Filter Breakthrough Caused by an Alum Feed Problem
You operate a 6 MGD conventional plant with four dual-media filters, each 420 ft², designed for 3 gpm/ft². Over the last two weeks filter run times have fallen from 48 hours to about 20 hours, and one filter has had two back-to-back runs under 16 hours. Filter effluent turbidity from that filter is running 0.30–0.38 NTU; the other three are 0.10–0.15. Settled water turbidity rose from 1.8 to 3.5 NTU, and filter head loss is climbing twice as fast as normal. The alum feed pump was found set at 12 mg/L, but last week's jar test showed optimum at 20 mg/L. Raw turbidity has been steady at 7–9 NTU. Raw alkalinity is 30 mg/L, filtered water pH is 6.3, and finished water free chlorine is 1.0 mg/L. The plant also has a caustic soda feed point after filtration, currently at 2 mg/L.
1. Identify the most likely cause of the filter problems and state the first three corrections, in order, with the calculations that support them.
Read the worked answer
First rule out hydraulic overloading. 6 MGD = 4,167 gpm; total filter area is 4 × 420 = 1,680 ft²; 4,167 ÷ 1,680 = 2.48 gpm/ft², below the 3 gpm/ft² design. The filters are not being pushed beyond design, so the problem is upstream: the plant is underdosing coagulant and the floc reaching the filters is small, weak, and easily carried through the media.
- Restore the alum dose to the jar-test optimum. Increasing from 12 to 20 mg/L is 8 mg/L × 6 MGD × 8.34 = about 400 lb/day of additional alum. Before changing anything, check why the pump was at 12: stroke setting, rotameter calibration, clogged suction, or a calibration error after maintenance. A dose correction on a miscalibrated pump will not hold.
- Check alkalinity and pH. Alum consumes roughly 0.45–0.5 mg/L of alkalinity per mg/L of alum, so 20 mg/L removes about 9–10 mg/L as CaCO3. Raw alkalinity is only 30 mg/L, leaving roughly 20 mg/L, consistent with the filtered pH of 6.3. That is too low for good floc and is corrosive downstream. Add caustic or lime ahead of the filters to bring post-sedimentation pH to roughly 6.8–7.2, and recalculate the caustic dose rather than leaving it at 2 mg/L.
- Verify with settled-water turbidity within 2 hours and adjust polymer or filtration rate if the response is incomplete.
2. What would you do with the individual filter now, and what information tells you when it can return to service?
Read the worked answer
Take it out of service rather than letting it drift. A filter running 0.30–0.38 NTU is above the 0.3 NTU combined-effluent target, even though the individual-filter reporting trigger is higher. Continued operation risks passing floc and, once the media is disturbed, a longer recovery.
After backwashing, put the filter on filter-to-waste until effluent turbidity is below about 0.10 NTU and stable. Filter ripening spikes are exactly where a struggling filter causes the most trouble. While it is down, inspect the media surface for cracks, boils, mudballs, and media loss at the walls; check backwash rate and air scour against the design; look for bed expansion problems or an underdrain issue; and confirm that the media depth and effective size are as built.
Return-to-service criteria should be written, not judgment-of-the-moment: stable effluent under 0.10 NTU for at least two consecutive readings after filter-to-waste, normal head-loss development at the current rate, and no turbidity rise during the run. Two back-to-back runs under 16 hours also justify reviewing backwash sequencing and whether the filter is being returned to service before it is fully rested. If the other filters hold 0.10–0.15 NTU once the coagulant is corrected, that confirms the problem was chemical, not mechanical.
3. What secondary risk does the low filtered pH create, and what would you check in the distribution system?
Read the worked answer
Filtered water at pH 6.3 is aggressive to distribution pipe scale and interior plumbing. The plant has a corrosion control obligation, and low pH with low alkalinity is exactly the condition that dissolves lead and copper. Even if the plant's finished water is adjusted downstream, the caustic feed at 2 mg/L is clearly not sized for this water.
I would check: distribution pH and alkalinity at representative sites, especially dead ends and the oldest service lines; recent lead and copper results and the standing of the sampling pool; whether any sample sites have lead service lines; and whether customers have reported discolored water or a change in taste. Low pH also shifts the chlorine equilibrium toward hypochlorous acid, which is actually more biocidal, so disinfection is not the concern here - pipe corrosion and metals release are.
Operationally, the fix is to add alkalinity and pH adjustment ahead of the filters, not just at the clearwell, because the floc itself needs the pH. Then verify with distribution monitoring before concluding the event is over. A low-pH excursion that lasts a few days can produce lead results weeks later, so the monitoring plan should extend past the immediate correction.