Calibration Is a Program, Not an Event

Most instruments in a water or wastewater plant were calibrated on the day they were installed. Some were calibrated again when an operator noticed the chlorine analyzer disagreeing with the bench, or when the state inspector asked. In between, the readings went into the historian, into the reports, and into control decisions, and nobody could say whether they were right. That is not a calibration program. It is a series of events, and the gap between events is where the trouble lives.

What a program is

A calibration program is a list of instruments, each with five things attached: the tolerance the reading must hold, the procedure that checks it, the standard it is checked against, the interval between checks, and the person responsible. It is run on the schedule the intervals produce, it records the as-found and as-left readings every time, and it is reviewed once a year to adjust the intervals from what the records show. The list is not every instrument in the plant. It is the instruments whose readings matter: the ones the permit is built on, the ones the control loops use, and the ones whose failure would cost money or safety. A pressure gauge on a hose bib is not on the list. The effluent flowmeter is at the top of it.

The as-found reading is the product

The moment that matters in a calibration is before anyone touches the instrument. The as-found reading, compared with the standard, says whether the instrument was within tolerance, which means every reading it produced since the last check can be defended. If it was outside tolerance, the record says by how much and since when it is unknown, and the plant has a decision to make about the data in between. A calibration record that shows only the as-left reading proves that the instrument was right for a moment on one day. It proves nothing about the month before, and an inspector who knows the difference will ask.

The as-found record is also what sets the interval. An instrument that has been within a quarter of its tolerance at every quarterly check for two years can go to semiannual; one that arrives out of tolerance at every check needs a shorter interval, or a look at why it drifts, or replacement. Intervals set by habit are wrong in both directions: too short for the stable instruments, which wastes the technician, and too long for the drifters, which loses data.

Adjusting is optional; recording is not

An instrument found within tolerance is left alone. Adjusting it adds the uncertainty of the standard and the technician to a reading that was already acceptable, and it erases the drift trend that would have told the plant when the sensor was aging. An instrument found outside tolerance is adjusted if the slope and offset stay within the healthy range the manufacturer states, and replaced if they do not. Either way, the record carries both numbers, the standard used with its lot and date, and the name.

Standards and test equipment

A program is only as good as what it calibrates against. Buffers expire and absorb carbon dioxide; conductivity standards evaporate; chlorine standards decay in hours; a pressure calibrator has its own certificate with a date on it. The compliance instruments are checked against standards traceable to a national reference, in date, stored as the label says. Control and monitoring instruments can be checked against the plant test instruments, which are themselves calibrated on a schedule. A technician who calibrates the effluent pH electrode with a buffer that has been open since spring has calibrated it to an unknown number.

Running it without a full-time technician

• Put the list and the schedule where the work orders come from, so a calibration is a work order like a pump inspection.
• Batch the work: all the analyzers on one day, all the pressures on another, so the standards are fresh and the technician is in the rhythm.
• Write the procedure for each instrument class once, one page, with the tolerance and the standard on it.
• Record on a template that has as-found, as-left, standard, lot, date, and name as required fields; a record with a blank as-found is not accepted.
• Review the records once a year: adjust intervals, retire instruments that drift, and add anything new.
• Keep the records where the permit records are, because that is where the inspector will look.

The moment it pays

The program pays on the day someone challenges a number. A permit exceedance the utility believes was an instrument error; a customer complaint about a meter; an engineer questioning a flow total in a capacity study. With as-found records the utility can say exactly how good the number was and when it was last proven. Without them the utility can say it calibrated the instrument when it was installed, and that ends the conversation in the wrong direction.