A utility inspector connects an analyser at your incoming supply, records for a week, and hands you a non-compliance notice. Nothing in your plant has failed. Motors run, drives control, product ships. Yet the measurement says your site is injecting more harmonic current than the connection is allowed to carry — and the IEEE 519 harmonic limits are the yardstick that decided it.
This is the uncomfortable position many variable frequency drive (VFD) users find themselves in. The drives that made the plant efficient are also the source of the distortion, and the standard that governs it is widely cited but poorly understood. This article explains what IEEE 519 actually limits, why the limits are structured the way they are, and why a plant can be perfectly healthy internally while failing at the meter.
What IEEE 519 limits, and where
The first misconception to clear up: IEEE 519 does not set limits inside your plant. It sets limits at the point of common coupling (PCC) — the boundary where your installation meets the utility, the point other customers also connect to. Distortion at an individual drive terminal can be high without breaching anything, because the standard cares about what you inject into the shared network, not what circulates inside your own switchboard.
The current edition, IEEE 519-2022, keeps the same two-part structure that has defined the standard for years. There are limits on voltage distortion, which the utility is largely responsible for holding, and limits on current distortion, which the consumer is responsible for. The split reflects a shared duty: the customer controls the harmonic current it draws; the utility controls the network stiffness that turns that current into voltage distortion.
Understanding that division is the key to the whole standard. Voltage distortion is the symptom everyone experiences; harmonic current is the cause each customer is accountable for.
Voltage distortion limits
Voltage limits are defined by the voltage level at the PCC, and they tighten as voltage rises — the higher-voltage network is shared by more customers, so less distortion is tolerated. The widely applied thresholds are:
| PCC voltage | Individual harmonic | Total harmonic distortion (THD) |
|---|---|---|
| ≤ 1 kV | 5.0% | 8.0% |
| 1 kV to 69 kV | 3.0% | 5.0% |
| 69 kV to 161 kV | 1.5% | 2.5% |
| Above 161 kV | 1.0% | 1.5% |
The lowest voltage band, at or below 1 kV, was formalised in the 2022 edition with the more relaxed 5% / 8% values shown above. The principle is intuitive: distortion on a 400 V board affects fewer parties than the same distortion on a 132 kV transmission bus, so the low-voltage limit is looser and the transmission limit is strict.
Current distortion: why the limit is not a fixed number
Here is where most misreadings of IEEE 519 begin. There is no single allowable current THD figure. The current limit is expressed as Total Demand Distortion (TDD) — harmonic current as a percentage of your maximum demand load current, not of the instantaneous current — and it is scaled to the short circuit ratio at your connection.
That short circuit ratio is the ratio of the available short circuit current at the PCC to your maximum demand load current. A stiff connection (high short circuit current relative to your load) can absorb more harmonic current before the voltage distorts, so it is allowed a higher TDD. A weak connection is allowed less. The TDD ladder runs roughly as follows for systems up to 69 kV:
| Short circuit ratio (Isc/IL) | Allowed TDD |
|---|---|
| Below 20 | 5.0% |
| 20 to 50 | 8.0% |
| 50 to 100 | 12.0% |
| 100 to 1000 | 15.0% |
| Above 1000 | 20.0% |
Within each row, the individual odd-harmonic components are capped in bands, with the lower-order harmonics allowed more than the higher-order ones. The 2022 edition also refined the treatment of even harmonics, holding them to a fraction of the neighbouring odd-harmonic allowance.
Two practical consequences follow. First, the same drive can pass at one site and fail at another purely because of connection strength — nothing about the equipment changed. Second, using TDD rather than instantaneous THD prevents a false failure at light load: a plant running one drive at low output shows high percentage distortion of a tiny current, which matters little; TDD measures it against full demand, which is what the network actually experiences.
Why VFD-heavy plants fail
A standard six-pulse VFD front end draws current in pulses, producing characteristic harmonics — the 5th, 7th, 11th, 13th and so on. Each individual drive may be within reason. The problem is aggregation: put twenty drives on one bus and their harmonic currents add, while your maximum demand does not rise in the same proportion. The ratio that IEEE 519 measures — harmonic current against demand — worsens.
Textile plants, cement grinding, water pumping stations and any facility built around large numbers of medium-size drives are the classic offenders. The failure is rarely one bad machine; it is the sum. This is also why buying a “compliant” drive does not guarantee a compliant site — the datasheet describes one drive into a defined impedance, not your drives into your actual connection.
Weak connections compound it. A plant at the end of a long feeder has a low short circuit ratio, so it faces the tightest TDD limit at exactly the point where its harmonic current produces the most voltage distortion. Rural industrial sites fail for this reason more often than urban ones with identical equipment.
Getting to compliance
The route to compliance follows a fixed logic, and skipping the first step is the most common and expensive mistake.
Measure before you specify. A harmonic survey at the PCC and at major drive groups establishes the actual distortion and its dominant orders. Mitigation designed from assumptions rather than measurement routinely targets the wrong harmonics.
Then choose mitigation to match. The realistic options, in rising order of capability and cost, are line reactors and DC chokes on individual drives, passive tuned filters for a dominant harmonic, multi-pulse drive front ends that cancel lower-order harmonics, and active harmonic filters that inject a corrective current across a broad spectrum. The right answer depends on the harmonic spectrum, the load profile and whether the plant will expand.
Model before you buy. Passive filters interact with the network. A filter tuned for one harmonic can form a resonant circuit with system or cable capacitance at another frequency and amplify it. A proper harmonic study models these interactions before hardware is ordered — which is far cheaper than discovering the resonance after commissioning. This is the core of a competent harmonic analysis, and it is where a modelled solution separates from a guessed one.
Re-measure to prove it. Compliance is demonstrated by measurement at the PCC against the applicable voltage and TDD limits, using the statistical (percentile-based) evaluation the standard specifies rather than a single worst-case instant.
FAQ
Is IEEE 519 mandatory?
IEEE 519 is a standard, not a law, but utilities and grid codes routinely adopt it as a connection condition, which makes it contractually binding for that connection. Many industrial supply agreements reference it directly, and non-compliance can carry penalties or connection restrictions.
What is the difference between THD and TDD?
THD measures harmonic content against the actual current at the moment of measurement; TDD measures it against the maximum demand current. IEEE 519 uses TDD for current limits precisely so that light-load operation does not produce a misleading high-percentage result. The distinction matters whenever load varies, which is most plants.
Why does my compliant drive still cause a compliance failure?
A drive datasheet describes one unit into a specified source impedance. On site, multiple drives inject harmonics that add, and your actual short circuit ratio may be lower than the reference used in the datasheet. Compliance is a property of the whole installation at the PCC, not of a single device.
Will a power factor capacitor bank help with harmonics?
Usually the opposite. Capacitor banks can resonate with system inductance near a harmonic frequency and amplify the very distortion they sit beside. Adding capacitors to a distorted network without a study is a common way to make a harmonic problem worse rather than better.
What does a harmonic study involve?
Measurement at the PCC and major loads, building and validating a network model, identifying dominant harmonics and any resonance, evaluating mitigation options, and confirming the chosen solution against IEEE 519 limits by simulation before procurement. It is typically a few weeks of work depending on network size and data availability.
SoftNitro Engineering provides harmonic analysis and power quality studies to IEEE 519 for industrial and utility clients, covering measurement, mitigation design and resonance assessment. See our full range of power system studies or request a scope and fee for a harmonic compliance study.
