Alum Overdose After a Storm Knocks Out Alkalinity

Chemistry. Alum destroys alkalinity at roughly 0.45–0.5 mg/L as CaCO3 per mg/L of alum. At 36 mg/L, demand is about 16–18 mg/L; the raw water only has 26 mg/L. That leaves under 10 mg/L of buffering, so pH becomes unstable and drops - the finished pH of 6.4 confirms it. Alum also produces roughly 0.26 mg/L of Al(OH)3 solids per mg/L of alum, so doubling the dose roughly doubles the sludge and solids load reaching the filters.

Diagnosis of the floc. The 0.28 mg/L aluminum in settled water is mostly particulate, not dissolved - check by filtering the sample; if filtered aluminum is low, the metal is riding on floc that never settled. Combined with thin, light floc and pin floc over the weirs, this points to poor coagulation/floc formation from a dose that is no longer matched to the water, not to pH-driven aluminum solubility (which would show up mainly below about pH 5.5).

First change. Do not add more coagulant. Run a jar test on the new raw water across a dose range, with and without alkalinity addition, and pick the dose that gives the best settled turbidity. Add lime or caustic to hold settled pH near the optimum and leave at least 20–25 mg/L alkalinity. Consider a coagulant aid polymer for floc strength, and check the clarifier sludge blanket and withdrawal rate before blaming the filters.

Mechanisms. Two things happened at once: the raw water delivered far more solids, and the plant added roughly twice as much precipitate per liter. On top of that, the floc is weaker and more shear-sensitive, so it breaks up in the flocculator and clarifier and the fines that reach the filters penetrate deeper into the media instead of forming a surface mat. Deep penetration means headloss and turbidity breakthrough arrive together and early.

Options, in order of cost:

• Optimize the clarifier first: verify sludge blanket depth and waste more often; slow the flocculator or reduce plant rate if you have that flexibility.
• Add a low-dose filter-aid polymer downstream of clarification to bind fines into filterable floc.
• Restore the chemical balance (alkalinity/pH) so the primary coagulant actually works - the polymer is a bandage, not a fix.
• Tighten backwash practice: confirm bed expansion, air scour time, and an adequate filter-to-waste or slow-start so you are not putting dirty water on line.
• Inspect for media loss, mudballs, and media shrinkage; a 9-hour run can also mean the bed is already compromised.

Do not simply increase coagulant again - that adds solids and repeats the cycle. Document each change so you can tie run length back to a specific adjustment.

Thresholds that matter (conventional filtration):

• Combined filter effluent must be ≤0.30 NTU in at least 95% of monthly measurements and must never exceed 1.0 NTU.
• Individual filter effluent above 1.0 NTU on two consecutive 15-minute readings, or above 0.5 NTU on four consecutive readings, triggers a required filter self-assessment. Those are the numbers that convert an operational nuisance into a compliance event.

Decision logic at 0.35 NTU and rising:

1. Confirm the reading - continuous analyzers drift. Pull a grab sample and check calibration and the sample line.
2. Identify which filter is responsible by reading each individual filter effluent, not just the combined value.
3. Try corrective steps first: backwash, place the filter on filter-to-waste, reduce rate, add filter aid.
4. Take the filter out of service if individual effluent cannot be held below the trigger values, if the bed has been disturbed, or if you cannot confidently account for the turbidity. Losing one filter's production is recoverable; a combined-effluent exceedance is not.
5. Document times, readings, actions, and notify the chain of command. If a trigger is crossed, start the self-assessment process and record everything you did.