Power transformer sound level — sound power per IEC 60076-10

The [guide on the Buchholz relay for oil-filled transformers](/guides/nen-3140/buchholzrelais-gasrelais-oliegevulde-transformator) and the guide on transformer oil analysis (BDV/DGA, IEC 60422) cover chemical and electrical condition indicators of a power transformer. This article covers a less obvious, but equally standardised indicator: sound level, determined per IEC 60076-10.

Three separate noise sources

The total noise of an operating power transformer is made up of three physically distinct sources, each behaving differently:

Core magnetostriction (no-load noise) — the transformer's core laminations deform minutely under the influence of the magnetic field (magnetostriction), with a dominant noise component at 100 Hz (twice the mains frequency of 50 Hz) and its harmonics. This noise is present as soon as the transformer is energised, regardless of load — an unloaded transformer already produces most of its characteristic hum.
Electrodynamic winding forces (load noise) — current through the windings causes electrodynamic forces that increase with the square of the current. For most oil-filled power transformers this contribution is considerably smaller than the no-load noise at rated load, but it is explicitly and separately specified in IEC 60076-10.
Cooling equipment noise — fans and oil pumps on forced-cooled transformers (for example ONAF or OFAF cooling) can, particularly at the highest cooling stage, dominate the total sound level over the transformer core itself.

Measurement method: sound power via a measurement surface

IEC 60076-10 describes a measurement-surface method: microphones are placed at a fixed distance (typically 0.3 m from the main radiating surface of the tank) around an imaginary rectangular measurement surface, after which the sound power level (LWA) in dB(A) is calculated from the measured sound pressure levels. Unlike a sound pressure level, sound power level is a property of the source itself, independent of the distance to the observer — making it comparable between transformers and usable as a guaranteed manufacturer value in a tender specification.

For measurements on an already installed transformer on site (rather than in a test laboratory), IEC 60076-10-1 provides additional guidance to correct for background noise and acoustic reflections from surrounding walls or buildings, which would otherwise pollute the measurement.

Note: IEC 60076-10 is a measurement method, not a noise limit standard. Permitted noise levels at a specific location (for example near residential buildings) follow from local permits and environmental legislation, not directly from IEC 60076-10 itself — the standard does provide the standardised figure (the manufacturer's guaranteed LWA) against which that local limit can be checked.

Why an increasing noise level is a condition indicator

Because no-load noise stems directly from the mechanical state of the core laminations, a noticeable, sustained increase in the measured noise level compared with an earlier baseline measurement can indicate:

Loosened core clamps or laminations, reducing mechanical damping of the magnetostrictive vibration.
Half-cycle core saturation caused by DC components (for example geomagnetically induced currents, GIC, during geomagnetic storms), which drives the core outside its normal operating region and causes noticeably louder, more harmonic-rich noise.
Mechanical degradation of cooling fans or pumps, for example through bearing wear, which increases or changes the character of the cooling noise.

As with thermographic inspection and trend analysis for DGA, a single measurement is less informative than the trend: a periodic noise measurement (or, more simply, a consistent trained-ear comparison by the same maintenance staff) that shows a gradual increase over months or years is an early sign of a developing mechanical or electrical problem, even when oil analysis and thermography show no abnormality at that point.

Practical relevance

During periodic inspection of a power transformer, it is worthwhile to record the noise character — not just the level, but also whether it remains a "clean" 100 Hz hum or takes on a rougher, more irregular character — as part of the maintenance record, alongside the usual oil and temperature data, so that a gradual change between successive inspections is noticed rather than only surfacing at an acute failure.

Common mistakes

Interpreting a single noise measurement as an absolute condition assessment, without a baseline measurement from an earlier, known-good period to compare against.
Confusing sound power (LWA) with sound pressure level — sound pressure depends on distance and the acoustic environment, sound power is a source-bound, comparable property.
Attributing an increase in cooling-fan noise to the transformer core itself, without assessing the three noise sources separately — a simple fan bearing problem then gets wrongly treated as a core problem, or vice versa.
Not accounting for background noise and reflections in a field measurement, which per IEC 60076-10-1 must be explicitly corrected for to obtain a usable result.