Coagulant Feed Drift After a Pump Change
A plant feeds liquid alum (about 48% strength, density 1.33 kg/L) through a diaphragm metering pump into a rapid mix. The target dose is 30 mg/L as alum at 40 ML/d. The operator replaced the metering pump with a different model and set the stroke to match the old pump's setting. Jar tests still show optimum at 30 mg/L. Two days later, settled turbidity has risen from 1.8 to 4.2 NTU, filtered turbidity from 0.08 to 0.22 NTU, and coagulation pH has dropped from 6.9 to 6.5. The operator's log shows a calculated dose of 30 mg/L based on the old pump's calibration curve.
1. What is the most likely cause of the drift, and why did the jar test and the log both look correct?
Read the worked answer
The most likely cause is that the new pump delivers a different actual volume per stroke, so the true dose is not 30 mg/L. The operator set stroke position, not verified delivered volume, and the log used the old pump's calibration curve - so the log reports 30 mg/L while the plant receives something else. The jar test looked correct because it tests the chemical at the target dose, not the pump's actual output; it validates the chemistry, not the feed system. The chemical evidence (pH drop from 6.9 to 6.5) suggests the true dose is higher than intended, since more alum means more acidity. Settled and filtered turbidity rising with a lower pH is consistent with overfeeding and being off the coagulation optimum.
2. How would you verify the actual dose and restore control, and what would you change in the log and calibration practice?
Read the worked answer
Verify by drawdown: measure the volume the pump actually delivers over a timed interval into a graduated container (or by tank-level drop over a known period), convert to mass using density 1.33 kg/L and 48% strength, and divide by flow to get the true mg/L. Compare that with the log. Adjust the stroke or speed to hit 30 mg/L as measured, not as assumed. Then re-check settled and filtered turbidity and coagulation pH over the next several hours. Change the practice: every pump replacement or rebuild gets a drawdown calibration before return to service, the calibration curve is tied to the specific pump, and the log records measured delivered volume rather than a stroke setting. Add a daily or per-shift dose verification so a drift is caught in hours, not days.