One Filter Breaks Through Early While the Rest Run Long
A plant has six rapid-rate filters, each 20 ft by 30 ft, with 30 inches of anthracite over 12 inches of sand. Average flow is 6.0 MGD, peak day flow is 8.0 MGD. Filter 3 has terminated on turbidity breakthrough at 0.35 NTU after 20 hours on three consecutive runs; the other five run about 50 hours to the 8 ft headloss limit. Backwash is set at 15 gpm/ft2 for 8 minutes with no air scour, using finished water that is 8 degrees C in December. Inspection of filter 3 shows mudballs and a slightly mounded surface but no visible media loss. Combined filter effluent is 0.12 NTU.
1. Calculate the filtration rate at average flow, at peak day flow, and at peak day flow with one filter out of service. Based on those numbers, is hydraulic loading the explanation for filter 3's behavior?
Read the worked answer
Filtration rate = gpm / media area.
Total media area: 6 filters x (20 ft x 30 ft) = 3,600 ft2
- Average: 6,000,000 gpd / 1,440 = 4,167 gpm; 4,167 / 3,600 = 1.16 gpm/ft2
- Peak day: 8,000,000 / 1,440 = 5,556 gpm; 5,556 / 3,600 = 1.54 gpm/ft2
- Peak with one filter out: area = 3,000 ft2; 5,556 / 3,000 = 1.85 gpm/ft2
Interpretation: all three rates are well under a typical 3-4 gpm/ft2 design limit for anthracite-over-sand, and even the worst case (1.85) is only about 60 percent of design. Flow is not the cause.
The real signal is the contrast between filter 3 at 20 hours and the others at 50 hours on headloss. Filters 1, 2, 4, 5, and 6 are carbon-copying each other, which means the influent water and the flow split are fine. One filter behaving differently points at the filter itself: media condition, underdrain, backwash distribution, or a mechanical problem like a sticking valve or an unseated air release.
2. Evaluate the backwash program. Compute the backwash volume and its share of production for a 20-hour run at 1.16 gpm/ft2, and explain why 15 gpm/ft2 at 8 degrees C may be inadequate.
Read the worked answer
Backwash volume:
- Rate: 15 gpm/ft2 x 600 ft2 = 9,000 gpm
- Volume: 9,000 gpm x 8 min = 72,000 gal per wash
Production per 20-hour run on filter 3:
- 1.16 gpm/ft2 x 600 ft2 = 696 gpm
- 696 x 20 x 60 = 835,000 gal
Backwash share: 72,000 / 835,000 = 8.6 percent of production. That is high (typical target 2-4 percent); the filter is being washed too often because it is failing early, which is a symptom, not the fix.
Why 15 gpm/ft2 may be short at 8 degrees C:
- Cold water is more viscous than warm water. The same backwash rate produces less bed expansion in December than in July.
- Anthracite needs roughly 20-25 percent expansion to fluidize and scrub the bed. At low temperature, 15 gpm/ft2 can give well under that.
- An under-expanded bed does not release attached floc or the agglomerations that become mudballs. Media then stays clumped, flow short-circuits through the clumps, and particles pass through before the rest of the bed is loaded. That is exactly the pattern: early turbidity breakthrough, a mounded surface, and mudballs on one filter, on top of a run that ends at 20 hours regardless of headloss.
Practical check: measure percent bed expansion and backwash turbidity profile (peak turbidity should be reached and then decline). If expansion is under 20 percent, raise the rate incrementally and watch for media loss, or add air scour, rather than lengthening the wash.
3. Give the diagnostic sequence you would follow on filter 3 before changing any setpoints, and state what you would do if the media inspection shows the anthracite layer has thinned.
Read the worked answer
Sequence:
- Backwash it and watch it. Observe the wash from above: boil pattern, dead spots, media boiling over the troughs, and the turbidity of the washwater over time. Uneven boil = backwash distribution problem (nozzle, lateral, or support gravel). Localized boiling = cracked or shrunken media, or an underdrain defect.
- Measure bed expansion at the current rate and at two higher rates, and record water temperature with each reading. This gives you a rate-versus-expansion curve to set the winter rate from.
- Core the media. Measure anthracite and sand depth against the original 30/12 inches, check for mudballs, and check whether media has migrated into the support gravel. Media loss without visible washout during the visit usually means it is leaving through a broken nozzle or it left during an earlier event.
- Check the valves and controls: inlet/effluent valve throttling, rate-of-flow controller, and whether the filter is being returned to service too fast after backwash. A fast start drives floc deep into the bed and can produce early breakthrough.
- Confirm influent quality is the same as the other filters (particle counts, settled turbidity at that filter's inlet), to rule out a piping or split problem.
If the anthracite has thinned:
- Take the filter out of service until the bed is restored; a thin bed cannot provide the 3-log credit the state granted for that unit, and the remaining sand depth may not meet the design L/d ratio.
- Fix the cause (nozzle, gravel, backwash rate) before adding media, or the new media will leave the same way.
- After refilling, disinfect, slowly bring the filter back to rate, and confirm a clean backwash and a full-length run before returning it to normal service; document the media depth and date in the filter records.