Filter Runs Collapse After a Storm
A 2.5 MGD conventional surface water plant draws from a reservoir and runs four dual-media filters with alum coagulation, flocculation, and sedimentation. On Monday the raw water is 4 NTU, 90 mg/L alkalinity as CaCO3, and pH 7.4, and the alum dose is 28 mg/L. Overnight a 1.6-inch storm pushes raw turbidity to 65 NTU; by Tuesday alkalinity is 42 mg/L and raw pH is 6.6. The operator leaves the alum dose at 28 mg/L and orders more frequent backwashing. Filter runs fall from 60 hours to 11 hours, combined filter effluent turbidity climbs from 0.08 to 0.42 NTU, and filter #3 reads 0.9 NTU on its last two 15-minute readings. The clearwell is at 70 percent and the distribution system needs full production by evening. No jar test has been run since Monday.
1. What is the most likely reason the filters are failing, and what are your first actions?
Read the worked answer
Diagnosis: the dose is fixed but the water changed.
- Particle load rose about 16-fold (4 to 65 NTU), so the coagulant demand rose with it.
- Alkalinity fell from 90 to 42 mg/L as CaCO3 and raw pH from 7.4 to 6.6. Alum consumes roughly 0.45 mg/L of alkalinity per mg/L of alum, so 28 mg/L of alum eats about 13 mg/L of alkalinity. Starting at 42 mg/L, the coagulation zone is now weakly buffered and the pH drops below the range where alum hydrolyzes well.
- Result: weak, pin-sized floc that passes the filters. Short runs plus rising effluent turbidity is carry-through, not a filter media problem.
First actions, in order:
- Run a jar test on today's raw water immediately; test a range of alum doses and a range of pH targets using caustic or lime. Do not simply raise the alum dose; at low alkalinity an overdose crashes pH further, worsens floc, and can carry soluble aluminum into the finished water.
- Confirm the coagulation pH with a sample taken after flash mix, not from the raw tap.
- Backwash filter #3 and inspect the bed for mudballs, media loss, or a disturbed surface.
- Check flash-mix and flocculator energy; a dose change without good initial mixing will not fix the floc.
- Consider reducing plant rate until the jar test result is in service, since lower loading buys time and improves settling.
2. Combined filter effluent is 0.42 NTU and filter #3 is at 0.9 NTU. How do you handle this from a compliance and operational standpoint?
Read the worked answer
Compliance view. Conventional filtration must keep combined filter effluent at or below 0.3 NTU in at least 95 percent of its measurements each month, and never above 1.0 NTU. A 0.42 NTU reading is an exceedance, not a data point to log and forget. Document it as a treatment upset and act.
Operational view. A single filter at 0.9 NTU is one reading away from the individual-filter threshold that triggers reporting and a filter self-assessment. Treat 0.9 as an alarm, not a curiosity.
Do now:
- Backwash filter #3 and confirm the backwash was effective (bed expansion, no media loss, no turbidity spike that lingers).
- If the filter has a filter-to-waste capability, use it; otherwise take #3 out of service if the remaining filters can carry the rate.
- Apply the jar-test-based dose and pH correction before the next filter run starts.
- Notify the plant superintendent and follow your state reporting obligation for the exceedance and for any individual filter that crosses 1.0 NTU on two consecutive 15-minute readings.
Do not respond by simply slowing the backwash triggers or widening the turbidity alarm.
3. How do you confirm the correction actually worked rather than just moving solids downstream?
Read the worked answer
Check the whole treatment train, not one number:
- Jar test repeatability: the new dose should produce a floc that settles clear in 10 to 15 minutes in the beaker before it is trusted in the plant.
- Settled water turbidity entering the filters should drop; if it does not, the problem is upstream (mixing, flocculation, or dose), and the filters will fail again within hours.
- Filter run length and headloss curve: runs should lengthen and headloss should climb smoothly. A run that ends with a turbidity rise but no headloss indicates carry-through, not loading.
- Post-backwash ripening: the turbidity spike after a backwash should be short and small. A long, high spike means the backwash is not cleaning the bed, or the bed has media problems.
- Finished water: confirm free chlorine residual and pH are back in target and check for aluminum carryover, which shows up as post-precipitation and dirty water complaints in the distribution system days later.
If only the turbidity number improves while run lengths stay short, you have masked the symptom and placed a heavier load on the filters later.