Data logger calibration: when your own numbers drift

The pH probe had been reading 0.6 high for five weeks. Nobody noticed, because a pH of 8.2 in a leachate sump is unremarkable and a pH of 8.8 is also unremarkable. What changed was the dose: the site had been adding acid to bring the pH down before gelling, based on a number that was wrong, and the polymer was being asked to work outside its happy range.

Instruments do not announce their drift. They keep producing numbers with the same confidence, and the decisions built on those numbers keep being made until something fails in a way that cannot be explained.

Which instruments drift, and how fast

pH electrodes are the worst offenders. A glass electrode in leachate fouls within weeks, and the fouling layer slows response and biases the reading. Expect meaningful drift in two to six weeks on an uncleaned probe, and plan for it.

Conductivity cells drift more slowly but are affected by fouling and by polarisation at the electrodes. Level sensors drift with build-up on the face, and ultrasonic units drift with temperature and foam. Flow meters drift with wear and with entrained gas. Load cells drift with temperature and with the mechanical condition of the frame.

The pattern is consistent: anything in contact with leachate drifts faster than anything that is not, and anything that is not cleaned drifts faster than anything that is.

Calibration intervals that match the duty

A pH probe used daily in leachate should be checked against buffers weekly and replaced on a fixed cycle rather than on failure. Conductivity should be checked monthly with a standard solution. Level sensors should be checked against a physical measurement — a dip tape, not another sensor — monthly. Load cells should be checked with a known mass quarterly.

These intervals are not universal, but they are close enough to start with. The point is that the interval is written down and owned, rather than being whatever happened last time someone remembered.

The drift that changed a decision

At one site, a level sensor on the sump had drifted 120 mm high over four months. The control system used the level to decide when to start a batch, so batches were starting early, at a lower volume, with the same polymer dose. The result was over-dosing by roughly 8 percent across every batch, invisible in the dose records because the records showed dose per batch rather than dose per cubic metre.

The site found it during an unrelated audit, when someone compared the sensor reading against a dip tape. Recalibration saved about 8 percent of the annual polymer spend, which was more than the calibration contract cost for three years.

The lesson is not “calibrate more”. It is “record the check in a form that lets you see drift”. A calibration log with only pass or fail tells you nothing. A log with the as-found reading tells you the trend, and the trend tells you when to replace rather than adjust.

A log format that works

Five columns: date, instrument tag, as-found reading, as-left reading, initials. Add a sixth for the reference method — buffer solution, dip tape, known mass. Two years of that log will show you which instruments in your plant are stable and which are consuming your accuracy budget.

Keep the log next to the instrument, not in the office. A log that requires a walk to the office gets filled in from memory, and memory is the least reliable instrument on site.

One more habit is worth the five minutes: note the weather and the batch number beside each check. When a result looks odd three months later, that context is usually what explains it — a probe checked on a 34 °C afternoon reads differently from one checked at dawn in January.

If your dose decisions rest on a number from a probe that has not been checked since the last outage, that number is a rumour. The leachate solidification guide covers which measurements actually drive the dose, so you can prioritise the instruments that matter.