Filter Runs Collapse After a Storm - Coagulant Dose vs. pH
A 12 MGD conventional alum coagulation plant is running 9 MGD. A two-inch rain pushed raw turbidity from 3 to 45 NTU and raw alkalinity from 90 to 140 mg/L as CaCO3. Alum feed was left at 25 mg/L. Settled turbidity is now 6 NTU against a goal of less than 2, filter runs have fallen from 40 hours to 11 hours, and finished turbidity is climbing past 0.30 NTU. Raw pH is 7.6; settled pH is 6.3. Filters are being returned to service as soon as the turbidity spike clears. The shift operator wants to jump the alum dose to 40 mg/L immediately and add a polymer at the same time.
1. What do the settled-water and filter data tell you, and why is jumping straight to 40 mg/L of alum risky without checking pH and alkalinity?
Read the worked answer
What the data say
- Settled turbidity of 6 NTU means floc is not forming or not settling; the filters are being forced to act as clarifiers. That is exactly why runs fell from 40 h to 11 h.
- The raw load roughly tripled in particles while the coagulant dose stayed flat, so the plant is underdosed relative to the new load.
Why a blind jump is risky
- Alum consumes alkalinity. A workable rule of thumb is about 0.45–0.5 mg/L of alkalinity (as CaCO3) per 1 mg/L of alum. Going from 25 to 40 mg/L adds roughly 7 mg/L of alkalinity demand on top of what is already being consumed.
- Settled pH is already 6.3. Pushing more alum without lime or caustic can drive pH below the range where aluminum hydroxide floc is stable, giving you soluble aluminum carryover, poorer floc, and possibly a worse settled turbidity than you started with.
- Adding polymer at the same time as a big dose change destroys your ability to tell which change did what.
Judgement Correct the dose, but pay for it in alkalinity and pH at the same time, and confirm the target dose in a jar test before committing the plant to it.
2. Lay out your next 24 hours of actions in priority order.
Read the worked answer
- Put filters on filter-to-waste where piping allows, so off-spec water is not pushed to the clearwell while the process is unstable. Keep them there until finished turbidity is consistently under target, not just under the spike.
- Run a jar test matrix now: alum at roughly 25, 30, 35 and 40 mg/L, each with and without caustic/lime to hold settled pH around 6.8–7.2. Pick the lowest dose that gives settled turbidity under 2 NTU at acceptable pH.
- Change one variable at a time, in steps no larger than about 5 mg/L, and hold 2–4 hours between steps so settled water responds.
- Monitor hourly: raw and settled turbidity, settled pH, alkalinity, filter headloss and run time, finished turbidity. Track alkalinity because it is your early warning that the coagulant is consuming your pH buffer.
- Do not return a filter to service on the basis of the turbidity spike clearing alone - use finished turbidity plus headloss trend plus (if available) particle counts. A filter that was recently overloaded can pass particles after the visual spike is gone.
3. What single process-control number would you use to decide the plant is stable again, and why that one?
Read the worked answer
Settled water turbidity, held under 2 NTU with a settled pH inside the coagulant's optimum window, is the number that matters. It is the last measurement before the filters, so it tells you whether the chemistry is working rather than whether the filters are absorbing the failure.
Reasoning: finished turbidity can look acceptable for hours while filters are simply storing up the problem - you will see that as shortening run times and rising headloss. If you watch finished turbidity only, you find out after the filters are already compromised. Watch settled turbidity plus filter run length trend and you get the same information 8–12 hours earlier.