Chasing Turbidity With Alum: Alkalinity Collapse at a 4-MGD Conventional Plant

Alkalinity consumption. Alum consumes roughly 0.5 mg/L of alkalinity as CaCO3 per 1 mg/L of alum (the hydrated salt destroys about 6 equivalents of alkalinity per mole). At 45 mg/L: 45 x 0.5 = about 22.5 mg/L as CaCO3 consumed, leaving only about 28 - 22.5 = 5.5 mg/L, dropping pH to 5.4 to 5.9. Below roughly 20 mg/L of alkalinity the water has almost no buffering capacity, so pH is now controlled by CO2 and swings with every dose change.

Consequences of that pH crash.

• Alum hydrolysis is less complete; the floc that does form is weak and pin-flocculated, so it survives settling as fine particles that reach the filters.
• Soluble aluminum residual risk rises.
• Finished water at pH 5.9 is corrosive to distribution piping and home plumbing, which conflicts with any lead/copper corrosion control the system is relying on.
• The one genuine benefit is that alum removes TOC better at low pH, which helps with the 62 ug/L TTHM number. That benefit does not justify an unstable finished water.

Why filter runs collapsed from 60 to 14 hours. Settled turbidity of 4 NTU is roughly double what the filters were designed to receive, so the solids loading rate on each filter has gone up substantially. Weak, poorly flocculated particles penetrate the media instead of being captured at the surface, plugging the deeper layers and terminating runs on head loss (or worse, on breakthrough). The 0.28 NTU finished number does not mean the filters are healthy: it means the operator is currently winning a race against breakthrough, and a 14-hour run leaves little margin for a slug of solids, a valve failure, or a backwash cycle to go wrong.

First, stop using alum dose as the turbidity lever. The plant is paying for turbidity control with alkalinity it does not have.

Sequence:

1. Stabilize chemistry immediately. Feed caustic soda post-settling (or split ahead of filtration) to bring finished pH to at least 7.5 and target a near-neutral or slightly positive LSI. This is protective of the distribution system and does not undo coagulation, provided it is added after floc formation. Verify by jar test that the caustic does not re-disperse floc.
2. Change the coagulant rather than the dose. A pre-hydrolyzed coagulant (polyaluminum chloride or similar) delivers charge neutralization with far less alkalinity consumption per unit of treatment, which is exactly the tool for low-alkalinity storm water. Even one truckload buys the days needed while raw alkalinity recovers. Run jar tests at plant temperature and mixing energy, not at bench temperature.
3. Bring the dose back toward what alkalinity can support (roughly 25 to 30 mg/L alum, or the PACI equivalent), and recover the lost settling performance with a modest polymer as floc aid, staying within the supplier's maximum feed rate and keeping the polymer from becoming a filter-plugging agent itself.
4. Protect the filters while the transition runs. Cut plant rate so filters are loaded near design, backwash on a shorter head-loss trigger with turbidity-based termination, and use filter-to-waste if it exists. Watch individual filter effluent, not just combined.
5. Address the 62 ug/L TTHM separately, not with more alum. Enhanced coagulation at moderately depressed pH is a legitimate TOC-removal tool, but it must be paired with alkalinity/pH recovery, not used as an excuse for unstable water. PAC is available if taste and odor or additional TOC control is needed.

Why this order. Stabilizing pH first prevents a distribution corrosion problem from compounding a treatment problem, and it costs almost nothing. Changing coagulant second fixes the root cause (alkalinity demand) rather than the symptom (turbidity). Rate reduction and filter discipline third protect public health while the chemistry is in transition. DBP strategy last, because the TTHM number is a running annual figure and will not be decided by this week's operation.

Board triggers for the next 72 hours:

• Settled water turbidity target under 2 NTU, with a written action point (reduce rate, adjust coagulant/polymer) if any 4-hour average exceeds that.
• Individual filter effluent turbidity continuous, with the permit's numeric limit and the utility's own lower action level noted separately. Combined filtered turbidity should be held comfortably below the permit limit, not at it.
• Finished pH at or above 7.5 and LSI not negative; caustic feed paced to flow and alarmed.
• Filter run length: investigate any run under 20 hours; log terminal head loss and backwash turbidity peak and duration.
• Alkalinity and pH on settled and finished water every 4 hours until raw alkalinity recovers above about 40 mg/L.
• Chlorine demand test daily to catch the change in TOC/oxidant demand as the storm runoff passes.

If an individual filter's effluent climbs while settled turbidity is still above 3 NTU: take that filter out of service, do not simply backwash and return it to a losing battle. Backwash it, inspect the media surface for mudballs or a disturbed bed, and return it on filter-to-waste or at a restricted rate only when its effluent is stable. Reduce plant production so the remaining filters are not overloaded, and accept a lower total output (with the distribution system's storage as the buffer) rather than pushing water through marginal filters. If settled turbidity cannot be brought below 3 NTU within the shift, the right answer is to cut rate further and communicate with the distribution side, not to run the filters harder.