Filter Media Treatment
Abstract
The post sedimentation addition of polyaluminum chloride (PACl) was investigated as a means to enhance particle removal efficiency in rapid sand filtration. The process modification was evaluated in laboratory studies and at the Cornell Water Filtration Plant (CWFP). PACl was continuously metered into CWFP filter influent to increase concentrations by 0.06 to 4.2 mg/L (as aluminum) during the filter-to-waste stage of the filter operation cycle to accelerate filter ripening. Lower influent PACl concentrations ranging from 0.056 to 0.43 mg Al/L were also continuously applied during filtration. In comparison to a control filter that received no PACl addition, the ripening time required decreased with PACl dose, and the incremental improvement in particle removal during filtration increased with PACl dose. The addition of 0.056 mg Al/L of PACl (the lowest concentration tested) significantly reduced initial filter ripening time at the CWFP from 10 hours to 2.5 hours, and effluent turbidity in the test filter over the 77 hour filter run was lower than the control filter by an average of 17%. Incremental head loss increase caused by the PACl feed was dose dependent and was negligible for the lowest dosage tested.
Method of application in Drinking Water Treatment Plants
The filter media treatment process was conducted in one of the rapid filters at the CWFP. An electrical metering pump (PULSAtron Series MP) proportionally controlled by the CWFP flow-paced computer system was employed to inject PACl. The PACl injection port was in the settled water pipe before the water reached the test filter. The filter media treatment process included two stages: (1) A high concentration of liquid PACl was applied to settled water during the filter-to-waste stage of operation for 13 min. (2) After filter-to-waste when the filter was placed in operation, a lower concentration of PACl was continuously injected during filter run. PACl addition was terminated when the test filter head loss reached 115 cm (45 inch) according to the standard operating procedure for the facility, but the filter was not backwashed until the filtered water volume reached 1.5 million gallons, filter online time reached 90 hours, or filter water quality deteriorated. Different concentrations of PACl were applied at both stages to evaluate the effect of PACl on filter ripening and long term performance. Table 2 provides details of PACl concentrations used in each trial. The PACl concentrations reported at each stage were dependent on the plant flow rate during filter-to-waste (0.9 mm/s or 240 gpm) and the average flow rate during filtration.
A control filter in the CWFP was operated simultaneously with the test filter to compare the filter performance, head loss accumulation, and aluminum concentration in the filtered water.
Table 2. Summary of full-scale experiments
Trial PACl used during FTW *† (mg Al/L) PACl used in Filtration (mg Al/L) PACl metering time (hr) Total Al mass used (g/m2) Incremental head loss ** (cm) Average filtration rate (mm/s)
1 0 0 0 0 -10 1.28
2 4 0.43 32.5 59.6 80 1.11
3 4.2 0.19 65.9 52.2 65 1.13
4 0.23 0.19 60 44.9 71 1.09
5 0.14 0.12 70 34.9 63 1.15
6 0.06 0.056 77 19.2 30 1.23
- FTW: Filter to waste
† The test PACl was 5.5% Al by mass and the PACl density was 1260 kg/m3
**Incremental head loss = final head loss in test filter – final head loss in control filter
Results of Filter Media Treatment
Figure 2 shows the effluent turbidities of the test filter with filter media treatment and the control filter for an entire filtration run (trial 4). The PACl concentrations (as aluminum) applied are presented in Table 2. The effluent turbidity of the test filter dropped from 0.064 to 0.046 NTU in 2 hours while the turbidity of the control filter increased initially and then declined to 0.054 NTU after 13 hours. Thus, the time for ripening of the test filter was significantly decreased through PACl addition. The filter performance of the test filter was better than the control filter by an average of 23% during the entire filter run. Operation of the CWFP is constrained to daytime intervals. Arrows in Figure 2 above the line for the control filter show the points in time when operation of the plant stopped, and the run time shown on the x-axis does not include times when the plant was not operating. Turbidity spikes appear in both filters on each occasion that operation resumed, and were likely due to particle detachment from the media grains in response to the onset of flow. Turbidity spikes in the test filter were significantly less than the control filter, and the recovery time of the test filter to the previous effluent turbidity was significantly shorter than for the control filter. Between the 55 to 56th hour a large spike occurred in settled water turbidity (data not shown) because of a malfunction of the CWFP plant PACl pump (the PACl pump for the test filter was still functioning). This influent turbidity spike resulted in effluent turbidity spikes from both filters. However, the transient increase in turbidity in the test filter was significantly smaller than the control filter. Collectively these results indicate that the filter media treatment process can significantly accelerate filter ripening, reduce the turbidity spikes when restarting the plant, effectively improve the filter performance, and make the filter less sensitive to fluctuations in influent turbidity. After 60 hours of operation addition of PACl to the test filter was terminated. Improved performance by the test filter (i.e., effluent turbidity below that of the control) continued after the PACl feed was terminated, but particle removal slowly began to approach that of the control. This result demonstrated that the filter media treatment process modified the filter media to enhance the ability to capture particles and that sustained improvement was dependent on continued addition of PACl as a filter aid. 
Figure 2. Effluent turbidities of the test and control filters for trial 4. Arrows above the control filter line show points at which daytime plant operation was terminated.
Advantages of Filter Media Treatment
The filter media treatment process can significantly accelerate filter ripening, reduce the turbidity spikes when restarting the plant, effectively improve the filter performance, and make the filter less sensitive to fluctuations in influent turbidity. In addition, it could reduce the coagulant dose used during flocculation.
The incremental cost of filter media treatment was estimated to be 4.70E-4 USD/m3 of filtered water based on the liquid PACl cost of 0.5 USD/L (2 USD/gallon) and an application rate of 0.06 mg/L as aluminum (slightly above the lowest dose used in this study). Filter media treatment can be accomplished with a single metering pump for an entire water treatment plant with all filter influent receiving a constant dose. Installation of the required PACl injection line from the existing PACl stock tanks to a point somewhere between the sedimentation tank effluent and the filter influent should be relatively straight forward. This economic analysis does not consider the possibility of using improved particle capture in filters as an opportunity to reduce the coagulant dose used before sedimentation.
Facilities Currently Using the Technology
The CWFP has implemented the PACl filter media treatment process as its normal operating procedure.
For more information
For more information please contact Po-Hsun Lin
Documents
Enhanced Filter Performance by Fluidized-Bed Pretreatment with Al(OH)3(am): Observations and Model Simulation
Po-Hsun Lin, Leonard W. Lion, and Monroe L. Weber-Shirk, Journal of Environmental Engineering 1, 389 (2011).
Enhanced Particle Capture through Aluminum Hydroxide Addition to Pores in Sand Media
Po-Hsun Lin, Leonard W. Lion, and Monroe L. Weber-Shirk, Journal of Environmental Engineering 1, 284 (2011).
Comparison of the Ability of Three Coagulants to Enhance Filter Performance
Po-Hsun Lin, Leonard W. Lion, and Monroe L. Weber-Shirk. Journal of Environmental Engineering 137, 371 (2011).