Ceramic membranes for reuse

In recent decades, water shortages in arid or heavily populated areas have caused water utilities to find alternative sources of drinking water. One source, wastewater, has been successfully integrated into the supply choices and other non-potable water uses for water utilities, but with a very high level of treatment if the water ultimately becomes a safe drinking water product. Using wastewater as a source of supply is referred to as reuse.

There are several non-potable uses of treated wastewater, such as irrigation, gray water supply for household toilets, laundry, and other uses, and industrial use. These can often be supplied with reuse water of a lower standard than of drinking water, but standards vary by country and region within a country (e.g., states in the United States).

The process trains for potable reuse applications, such as source augmentation and aquifer storage, typically contain reverse osmosis (RO) and an advanced oxidation process (AOP; typically ultraviolet light, UV, irradiation with hydrogen peroxide), because these processes provide excellent purification of the wastewater. With this treatment train, RO requires reliable and robust pre-treatment, which is typically with coagulation, possibly clarification, and some form of polymeric membrane filtration (either ultrafiltration, UF, or microfiltration, MF). While the combination of these processes achieves a very high finished water quality with multiple barriers to prevent pathogen break-through to the finished water, this combination of processes is expensive to operate, wastes a high percentage of water being treated, and in some cases, produces a water quality that surpasses what is actually needed.

An alternative treatment train is now being discussed and researched in the water reuse industry: coagulation, ozone, ceramic membrane, and biological granular activated carbon (BAC). This train has the potential to be less expensive to operate, waste less source water, and produce a finished water quality that is suitable for many applications. Likewise, when an AOP is necessary, for example to remove micro-pollutants such as endocrine disruptors, either hydrogen peroxide can be added to the ozone to make an AOP, or peroxide with UV can be installed downstream.

There are many ‘levels’ of reuse used throughout the world. Indirect potable reuse is the treatment of drinking water to a quality that allows it to be put into a water supply (e.g., reservoir, sand dunes, underground storage), such that it can undergo some natural ‘treatment’ while in the environment before become the source of drinking water at a water treatment plant. This is the most common form of reuse, because consumers are not always accepting of the alternative, which is direct potable reuse. Direct potable reuse is the treatment of wastewater followed by direct distribution to customers.

The paper will evaluate results of the Metawater ceramic membrane (i.e., 0.1 micron nominal pore size, 0.4 or 25 m² filtration surface area) from pilot testing secondary effluent at four different locations (i.e., two in Singapore, one in Australia and one in the Netherlands). These pilots had ozone and coagulation with ceramic filtration, and one of the pilots includes BAC after the membrane. Results from each location and membrane operation condition will be presented.

The aim of this paper is to compare the performance of the monolith ceramic membrane by Metawater for treating secondary effluent from different locations with different pre-treatments for reuse purpose. Results from different pilot studies have been reviewed and are compared herein.

Each pilot installation was slightly different in pre-treatment configuration, but each included a Metawater membrane. The pilots are summarized as follows:

• Bedok, Singapore pilot had a 25 m² membrane area with pre-coagulation and pre-ozone;

• Eastern Singapore, Singapore pilot had a 0.4 m² membrane area with pre-coagulation and pre-ozone (both aluminum sulfate and polyaluminum chloride were evaluated). Testing included inline coagulation, pre-ozone and coagulation, pre-ozone alone, and no pretreatment; Figure 1);

• Western Australia, Australia pilot had a 25 m² membrane with pre-coagulation and pre-ozone (Figure 2).

• Medemblik, the Netherlands has the same pilot equipment as the Eastern Singapore, Singapore pilot. An additional BAC system is installed downstream of ceramic membrane;

The membrane performance is evaluated by trans-membrane pressure (TMP) development at various fluxes. Generally, the critical flux is tested by running the membrane at different fluxes with a fixed amount loading until a backwash (BW). The critical flux test is varied at each site, depending on the TMP development that is observed in testing. Chemically enhanced backwashes (CEBWs) also varied for each site, but each includes at least a daily acid clean (i.e., pH < 2 with either hydrochloric or sulfuric acid) sometimes with 100 mg/L hydrogen peroxide, and also a CEBW with 100 mg/L chlorine, and the frequency varied based on the membrane's performance. For some of the pilots, clean-in-place (CIP) with traditional chemicals or ozone is possible. After each test run, the membranes are cleaned before beginning the next run.

There was a pilot study, which was conducted in the USA. The ceramic membrane pilot in the USA did show promise with operation at 200 lmh when having the combined ozone and coagulation pre-treatment. Unfortunately, the pre-ozone treatment system could not provide a consistently sufficient ozone residual to the membrane, due to the highly variable feed water quality (i.e., ozone demanding substances) upstream of the membrane; pilot testing results were inconclusive and not included herein.

One of the early pilot studies (Clement et al. 2009) of ceramic membranes for reuse was performed in 2007 and 2008, where coagulation and pre-ozonation was tested over about 12 months. Initial testing with coagulation showed that the polyaluminum chloride (PACl; 2 mg/L) resulted in a slower TMP build-up than ferric chloride (2 mg/L), so PACl use was continued. This coagulant comparison test was performed at 200 lmh with a backwash interval of 60 minutes, and the fouling rate for PACl was from about 17 to 35 kPa/day. In a subsequent test, ozone was applied to the membrane, at dosages of 0, 2, 4, and 6 mg/L. With no ozone, the fouling rate was 4 kPa/day (which is much lower than the previous tests), and with the ozone, at 2 to 6 mg/L, the fouling rate was 0.5 to 0.7 kPa/day, when operating at 200 lmh. This was a remarkable result, showing the self-cleaning action of ozone when applied to the membrane surface. Limited data are available about the cleaning regime and changes in source water quality to examine this more thoroughly, but it did show the potential for ozone with ceramic membranes for water and reuse applications.

A detailed summary of the pilot results was presented by Zheng et al. (2015), but pertinent details are shown here to allow for comparison of performance.

• At 100 lmh, the fouling rate was 0.5 kPa/day, and

• At 200 lmh, the fouling rate was still acceptable at 9.3 kPa/day.

Note that during run 2, at 150 lmh, there was a feed pump malfunction with air getting into the pipeline.

Subsequent filtration runs showed the impact of pre-treatment on the membrane TMP development (Figure 4):

• With no-pretreatment, the membrane operated at 100 lmh;

• With only coagulation, the membrane operated at 200 lmh;

• With ozone alone, the membrane operated at 200 lmh; and,

• With ozone and coagulation, the membrane operated at 300 lmh.

The filtration load until a BW was constant at 75 l/m². The data presented in Figures 4 and 5 is from the last two filtration terms (a filtration term defined as the filtration between two BWs) at the end of each run.

The TMP was high when there was no pretreatment or using the pre-ozone pretreatment. The TMP was much lower when having inline coagulation pretreatment and a combined pre-ozone and inline coagulation pretreatment. Figure 5 shows the influence of pretreatment on the specific flux. Without any pretreatment of with pre-ozone alone pretreatment caused a serious specific flux reduction. High specific flux can be maintained with inline coagulation pretreatment and a combined pre-ozone and inline coagulation pretreatment.

Figure 6 shows the average TMP as a function of treated volume in a relative long term (treating 2 m³ secondary effluents). It clearly illustrates that the membrane operation was stable at a flux of 300 lmh for the combined pre-coagulation and pre-ozone treatments; and it was stable at flux of 200 lmh for the pre-coagulant alone. The direct filtration and pre-ozone alone resulted in an unstable membrane operation.

The advantage of the combined pre-ozone and inline coagulation pretreatment as compared to inline coagulation alone was also clear. By introducing ozone, the membrane could be operated at high flux (i.e., 300 lmh), and at same time with high specific flux. Unfortunately, the ozone demand of this water was very high, and a dose of approximately 10 mg/L ozone was necessary to achieve an ozone residual on the membrane surface of 0.8 to 1.1 mg/L.

Operating with coagulant only was both technical and economic feasible. Therefore, it was further optimized. Figure 7 shows the results when running at 189 lmh with various coagulant dosages and the data suggest that the coagulant dosage could be optimized to yield very stable operation. The Metawater membrane feed channels can hold a high amount of solids, and coagulated solids act as a cake-layer on the membrane surface when coagulating under optimized conditions. This cake-layer has a high permeability and reduces the TMP increase. The layer has less affinity for the membrane surface and is backwashed away.

Data from Zheng et al. (2015) showed that the SDI of the membrane filtrate was low, and below the target level of 3, which is the common requirement for RO feed water, when the feed water to the membrane was chlorinated (see Figure 8). Ozone also helped to lower the SDI, and this is an important consideration if downstream process requires a low SDI.

For organics removal, the raw water DOC was in the range from 8.4 to 10.3 mg/L, and the filtrate was 6.7 to 8.0 mg/L according to liquid chromatography – organic carbon detection (LC-OCD) analysis. The fraction of biopolymers was rejected by the membrane, and this caused fouling. When using pre-coagulation and pre-ozone, the biopolymer fraction was removed well, meaning the biopolymer concentration in the membrane feed was low. At the same time, the properties of biopolymers were altered, as indicated by the low biopolymer rejection by the membrane (Zheng et al. 2015). This is the reason why the combined pre-ozone and pre-coagulation led to best membrane performance. To illustrate the concept, Figure 9 shows the change in the organics throughout the process of pre-ozone, pre-coagulation, and ceramic membrane. Figure 10 shows the TOC removal achieved by a higher coagulant dose during a coagulation test run at 189 lmh. This shows there is still a significant organic concentration in the filtrate.

A Metawater membrane pilot plant was refurbished to a single container size mobile pilot. The pilot is currently located at a secondary effluent treatment plant. The testing is still in progress and more data can be shared during the presentation.

Polyaluminum chloride (PACl) is used as the coagulant. The preliminary start-up trials were conducted,

• At 70 lmh, with coagulant dose of 6 mg/l Al³⁺ without pre-ozone;

• At 100 lmh, with coagulant dose of 6 mg/l Al³⁺ and pre-ozone.

Figure 11 shows the TMP and specific flux graphs when operating with coagulation only. It clearly illustrates that the membrane operation is unsustainable with coagulation as pre-treatment only.

Figure 12 shows the TMP and specific flux graphs when operating with coagulation and pre-ozone. The filtration load until a BW was constant at 50 l/m². By introducing ozone, the membrane could be operated at higher flux (i.e. ≥100 lmh), and at same time with stable specific flux. A dose of approximately 16 mg/L ozone was necessary to achieve an ozone residual on the membrane surface of 0.8 to 1.1 mg/L.

The result for operation with coagulation and pre-ozone is promising. The application of ozone upstream of the membrane results in lower rates of membrane fouling and higher sustainable flux rates. Ozone reacts with organic matters in the water and provides a ‘continuous’ in-situ cleaning to the membrane.

The trials are still on-going to find the maximum sustainable operating parameter.

A pilot evaluating the Metawater membrane for treating secondary effluent is currently under progress. This pilot is evaluating several different pre-treatments with biologically active carbon downstream of the membrane to determine if this treatment train can provide a suitable water quality for a reuse application. The study plan includes pre-treatment with ozone and coagulation, as well as bench-scale trials of ion exchange for pre-treatment. So far, it found that the ion exchange removed a large part of DOC and the results are promising.

Ceramic membranes can be used to treat secondary effluent for reuse applications, and the operational flux can be high (Table 1) as long as the pre-treatment is optimized for the water quality conditions. Stable operation can be achieved at high fluxes with only coagulation, but coagulation must be optimized and dosages are sometimes high to treat secondary effluent. When ozone is applied, a very high flux is possible and stable yet constant specific flux could be achieved; however, in practice the high ozone demand may be cost prohibitive. Further research is needed to observe the other water quality benefits that can be achieved, and to determine ways to optimize the process.

The authors wish to thank the pilot operations staff and Metawater.