DOC among the 3 samples with a measured value of 6773.60 ppb (=6.77360 mg/L). This total DOC is composed of 1940.05 ppb of hydrophobic-DOC fraction and 4833.56 ppb of hydrophilic-DOC fraction. The hydrophilic-DOC fraction, in turn, can be further categorised into the different classes: biopolymers (1.38 ppb), humics (3082.18 ppb), building blocks (865.36 ppb), LMW neutrals (884.64 ppb) and LMW organic acids (n.q.).

Analysing in more detail between sample (a) conventional treatment effluent, and sample (b) MBR filtrate, the following points can be made:

• i) The superiority of the MBR filtrate in total DOC is attributed to both reduced hydrophobic-DOC fraction and hydrophilic-DOC fraction.

• ii) There is significant difference in DOC value for the fraction biopolymers, indicating that this class of organics has been effectively removed by the MBR but not by the conventional treatment.

• iii) There is no significant difference in DOC value for the fraction humics for both MBR and conventional treatment, indicating that this class of organics is neither effectively removed by the MBR nor by the conventional treatment.

• iv) There is significant difference in DOC value for the fractions building blocks and LMW neutrals, indicating that these classes of organics have been effectively removed by the MBR but not by the conventional treatment.

• v) The DOC value for the fraction LMW acids for both MBR and conventional treatment is either not quantifiable or in negligibly small amount, indicating that this class of organics is not significantly present in all water samples and therefore needs no further discussion.

The above points (i) and (ii) can be well understood as colloidal and macromolecular compounds (e.g. biopolymers) could be effectively retained by the MBR3.1, which uses UF-membrane with nominal pore size of 0.04 μm that may correspond to molecular-weight-cut-off in the range >10,000 g/mol (Judd 2006; Le-Clech et al. 2006). The same colloidal and macromolecular compounds, however, would be carried over into the effluent in the SST of conventional treatment and hence detected in that sample. It should be noted that the total DOC values were separately checked and validated by TOC-analysis carried out by another independent laboratory.

Point (iii) can be explained as humics are non-readily biodegradable substances that are small enough to pass through the UF-membrane in the MBR, and therefore are present in all water samples (Rosenberger et al. 2006; Lyko et al. 2008).

Point (iv), however, is noteworthy. In terms of MW, building blocks (300–500 g/mol) and LMW neutrals (<350 g/mol) could not be retained by the UF-membrane of the MBR, and should be detected in similar concentration in the conventional treatment effluent as well as in the MBR filtrate as is the case for humics. However, result conclusively shows that there is enhanced removal efficiency of lower molecular weight organic compounds by the MBR over the conventional treatment. The conclusion is that there is enhanced removal efficiency of lower molecular weight organic compounds including organic micro-pollutants by the MBR as a system that is not due to physical separation of the membrane alone. Similar observations and conclusions are also well documented in the literature. This is a distinctive advantage of MBR system that utilises the synergistic interaction between membrane and biological process. Current theory is that the enhanced removal efficiency of lower molecular weight organic compounds can be attributed to the formation of a gel/biofilm layer on the membrane surface that improves the separation performance of MBR system (Tao et al. 2005; Qin et al. 2006; Melin et al. 2006; Sahar et al. 2011).

Moving forward, PUB continues the efforts in developing the MBR technology. The R&D approach is to progressively advance innovative technology up the value chain from fundamental research to pilot and validation studies, demonstration testing and finally full-scale implementation. Key objectives are to increase water recovery, maintain water quality and security and reduce production and distribution cost by minimising footprint of resources including energy, land area, chemicals and manpower while coping with the warm climate condition in Singapore.

Ongoing R&D efforts include the 1-MGD (=4,550 m³/d) demonstration testing of ceramic MBR system to treat industrial used water at the Jurong WRP and the 0.22-MGD (=1,000 m³/d) integrated validation plant (IVP) to treat domestic used water at the Ulu Pandan WRP. Integral in the IVP is a process combination of enhanced primary treatment and low-energy MBR technology that was conceptualised based on extensive review of potential technologies and study on global trends in municipal used water treatment (Lee et al. 2013).

The IVP has been able to produce high quality MBR filtrate with nutrient removal at short SRT of 5 days. Further optimisation studies at the IVP are ongoing. Based on studies carried out at the IVP, a tender for the construction for of a demonstration plant with capacity of 2.75-MGD (=12,500 m³/d) has recently been awarded to gain operational experiences with the technology. The IVP and the demonstration plant serve as part of a greater effort in developing innovative technology that may be applied at the upcoming Tuas WRP under the Phase 2 of the Singapore's Deep Tunnel Sewerage System (DTSS) project targeted for completion by 2024. The DTSS is an integral part of PUB's strategy to meet Singapore's long-term needs for water reclamation. The Tuas WRP is planned to tie in with the phasing out of Jurong WRP and Ulu Pandan WRP, and to consist of both domestic module and industrial module in the approximate proportion of 80% to 20%. The Tuas WRP is conceptualised to leverage on energy- and space-efficient MBR-RO process for NEWater production.

The authors acknowledge and thank research groups of Associate Professor Hu Jiang Yong and Associate Professor Ng How Yong from the National University of Singapore for the performance the LC-OCD analyses and Bernard Lim for the assistance in preparing some of the figures in this manuscript as part of his internship at PUB.