Investigation of E. coli and Virus Reductions Using Replicate, Bench-Scale Biosand Filter Columns and Two Filter Media

Abstract

The biosand filter (BSF) is an intermittently operated, household-scale slow sand filter for which little data are available on the effect of sand composition on treatment performance. Therefore, bench-scale columns were prepared according to the then-current (2006-2007) guidance on BSF design and run in parallel to conduct two microbial challenge experiments of eight-week duration. Triplicate columns were loaded with Accusand silica or crushed granite to compare virus and E. coli reduction performance. Bench-scale experiments provided confirmation that increased schmutzdecke growth, as indicated by decline in filtration rate, is the primary factor causing increased E. coli reductions of up to 5-log10. However, reductions of challenge viruses improved only modestly with increased schmutzdecke growth. Filter media type (Accusand silica vs. crushed granite) did not influence reduction of E. coli bacteria. The granite media without backwashing yielded superior virus reductions when compared to Accusand. However, for columns in which the granite media was first backwashed (to yield a more consistent distribution of grains and remove the finest size fraction), virus reductions were not significantly greater than in columns with Accusand media. It was postulated that a decline in surface area with backwashing decreased the sites and surface area available for virus sorption and/or biofilm growth and thus decreased the extent of virus reduction. Additionally, backwashing caused preferential flow paths and deviation from plug flow; backwashing is not part of standard BSF field preparation and is not recommended for BSF column studies. Overall, virus reductions were modest and did not meet the 5- or 3-log10 World Health Organization performance targets.

Keywords: household water treatment and safe storage (HWTS); intermittent; point-of-use (POU); slow sand filter.

Figure 1

Cross-section of bench-scale columns used in Column Test No. 1. Six columns of this design were used, three loaded with Accusand silica and three with crushed granite media.

Figure 2

Cross-section of bench-scale columns used in Column Test No. 2. Six columns of this design were used, three loaded with Accusand silica and three with crushed granite media.

Figure 3

Sieve analysis of backwashed Accusand and granite media from Column Test No. 2.

Figure 4

Initial daily filtration rate during Column Test No. 1. Accusand-loaded filter columns A1-A3 (a); and Granite-loaded filter columns G1-G3 (b). Spikes in filtration rate are due to either intentional removal (cleaning) of the schmutzdecke (A1 day 39, A2 day 29, G1 day 47) or unintentional disturbance of the schmutzdecke during filter loading (A3 day 17, G3 day 41).

Figure 5

Initial daily filtration rate during Column Test No. 2. Accusand-loaded filter columns A1-A3 (a); and Granite-loaded filter columns G1-G3 (b). Spikes in filtration rate are due to intentional removal (cleaning) of the schmutzdecke (A1 day 49; A2 day 37; A3 day 37; G1 days 37 and 56; G2 days 29 and 56; G3 days 23 and 56).

Figure 6

E. coli reductions for (a) Accusand-filled columns; and (b) granite-filled columns in Column Test No. 1 for samples collected throughout the eight-week experiment. N = 10 for G3; 11 for A1, A2 and G1; 12 for A3; 13 for G2.

Figure 7

E. coli reductions for (a) Accusand-filled columns; and (b) granite-filled columns in Column Test No. 2 for samples collected throughout the eight-week experiment. N = 17 for G2; 18 for A1; 19 for A2; 20 for G1, G3 and A3.

Figure 8

E. coli reductions in Column Test No. 1 and Column Test No. 2 organized by bins of normalized filtration rate where Q_I is the initial daily filtration rate on each day of charging the column and Q_I,o is the comparable initial filtration rate on the first day. N for Column Test No. 1: 14 for <0.2, 13 for 0.2–0.8, 23 for >0.8. N for Column Test No. 2: 15 for <0.2, 34 for 0.2–0.8, 30 for >0.8.

Figure 9

MS2 reductions for (a) Accusand media columns; and (b) granite media columns in Column Test No. 1 for samples collected throughout the eight-week experiment. N = 7 for G3; 8 for A1; 9 for G1, G2, A2 and A3.

Figure 10

Echovirus 12 reductions for Accusand media columns and granite media columns in Column Test No. 1 for samples collected throughout the eight-week experiment. N = 11 for Accusand and 11 for granite media.

Figure 11

MS2 reductions for (a) Accusand media columns; and (b) granite media columns in Column Test No. 2 for samples collected throughout the eight-week experiment. N = 13 for G2; 14 for A1–A3; 15 for G1 and G2.

Figure 12

PRD-1 reductions for (a) Accusand media columns; and (b) granite media columns in Column Test No. 2 for samples collected throughout the eight-week experiment. N = 13 for G2; 14 for A1–A3; 15 for G1 and G2.

Figure 13

MS2 reductions for (a) Accusand media columns; and (b) granite media columns in Column Tests No. 1 and No. 2 organized by bins of normalized filtration rate, where Q_I is the initial daily filtration rate on each day of charging the column and Q_I,o is the initial filtration rate on the first day. N for Accusand: 16 for <0.2, 17 for 0.2–0.8, 17 for >0.8. N for granite: 12 for <0.2, 25 for 0.2–0.8, 11 for >0.8.

Figure 14

PRD-1 reductions for (a) Accusand media columns and (b) granite media columns in Column Tests No. 2 organized by bins of normalized filtration rate where Q_I is the initial daily filtration rate on each day of charging the column and Q_I,o is the initial filtration rate on the first day. N for Accusand: 6 for <0.2, 10 for 0.2–0.8, 9 for >0.8. N for granite: 6 for <0.2, 15 for 0.2–0.8, 6 for >0.8.