Optimizing Microfiltration Membranes

May 13, 2013
Pilot study investigates effects of ozone pretreatment on PVDF membranes

About the author: Fredrick W. Gerringer is supervising engineer for Trussell Technologies Inc. Gerringer can be reached at [email protected] or 626.486.0560. R. Shane Trussell is president of Trussell Technologies Inc. Teresa Venezia is engineer II for Trussell Technologies Inc. Rajen Budhia is water resources engineer for West Basin Municipal Water District. Marc Serna is manager of engineering for West Basin Municipal Water District.


West Basin Municipal Water District is implementing ozone pretreatment of non-nitrified secondary effluent as part of an expansion of the Edward C. Little Water Recycling Plant (ECLWRF) in El Segundo, Calif. This project is intended to improve water production for existing polypropylene microfiltration (MF) membranes in the seawater intrusion barrier water treatment train at ECLWRF.

Pilot Testing

Previous pilot testing demonstrated that preozonation improved the performance of polypropylene MF membranes by reducing the fouling potential of effluent organic matter; however, additional MF capacity in the plant expansion is provided by polyvinylidene fluoride (PVDF) membranes. In preparation for full-scale design and operation of ozonation with the new MF units, West Basin commissioned a pilot study to examine the effects of ozone pretreatment on the operation of PVDF membranes.

Pilot testing involved two parallel treatment trains with PVDF MF modules (0.1 μm nominal pore size) and pilot equipment provided by Pall Corp. Membrane performance was compared using transmembrane pressure (TMP) data with a goal of 21 days of cumulative operation without a clean-in-place (CIP). One MF train operated with 10 mg/L of ferric chloride, a flux of 27 gal per cu ft per day (gfd), a daily chlorine enhanced flux maintenance (EFM) cleaning procedure, and a citric acid EFM procedure every 10 days. These test conditions were chosen because they followed the design of the full-scale PVDF MF process that would be installed at ECLWRF. 

The second MF train operated at 27 gfd and a transferred ozone dose of 4 to 16 mg/L, depending on the experiment. Ozone was added to the process flow through a venturi injector, with the dose being controlled by manual or automatic adjustments to ozone generator power. After ozonation, the water flowed through two vertical columns to provide contact time and to separate foam produced by ozonation.   

After the columns, the water entered an equalization tank before being pumped to the MF pilot unit. The treatment trains with MF pretreatment by ferric chloride and ozone were designated FeCl3-MF and O3-MF, respectively.

During pilot testing, water quality within each treatment train was characterized by fluorescence excitation emission matrix (EEM) spectroscopy. This analysis can identify soluble microbial products, fulvic-like substances and humic-like substances in EEM images produced from samples collected from the FeCl3-MF and O3-MF treatment trains.  Ozonation significantly reduced fluorescence intensity in all three regions, resulting in MF having little effect on the EEM spectra.

The EEM spectra of FeCl3-MF filtrate resembled O3-MF filtrate, but the most significant difference between these treatment trains can be detected by comparing MF backwash. Ozonation altered the chemical structure of effluent organic matter, thereby reducing EEM intensity in O3-MF feed and backwash samples. FeCl3-MF physically removed fluorescent effluent organic matter, lowering fluorescence in FeCl3-MF filtrate but increasing fluorescence in FeCl3-MF backwash. The higher fluorescence intensity in FeCl3-MF backwash compared with O3-MF backwash indicated more organic foulants were expected to be present on the MF module of the FeCl3-MF treatment train.

Transmembrane Pressure

TMP data from the two treatment trains during the first experiment revealed that the MF units performed differently: While there was a rise in the TMP data and a temporary shutdown caused by a mechanical problem, FeCl3-MF maintained good performance during 21 days of cumulative operation without a CIP. The final TMP was shown to be only 5 psi higher than the initial. The TMP trend for O3-MF showed a gradual increase during the first 12 days of operation, followed by a rapid rise after that. O3-MF automatically shut down when TMP exceeded 40 psi, prompting the researchers to trigger a manual chlorine EFM, followed by a manual citric acid EFM. These cleanings decreased TMP to 5 psi, allowing testing to resume. 

After a few days, TMP began rising again, triggering another automatic shutdown. O3-MF achieved 20 days of cumulative operation, only one day shy of the 21-day goal, with one chlorine EFM and one citric acid EFM. These data suggested that including two or more appropriately scheduled chlorine EFMs and citric acid EFMs could have been sufficient to achieve the 21-day goal without any shutdowns from high TMP.

A second set of chlorine and citric acid EFMs were performed before resuming filtration without ozone pretreatment. Comparing the TMP trend after the first chlorine and citric acid EFMs with the trend after the second chlorine and citric acid EFMs suggested ozonation was improving daily performance by reducing organic fouling.  

Considering TMP data for FeCl3-MF, there were two likely explanations for the rapid rise in TMP that triggered the automatic shutdowns: Ozonation either was not addressing a foulant that was removed by FeCl3, or it was creating a foulant not produced by FeCl3. Based on a significant improvement in clean water flux after a citric acid CIP of the O3-MF membrane, ozone oxidation chemistry and MF fouling mechanisms, manganese was determined to be the likely cause of fouling. Soluble manganese was present in the pilot plant influent, and ozone should have oxidized it to insoluble manganese that could cause deep pore fouling unresolved by backwashing.

Additional Testing

Later testing demonstrated ferric chloride addition after ozonation removed the foulant that caused TMP to rise. The probable explanation for this result was ferric chloride coagulated insoluble manganese, thereby capturing it in the cake layer, which was easily removed by backwashing. However, researchers continued to optimize the O3-MF treatment train to achieve acceptable performance without addition of ferric chloride.

The successful strategy behind this involved implementing automatic adjustments to the ozone dose based on the ultraviolet transmittance (UVT) at 254 nm after ozonation. A target range, or dead band, was set for the ozone effluent UVT. As influent water quality varied, the ozone effluent UVT would drift outside the dead band and trigger a computer to automatically adjust the power of the ozone generator to either increase or decrease the ozone dose. This approach allowed the O3-MF train to achieve greater than 30 days of operation at 30 gfd, a chlorine/caustic EFM every three days and a citric acid EFM every nine days. Transferred ozone doses ranged from 5 to 17 mg/L.

This pilot study demonstrated the successful operation of PVDF MF membranes after ozonation with the implementation of ozone dose control based on ozone effluent UVT. Further optimization would be required to determine the optimal ozone effluent UVT dead band that would provide adequate effluent organic matter oxidation to reduce organic fouling, while also minimizing the ozone dose to reduce costs associated with implementing this MF pretreatment strategy. Additional benefits, such as improved MF backwash water quality that could reduce coagulant doses required for recovering backwash water, also need
to be investigated. 

Download: Here