Industrial Water: Separate Online Control From Lab Verification

Online instruments and laboratory instruments answer different questions. The online analyser answers “is the process inside its window right now”. The laboratory answers “what is the value, and can I defend it”. Treating one as a substitute for the other is the most common reason a plant ends up with two sets of numbers and no rule for choosing between them.

Two jobs, two standards of evidence

The process reading needs speed and continuity; it does not need four significant figures. The laboratory result needs traceability: a documented sample point, a documented preservation, and a calibration that can be inspected. A plant that asks the online analyser to be defensible, or the laboratory to be continuous, will buy the wrong instrument twice.

Stating the two jobs separately also settles who owns each number. The operating team acts on the online value; the quality function signs the laboratory value. When the two disagree, the rule for resolving it has to exist before the disagreement, not after.

Where online readings drift

Four causes account for most disagreements: fouling on the sensor surface, a calibration that has aged without anyone noticing, a sample line that changes the water before the sensor sees it — degassing, cooling or depositing — and a reference electrode that has drifted. None of them raises an error message, and all of them produce a plausible number.

The verification schedule

Sampling points must match

A verification is only valid if both instruments see the same water. Take the grab sample at the sensor, at the same depth, at the same moment, and record the online value at that moment. Comparing a laboratory result from the outlet with an online value from the inlet produces a disagreement that says nothing about either instrument.

Write the sampling point into the procedure by name and by position. “Sample point 3, sensor depth, valve A” survives staff changes; “the tank” does not.

When to trust the laboratory

The laboratory value settles the argument when the sample was properly taken and preserved, the method is documented, and the result sits inside a calibration verified in the same batch. The online value wins when the question is about a trend or a transient that a grab sample cannot catch. Most plants need both statements written down, so that the operating team knows which number to act on and the quality function knows which number to sign.

A verification plan in one page

List the online points, the laboratory method used for each, the frequency of comparison, the acceptance window and the action when the window is exceeded. Name who takes the sample and where the record is kept. A plan that fits on one page gets followed; a forty-page procedure does not.