Cost-effectiveness and efficiency are important factors in selecting a treatment process that will comply with both current and anticipated water quality regulations. One cost-saving process that shows great promise for treating high-quality water sources is extreme/high-rate serial filtration (EHRSF). Pilot testing was conducted on EHRSF processes by Kennedy/Jenks Consultants, at the Humboldt Bay Municipal Water District (HBMWD) in northern California. The pilot tests demonstrated that EHRSF could meet the requirements of the Surface Water Treatment Rule and provide treatment equivalent to conventional methods, while operating at higher loading rates than are currently permitted.
Serial filtration is a two-stage treatment process well-suited for high quality water supplies with low-turbidity. The process uses contact clarifiers (roughing filters) in the pretreatment stage that flocculate and clarify the raw water. This process is then followed with polishing filters. The two-stage treatment process cuts construction costs because it uses compact unit processes that operate at significantly higher surface loading rates than conventional sedimentation clarifier processes.
The serial filtration process also requires far less coagulant chemical and lower operating and maintenance costs. In addition, the District could significantly lower costs by operating a serial filter system at higher than approved surface loading rates.
With these advantages in mind, Kennedy/Jenks conducted pilot testing to determine if the treatment process would produce a satisfactory water quality when the contact clarifiers and filters were operated at surface loading rates that were 20 to 50 percent higher than the currently-approved rates.
High Quality Water Source
HBMWD's source water is supplied by a 25 mgd Ranney Collector System, (see pg. 18), that collects groundwater which has flowed through the alluvium below the bottom of the Mad River. Since the source water is prefiltered through gravel and sand in the riverbed, turbidity is typically quite low, and the water contains little organic carbon that can react with chlorine to form disinfection by-products. Turbidity is normally less than one nephelometric turbidity unit (NTU) for most of the year. However, during winter high-flow conditions, the turbidity in the Ranney Collector source water has exceeded 20 NTU.
When the California Department of Health Services (DHS) found that turbidity and temperature changes in the Ranney Collector filtered water supply seemed to correlate with turbidity and temperature fluctuations occurring in the Mad River's surface water, DHS reclassified the Ranney Collector source water as groundwater under the direct influence of a surface water. This classification meant that HBMWD would have to filter its source water to comply with the Surface Water Treatment Rule (SWTR).
After negotiating with DHS, the District agreed to build a new water treatment plant by December 1998. The district hired Kennedy/Jenks Consultants to evaluate District options and recommend a cost-effective treatment process to comply with the SWTR requirements. In 1991 the estimated cost of building a conventional 21 mgd water treatment facility (with flocculation, sedimentation pretreatment and filtration) was about $18.3 million. According to a preliminary design study, the most cost-effective alternative for treating this type of low-turbidity source water was serial filtration, costing over $6 million less than a conventional water treatment plant.
Serial Filtration Processes
Serial filtration is an alternative treatment process for treating potable water supplies. In California, two types of serial filtration processes have been classified by DHS as "equivalent to conventional treatment processes." Conventional processes (including flocculation, sedimentation and filtration) are granted a 2.5-log removal credit for Giardia cysts and a 2-log removal credit for enteric viruses.
Both DHS-approved serial filtration processes have been granted "equivalence to conventional treatment" when operated at surface loading rates of 10 gpm/ft2 for the contact clarifiers and 5 gpm/ft2 for the filters. DHS also considers serial filtration treatment plants with contact clarifiers operating at greater than 10 gpm/ft2 and filters operating at less than 6 gpm/ft2 to be "equivalent to a direct filtration process." These processes are given a 2-log removal credit for Giardia cysts and 1-log removal credit for enteric viruses.
Kennedy/Jenks recommended an evaluation of extreme/high-rate serial filtration because they had prior experience with high-rate serial filtration processes and it offered substantial cost savings to the district. Potential savings amounted to $1 million compared to currently approved high-rate serial filtration processes, $3.5 million compared to direct filtration treatment, and $6.5 million compared to conventional treatment.
The pilot test evaluated high-rate serial filtration operating at the currently-approved surface loading rates and at extreme/high-rates. Since the proposed loading rates exceed those currently accepted by DHS for serial filtration systems, the pilot scale study was needed to determine the removal credits (for Giardia cysts) that would be granted to the EHRSF treatment process.
At the request of DHS, the pilot study also investigated a serial filtration process, provided by a third manufacturer, to increase the number of qualified equipment suppliers. This manufacturer's equipment had not been previously accepted by DHS as equivalent to either conventional treatment or direct filtration processes.
Pilot Testing Six High-Rate Systems
A total of six alternative high-rate serial filtration treatment systems were pilot tested under "worst case conditions," including critical low temperature and high turbidity. Three different types of proprietary contact clarifier designs were tested side-by-side, with two polishing filter designs. Using the District's Ranney Collector source water, the pilot plant study team evaluated the EHRSF process with the contact clarifiers operating at 12 to 15 gpm/ft2 and filters operating at 6 to 10 gpm/ft2 on high turbidity blended source water. The first pilot tests evaluated high-rate contact clarification (operating at 12 gpm/ft2) with conventional rate multi-media filters (operating at 6 gpm/ft2) and high-rate dual-media filtration (operating at 8 gpm/ft2).
After pilot testing had successfully demonstrated that the serial filtration systems could meet treatment objectives at the high rates classified by DHS as equivalent to "direct filtration," the project team then proceeded to operate four serial filtration systems at "extreme/high rates." Two contact clarifiers were operated at 15 gpm/ft2 along with two multi-media filters at 7.5 and 9 gpm/ft2 and two dual-media filters at 10 gpm/ft2, using low-turbidity source water. These tests evaluated the effectiveness of the EHRSF process in treating the high quality source water normally provided by the Ranney Collectors. To provide a basis for comparing performance, one contact clarifier was operated at 12 gpm/ft2, one multimedia filter at 6 gpm/ft2, and one dual media filter at 8 gpm/ft2, in parallel to the four EHRSF systems.
Grab samples of blended source water, clarified water and filtered water were collected every four hours, and analyzed by conducting particle counts and measuring turbidity. The study team took daily water samples to analyze for alkalinity, hardness, temperature, dissolved oxygen, coliforms and heterotrophic bacteria. A sample was also collected to evaluate the impact of the treatment process on the corrosivity of the water for compliance with the Lead and Copper Rule.
During the last five days of the investigation, a major storm dropped more than 8 in. of rain and raised the turbidity in the Mad River to 600 NTU (peak) and the turbidity in the Ranney Collector source water to 5 NTU (maximum).
For three days during this period, the six serial filtration systems were operated with source water taken directly from the Ranney Collectors with average source water turbidities ranging from 0.3 to 4.5 NTU. To test the serial filtration system's performance under peak turbidity conditions, the final pilot run used a blend of Ranney Collector source water and raw Mad River water with an average turbidity of 24.5 NTU.
Evaluating the Results
The pilot study showed that a high-quality water supply, such as found in the HBMWD's Ranney Collectors, can comply with EPA's SWTR and related California DHS regulations, when it is treated using the high/extreme-rate serial filtration process. The pilot plant evaluation verified that EHRSF processes could remove particles and turbidity as well as the serial filtration processes that are accepted by DHS "as equivalent to conventional treatment" in treating water with turbidities ranging from 0.2 NTU to 25 NTU. As required by the SWTR, the process removed more than 99.7 percent (2.5 log) of Giardia cyst-sized particles (4p;10 µm) when proper coagulant doses were applied. Treated water turbidities averaged less than 0.1 NTU.
When the contact clarifiers and the two polishing filters were operated at both high and extreme/high surface loading rates, all three clarifiers performed comparably in removing particles when used with both the multi-media and the dual-media filter designs. In addition, the performance was not significantly different when the serial filtration systems used blended source water and were operated at higher surface loading rates, and when these processes were operated at the lower surface loading rates currently approved by DHS.
Each of the six serial filtration systems produced an average filtered water turbidity of less than 0.1 NTU for at least 22 hours in seven out of the nine pilot runs and satisfactorily removed over 99.7 percent of the particles in the 4p;10 µm particle size range.
These results were achieved whether high-turbidity blended source water was used or water was taken directly from the Ranney Collectors. The reduction of 4p;10 µm size particles from the total treatment process (Ranney Collectors plus serial filtration) ranged between 2.5-log (99.7 percent) and 5.7 log (99.9997 percent) removal. The higher removal rates correlated with higher turbidity in the raw Mad River water.
The Ranney Collectors provided most of the log reduction for the low-turbidity raw water during the test period. (See Table 1, pg. 19.) The average count of 4p;10 µm size particles in the raw Mad River water ranged between 9,400 and 1,630,000 particles per milliliter (ml); the Ranney Collector source water particles ranged between 40 and 950 particles/ml; and filtered water counts ranged between an average of 2.5 and 3.3 particles/ml during this ten day period. When particle counts in the Ranney Collector source water were low, the average particle removal by the serial filtration processes was also low, ranging between 0.93- and 1.47-log. The multi-media filters performed slightly better than the dual-media filters during these runs. However, the overall serial filtration process performance was nearly the same when the three contact clarifiers were operated with either the multimedia filters or the dual-media filters. The removal rate of Giardia and Cryptosporidium size particles increased as raw water turbidity and particle counts increased. Therefore, this reasons that the Ranney Collectors provide a significant barrier to changes in varying raw water quality and help ensure that the quality of the source water for this cost-effective serial filtration process remains relatively high and consistent.
The successful operation of the pilot filters at 8 to 10 gpm/ft2 and 7.5 to 9 gpm/ft2, established that both the dual-media filter and multi-media filter designs, can produce filtered water turbidity below 0.5 NTU all of the time.
They can also produce an average daily filtered water turbidity below 0.1 NTU, when operated at surface loading rates that are higher than currently accepted by DHS. The higher loading rates mean that the filters can be up to one-third smaller than conventional rate filters, resulting in significant cost savings. It was possible to achieve filter runs lasting more than 48 hours for both filter media designs at the high surface loading rates.