Learn
What is power quality?
Power quality is how closely the voltage and current supplied to equipment match an ideal, undistorted sine wave at nominal magnitude and frequency. Poor power quality means disturbances — sags, harmonics, imbalance, transients — that cause equipment to trip, overheat or fail early.
The seven disturbance types, with their bounds
Power quality is not one phenomenon. IEEE 1159 classifies electromagnetic disturbances by duration and magnitude, and the categories behave completely differently.
| Disturbance | What it is | Typical bound |
|---|---|---|
| Sag (dip) | RMS voltage falls to 10–90% of nominal | 0.5 cycle to 1 minute |
| Swell | RMS voltage rises above 110% of nominal | 0.5 cycle to 1 minute |
| Interruption | Voltage below 10% of nominal | Momentary to sustained |
| Harmonics | Current or voltage at integer multiples of 50/60 Hz | Steady state, assessed statistically |
| Imbalance | Unequal magnitude or phase spacing between the three phases | Steady state; 2% is a common alarm point |
| Transient | Sub-cycle impulse or oscillation | Microseconds to milliseconds |
| Flicker | Rapid voltage fluctuation causing visible light modulation | Assessed per IEC 61000-4-15 |
The reason this matters practically: each one requires a different measurement to see it. An energy meter reporting averages every 10 seconds sees none of them except gross imbalance. That is not a defect in the meter; it is a different instrument for a different job.
The standards that actually govern it
- IEC 61000-4-30 defines the measurement methods and two instrument classes. Class S is adequate for surveys and statistical assessment. Class A is required where a measurement will be used to bill, penalise or contest — its results are reproducible between compliant instruments, which is the whole point of a dispute.
- IEEE 519 governs harmonic limits at the point of common coupling, and it is a statistical procedure, not a threshold reading. It requires 3-second very-short-time and 10-minute short-time aggregation per IEC 61000-4-7, then evaluation of the daily 99th percentile and the weekly 95th and 99th percentiles. A single spot reading of THD does not demonstrate compliance or non-compliance with anything.
- IEEE 1159 classifies and characterises the disturbances themselves.
- ITIC (formerly CBEMA) and SEMI F47 define what voltage disturbance equipment is expected to ride through. See what is a voltage sag.
Harmonics, in the detail the standard actually specifies
Harmonic current is drawn by any load with a rectifier front end: variable-frequency drives, UPS systems, LED drivers, servo amplifiers, induction heating. It does no useful work, but it heats conductors and transformers, and it can excite resonance with power-factor correction capacitors.
IEEE 519 sets current limits as total demand distortion — distortion relative to the maximum demand load current, not to the instantaneous current — and the allowance scales with the ratio of short-circuit current to load current at the point of common coupling. The stiffer the supply relative to the load, the more distortion it can absorb:
| Short-circuit ratio (I_sc / I_L) | Total demand distortion limit |
|---|---|
| < 20 | 5.0% |
| 20 – 50 | 8.0% |
| 50 – 100 | 12.0% |
| 100 – 1,000 | 15.0% |
| > 1,000 | 20.0% |
Voltage distortion limits for systems at or below 69 kV are 5.0% individual harmonic and 8.0% total harmonic distortion.
There is no defensible published figure for money saved by fixing harmonics. OptimizeOS does not quote one, and any vendor that does should be asked for the derivation. The honest case for harmonic assessment is transformer derating, nuisance breaker tripping, neutral conductor heating and capacitor failure — all real, all site-specific, none of them a percentage anyone can generalise.
Worked example: putting a number on the events
Two independent sources give a defensible way to size this.
The scale of it. Lawrence Berkeley National Laboratory estimated the annual US cost of power interruptions at $79 billion, of which 67% — $52.3 billion — comes from momentary events of five minutes or less, against $26.3 billion from sustained interruptions. Two-thirds of the money is lost to events short enough that nobody logs them. Per interruption, the momentary figures are $605 commercial and $1,893 industrial. (LaCommare & Eto, LBNL-55718, 2004; figures in 2002 dollars.)
The per-event cost. A Brazilian field survey of 33 companies across selected sectors, conducted on site in 2019, measured an average cost of $7,365 per event and $6.72 per kilowatt interrupted, at roughly eight events a month. By sector: metallurgical $10,788, automotive $7,975, food $2,898. (Motoki et al., Energies 2021, 14(10), 2874. Brazilian sample; the averages cover selected sectors and omit several higher-cost ones.)
Applying the per-kilowatt figure to a specific line:
- A sag drops a 900 kW packaging line
- 900 kW × $6.72/kW interrupted = $6,048 for the event
- At four such events a year: $24,192
The right way to use these numbers is as a sizing check, not as a quotation. A facility that logs its own events — timestamp, feeder, what stopped, how long recovery took, what was scrapped — replaces the survey figure with its own within a quarter, and that number is the one that funds anything.
What power quality is not
It is not reliability. Reliability is about whether the power is there, measured in interruption counts and minutes. Power quality is about what the power is doing while it is there. A facility with excellent SAIDI numbers can have a serious sag problem, and usually does, because the sags come from faults on other people's feeders that clear in a few cycles and never register as an outage.
It is not measurable from energy data. A meter reporting averages every 10 seconds cannot see a sag, cannot perform IEEE 519 aggregation, and cannot report true power factor without a voltage reference. Power quality assessment requires an instrument at the point of common coupling and at one or two sensitive process feeds — one good instrument in the right place, not many cheap ones everywhere.
It is not fixed by a single device. "Power conditioning" covers at least six different technologies addressing different disturbances. A ride-through solution for sags does nothing for harmonics; a harmonic filter does nothing for transients.
Poor power factor is not a power quality problem in the standards sense. It is a billing and capacity condition. The two are related, they are measured by the same instruments, and they are different subjects. See what is power factor.
Common questions
What causes poor power quality?
Most voltage sags originate outside the facility, from faults on the utility's transmission or distribution system that clear in a few cycles. Inside the facility, the common causes are large motor starting, transformer energisation, harmonic-producing electronics such as variable-frequency drives, and unbalanced single-phase loading across the three phases.
What is the difference between IEC 61000-4-30 Class A and Class S?
Class A specifies measurement methods precise enough that two compliant instruments produce the same result on the same signal, which is what makes a measurement usable in a billing dispute or a penalty assessment. Class S is a less demanding class intended for statistical surveys and general monitoring.
Does IEEE 519 apply to my facility?
IEEE 519 sets harmonic limits at the point of common coupling — the boundary between the utility and the customer — and assesses them statistically over a week, using daily 99th percentile and weekly 95th and 99th percentile values. A single spot THD reading demonstrates nothing about compliance either way.
Can an energy meter measure power quality?
No, beyond gross voltage imbalance. Sags occur at half-cycle resolution, roughly 8 milliseconds, and harmonic assessment requires 3-second and 10-minute aggregation of a full spectrum. An energy meter reporting averages every 10 seconds or 15 minutes averages all of it away.
Related pages
OptimizeOS records what the electrical system was doing at the moment a machine stopped, and ingests data from a dedicated power quality instrument where one is installed — see power quality.