Interpreting power-quality data: waveform first, flag second

A power-quality recorder is a computer that classifies voltage and current against thresholds. If the threshold is wrong, you get a library of “events” that are ordinary motor starts. If the threshold is right but you only keep the classification, you throw away the only evidence that distinguishes a utility sag from a local fault, from a blown fuse on one phase, from an instrument that was connected to the wrong reference. Power quality work starts with the waveform and the connection diagram, then uses the flags as an index.

How the investigation is actually run

Print the connection sketch on the first page of the report, including probe type and whether you were phase-to-neutral or phase-to-phase. Then pick three events that represent the classes in the log, not the ugliest screenshot. Export raw cycles. Sit with operations and mark what the plant was doing. If the historian clock and the recorder clock disagree, fix that before you tell a causal story.

Teach the reader how you classified: threshold, hysteresis, and whether the software used a sliding rms window. A sag that is really a motor start should be labelled as such even if the default report says “utility.” Conversely, a simultaneous three-phase sag with collapsing current is not a chiller. Harmonic tables without an operating state are decorative.

End with a short list of what the next recorder deployment should change: probe location, duration, analog input from a process contact, or a second meter on the UPS output. Power-quality data is only as good as the question you programmed it to answer.

Symptoms in the file

Repeated sag events at the same clock time as a chiller or a punch press. Flicker complaints from offices on the same transformer as welders. Unbalance that tracks a single large load. Harmonic snapshots taken at Sunday no-load that are then compared to a weekday standard. UPS transfers that the PQ log never saw because the probe was upstream of the bypass. Trips that have no voltage event at all — those belong in breaker diagnostics, not a voltage novel.

Causes of misreading

Phase rotation and probe polarity swapped. Neutral used as “ground” on a three-phase four-wire system so common-mode looks like a sag. Current clamps saturating on inrush, inventing harmonics. Time clocks not synchronized with the process historian, so correlation is folklore. Classifying every rms dip as “utility” because that is the default software story.

Investigation

Draw where the probes sat. Export a few raw cycles around each class of event, not only the CBEMA plot. Compare voltage and current: a sag with rising current is a load; a sag with falling current is more likely source or upstream opening. Look at all three phases. For distortion, read nonlinear loads before you specify a filter. For ride-through, read UPS under real load.

The conclusion is a mechanism sentence: this envelope violation is the starting current of compressor C on bus 2; this notch is a six-pulse drive on a weak transformer; this “swell” is a lost load plus regulator overshoot. Incomplete files stay incomplete in writing. Related electrical environment: when the network problem is electrical.