Filter Runs Collapse After a Storm at a 6-MGD Conventional Plant
Your 6-MGD conventional surface-water plant (two 3-MGD trains) draws from a Piedmont river. For ten days after a 2.5-inch storm, raw turbidity has run 18-30 NTU (normal 4-8 NTU), raw alkalinity has fallen from 42 to 30 mg/L as CaCO3, and raw TOC has risen from 3.1 to 5.4 mg/L. Alum feed is unchanged at 28 mg/L; the filter-aid polymer and the post-filter lime feed are also unchanged. Settled turbidity is 2.4 NTU against a 1.0 NTU internal trigger, individual filter effluent turbidity has climbed from 0.08 to 0.26 NTU, and filter runs have fallen from 58 hours to 19 hours. Finished-water pH is 7.1 against a 7.6-7.9 target. Combined filter effluent is 0.19 NTU, the clearwell is at 71 percent, and system demand is normal for a weekday. The jar tester works, but the last recorded jar test was run before the storm. You are the shift operator; the next shift arrives in six hours.
1. Interpret the alkalinity, pH, and turbidity pattern. What is most likely happening to coagulation, and what is the first thing you do before you touch the alum dose?
Read the worked answer
Alum consumes alkalinity at roughly 0.45-0.5 mg/L as CaCO3 per mg/L of alum. At 28 mg/L that is about 13 mg/L of alkalinity demand. Raw alkalinity fell from 42 to 30 mg/L as CaCO3, so the water entering the rapid mix has roughly a third less buffering than the dose was set for, and the coagulation pH drops below the range where alum floc is strongest (about 6.0-6.8 for most surface waters). The higher TOC makes it worse: more natural organic matter competes for the coagulant, so the effective dose per unit of turbidity is lower than it was.
The symptoms line up with underdosing plus pH suppression, not with filter failure. Settled turbidity of 2.4 NTU means floc is escaping the settling basin, so the filters are receiving a load they were never designed for. That explains the drop from 58 to 19 hours and the rising individual filter effluent turbidity. The filters are the messenger, not the cause.
The first action is a jar test matrix, not a dose change. Run four or five alum doses bracketing 28 mg/L, each with and without caustic or lime to hold coagulation pH near 6.3, and measure settled turbidity at 20-30 minutes plus supernatant turbidity, not just floc appearance. If streaming current or zeta potential is available, use it to bracket the charge-neutral point.
Do not simply raise alum. An extra 10 mg/L of alum would consume another 4.5 mg/L of alkalinity and push pH even lower, which can make floc worse and also increases the lime needed downstream.
2. Individual filter effluent turbidity is rising and runs are 19 hours. What immediate operational actions protect the finished water, and what triggers taking a filter out of service?
Read the worked answer
Protect the finished water first, then chase the cause.
Monitoring: combined filter effluent is at 0.19 NTU, which is inside the 0.3 NTU limit but with almost no margin. Move to continuous individual filter effluent turbidity if the equipment is available, and at minimum read and record every two hours. Record filter flow rate, headloss, and run time for each filter - at 6 MGD across two trains the filters may be running well above design loading.
Diagnosis: a filter with 19 hours of run time, low headloss, and rising effluent turbidity is showing breakthrough, not loading. That points to media upset, air binding, or a backwash problem rather than raw-water quality alone. Probe the bed for mudballs and cracks, check bed expansion and backwash rate, verify the air scour, and look for media in the wash troughs.
Removal triggers: take a filter out of service when its individual effluent turbidity reaches 1.0 NTU, and evaluate any filter above 0.15 NTU on two consecutive 15-minute readings. Start every filter on filter-to-waste after backwash and do not return it to service until its effluent is below 0.10 NTU.
Also notify the lab and water quality supervisor, check whether the polymer feeder is actually delivering (draw-down test on the day tank), and document the times. If combined effluent reaches 0.3 NTU, the plant is out of compliance and the state must be notified under the required reporting timeline.
3. Finished-water pH is 7.1 against a 7.6-7.9 target. What is the risk, and how do you correct it without making the filters worse?
Read the worked answer
pH 7.1 with alkalinity in the low 30s is aggressive water: negative Langelier index, elevated lead and copper solubility at the tap, red water complaints, and higher chlorine demand in the distribution system. The immediate cause is that the higher alum demand is eating alkalinity the lime feed was never sized for - a 10 mg/L alum increase consumes about 4.5 mg/L of alkalinity, which is roughly 5-6 mg/L of hydrated lime.
Do not fix this by dumping lime into the rapid mix. Coagulation pH and finished-water pH are two different setpoints. Raising pre-filter pH without re-running the jar tests can break the floc you are trying to build, and it can also move you off the optimum for the polymer.
Sequence: confirm the lime feeder is actually delivering (day-tank draw-down, slaker condition, scale in the line and rotameters); run the jar tests at the corrected coagulation pH; set the post-filter lime to hold finished water at 7.6-7.9; then verify pH, alkalinity, and calcium at the plant and at distribution sample points, because the number that matters for corrosion control is the one at the dead ends, not the one on the bench.
Log every change with times and tell the incoming shift what was changed and why.