Filter 2 Breaks Through After a Coagulant Cut
Riverview Water Treatment Plant is a 4.0 MGD conventional plant - alum plus anionic polymer, three dual-media filters - serving 11,000 people. After a weekend storm, raw alkalinity fell from 62 to 34 mg/L as CaCO3 and raw turbidity rose from 4 to 38 NTU. To protect finished-water pH, the operator on duty cut alum from 45 to 30 mg/L at 07:00 and raised polymer to 0.4 mg/L. Settled water turbidity is now 3.1 NTU (normally 1.2). Combined filter effluent is 0.08 NTU, but Filter 2's individual effluent climbed from 0.05 to 0.28 NTU in four hours. Filter 2 headloss is only 2.1 ft, well below the 4.5 ft it normally reaches before a backwash, and it was backwashed six hours ago. Finished water: pH 7.4, free chlorine 1.05 mg/L, alkalinity 28 mg/L.
1. Filter 2 is passing something. What is the most likely mechanism, and what would you check before you change the alum dose again?
Read the worked answer
Mechanism: weak floc, deep-bed penetration - not a damaged or blinded filter.
The combination of low headloss and a rising effluent turbidity shortly after a backwash points to small, poorly aggregated particles traveling deep into the bed and breaking through, rather than a surface mat blinding the media. Media loss is a weaker explanation: that would usually show as a sudden jump, a shortened run, and a measurable drop in bed depth.
The real problem is upstream. A 33% alum cut against a 45% drop in raw alkalinity did not preserve the coagulant-to-alkalinity ratio - it worsened it, and the 0.4 mg/L polymer cannot build dense floc on its own. Settled water at 3.1 NTU confirms the filters are being asked to remove particles the coagulation step failed to make removable.
Check, in this order:
- Settled water turbidity and particle counts since 07:00 - is the trend still worsening?
- A jar test on the current raw water, not the pre-storm water.
- Streaming current or zeta potential, if you have it.
- Filter 2 effluent particle counts - they rise before turbidity does.
- The last backwash record: bed expansion %, air scour, turbidity peak and time to clear.
- Media depth and condition in Filter 2 versus the other two.
- Whether Filter 2 is carrying more than its share of plant flow because of a throttled valve elsewhere.
Do not simply backwash again. That resets the turbidity curve for a few hours and hides a chemistry problem that is getting worse.
2. You have until the 18:00 demand peak. Lay out what you actually do, in order.
Read the worked answer
- Jar test immediately on the current raw water. Run alum at roughly 38, 45, and 52 mg/L, with and without lime or soda ash to bring alkalinity back toward 50–60 mg/L. Do not set the dose from a table or from the pre-storm history.
- Plan for alkalinity, not just coagulant. Restoring alum will consume more of the 34 mg/L of alkalinity and pull pH down further. Finished water is already at 7.4 and 28 mg/L alkalinity, which is a corrosion risk in the distribution system. Feed lime or soda ash with the coagulant and pick a post-filtration pH you can hold.
- Verify the dose that is actually being applied. A one-shift cut this large is often a feeder problem (calibration, specific gravity, dilution, polymer make-down age) rather than a decision. Check it before you argue about chemistry.
- Protect the filtrate. Consider taking Filter 2 out of service and running on two filters at a reduced plant rate, or placing it on filter-to-waste until its effluent turbidity is stable. Two well-run filters beat three mediocre ones.
- Watch the chlorine. Extra particles and organics raise chlorine demand; a 1.05 mg/L residual can disappear quickly. Confirm residual downstream of the clearwell, not at the filter gallery.
Document the sequence and the reasoning. If the plant goes out of compliance tonight, the record of what you tried and when matters as much as the outcome.
3. At 18:00 Filter 2 records 1.2 NTU on two consecutive 15-minute readings. What is required, and what is not?
Read the worked answer
Required:
- This is an individual filter effluent exceedance. Report it to NJDEP within 24 hours. The reporting obligation is triggered by the recorded readings and is not discretionary - do not wait to see whether the next shift fixes it.
- Preserve the data: the raw turbidity record, the continuous IFE trace, the backwash log, and the settled water data for the whole day.
- Run a filter profile - turbidity and particle counts through the bed depth - to show where the floc is penetrating, and be ready for a self-assessment requirement if this recurs in a following month. Confirm the exact trigger and deliverables with NJDEP.
Not automatic:
- This single event does not by itself put you outside the combined filter effluent limit. That limit is judged on the combined filtrate over the month at the 95th percentile, and 0.08 NTU is fine. If the combined value starts climbing too, you have a treatment technique problem, not just one bad filter.
- Backwashing is not a compliance strategy. The exceedance happened on the record and the report goes in.
Between the exceedance report and the self-assessment there is a real judgment call about how fast you escalate. Given an obviously upset coagulation process, escalate now rather than at the next quarterly review.