How in-situ flow meter calibration works

A high-accuracy reference flow meter — the REF in a method document — with its own documented, traceable calibration. It is temporarily installed in series with your meter so the two can be compared under identical flow, then removed. It is selected for the maximum flow rate of the line, so your working range sits inside its calibrated range, and its accuracy needs to be substantially better than the tolerance being proven.

The meter under test, abbreviated MUT — or DUT, device under test, in method documents and uncertainty budgets — is the instrument already installed in your plant, the one whose readings you use. The calibration establishes how far it deviates from the reference under real operating conditions.

In series, the identical fluid at the identical rate passes through both, so any difference in readings is attributable to the instruments. In parallel branches the two would see different flows, and the comparison would measure the split rather than the instruments.

Three to five flow points across the actual working range, chosen by you or by the calibration engineer, and each point repeated three to five times. Every run records both meters simultaneously for at least 180 seconds — longer on request. The repeats are not padding: they are what confirm linearity across the range and produce a repeatability figure instead of a single lucky reading.

The recording itself is arithmetic: three to five points, repeated three to five times, at a minimum of 180 seconds per run, is roughly half an hour to an hour and a quarter of pure measurement. Add installation, filling and leak checks, purging air, zeroing and removal, and a single accessible point with a prepared spool piece is typically a few hours. Multiple points on the same line share the setup and are considerably faster per point. The variable is almost always access and isolation, not the measurement.

Gravimetric weighing is the lower-uncertainty method and it is what many national metrology institutes are built on — but it needs the meter in a laboratory, on a rig with a weighing vessel, out of your process. The master meter method reaches a comparable job with a far simpler flow configuration, because the reference meter itself indicates the reference flow rate traceable to the SI units, and that is what makes an on-site rig possible at all. Where your tolerance genuinely requires laboratory uncertainty, we will say so rather than sell you a field comparison.

Because several other things contribute. The repeatability observed on your meter is a Type A contribution evaluated from the runs themselves. The reference's long-term stability between its own calibrations, and the flow computer, temperature and pressure instruments, are all Type B contributions. And the two flow rates are not independent — the continuity equation ties them together along the same pipeline — so a cross-correlation term between reference and meter under test belongs in the budget, as the WGFF guideline sets out. Leaving those terms out produces a smaller number than the truth.

It is the coverage factor. The combined standard uncertainty is multiplied by k to give the expanded uncertainty, the interval expected to contain a large fraction of the plausible values; k is typically between 2 and 3, and k = 2 corresponds to approximately 95 % confidence. It matters that the certificate states it: the same measurement quoted at k = 1 shows an uncertainty half the size for no better reason.

Annual verification is a reasonable default for a critical point, more frequent for custody transfer or billing. The better answer comes from your own data: once several as-found records exist, the observed drift shows whether the interval should be stretched or tightened.

Verification confirms the instrument is still inside its expected tolerance — a pass or fail. Calibration quantifies the deviation against a traceable reference and documents it, so a correction can be applied and drift can be tracked. Verification answers "is it still good enough?"; calibration answers "by how much is it out?".

Yes — and where an adjustment is made it is two sheets, not one: the as-found verification certificate before the correction and the as-left certificate after it. Between them they state the points tested and the number of runs, the deviation and repeatability at each point, the reference instrument and its traceability chain, the expanded uncertainty with its coverage factor, the fluid temperature and pressure, and the recommended next interval — which is what an auditor asks for.