What Causes a Power Transformer to Fail? Why It Hums

Power transformers fail when thermal, electrical or mechanical stress exceeds the strength of their insulation, windings, connections or accessories. Common causes include overheating, moisture, partial discharge, short-circuit forces, overvoltage and defective bushings or tap changers. Electrical transformers hum mainly because alternating magnetic flux makes the core change dimensions through magnetostriction. A steady hum can be normal; a changed hum needs investigation. The useful question is what changed alongside it: excitation, load, cooling, installation or insulation condition.

Why Do Electrical Transformers Hum?

Core vibration continues without load

Core magnetostriction explains why an energized transformer can hum with its secondary open. Alternating flux produces microscopic dimensional changes in the electrical steel. Supports, clamps and the enclosure transmit this vibration into the surrounding structure and air.

The sound reaching an inspector depends on the installation as well as the transformer. Structural resonance, rigid connections and room acoustics can amplify vibration. A louder electrical room does not automatically mean deteriorating windings.

Excessive volts per hertz can increase noise

Voltage and frequency must be evaluated together when core noise changes. For a fixed winding and approximately sinusoidal excitation, magnetic flux density follows the voltage-to-frequency ratio. Higher voltage or lower frequency increases excitation and can raise magnetizing current, noise and heating.

A 60 Hz transformer supplied at its original voltage on 50 Hz experiences a 20% increase in V/Hz. This calculated example explains the risk; it does not establish an acceptable operating condition. Any proposed voltage reduction or capacity change requires manufacturer confirmation.

Load, cooling and mounting create other sound paths

A sound change that follows loading needs a different investigation from one present at light load. Winding forces, load harmonics, cooling fans and pumps can contribute to the acoustic signature. DC bias can also produce asymmetric core magnetization and additional heating.

Isolation pads cannot correct every noise problem. Incorrect support loading, uneven compression or rigid pipework can bypass the intended isolation. Corrective work must preserve grounding, anchorage, ventilation and mechanical strength.

What Causes a Power Transformer to Fail?

Heat reduces insulation strength before breakdown occurs

Overheating can weaken insulation long before the transformer stops delivering normal voltage. Sustained overload, blocked ventilation, failed cooling, high ambient temperature and harmonic losses raise winding hot-spot temperature. Average temperature and total loading can conceal local damage.

Moisture and contamination add dielectric risk to thermal ageing. In liquid-filled equipment, water, particles and oxidation products can impair insulation and heat transfer. Dry-type equipment faces dust, condensation, conductive contamination and surface tracking.

Short circuits can leave hidden mechanical damage

An external short circuit can damage the transformer even when protection clears the fault successfully. High current produces thermal stress and axial or radial winding forces. These forces can displace conductors, deform windings or loosen supports.

Normal winding resistance after the event does not exclude displacement. Resistance checks electrical continuity and connection condition; it does not describe the complete winding geometry. Significant fault exposure may justify frequency response analysis, impedance comparison and other assessments.

Local defects can defeat a healthy-looking load reading

A small defective component can interrupt an otherwise adequately rated transformer. Loose connections create local resistance heating. Tap-changer contact wear or drive problems can produce overheating, arcing or incorrect voltage. Bushing moisture, seal failure and internal defects introduce separate insulation risks.

Partial discharge can progressively erode insulation without an obvious external defect. Voids, cracks, contaminated surfaces and concentrated electric fields can support discharge activity. Interpretation requires suitable measurements, background-noise assessment and equipment-specific criteria.

The final event may differ from the underlying cause

A useful failure investigation separates the initiating condition, damage mechanism and final trigger. Cooling deterioration may accelerate insulation ageing; a later switching surge may then trigger breakdown. Recording only “overvoltage failure” would leave the cooling problem unresolved.

Stress or defect	Damage mechanism	Evidence to investigate
Cooling deterioration or overload	Hot-spot heating and insulation ageing	Load history, ambient temperature, cooling operation, temperature trends
Moisture or contamination	Reduced dielectric strength, tracking or discharge	Environmental history, insulation tests, liquid condition where applicable
External short circuit	Winding deformation or support movement	Fault records, FRA, impedance and complementary electrical tests
Excessive V/Hz or DC bias	Abnormal excitation and heating	Voltage, frequency, tap position and relevant system measurements
Poor connections or tap-changer defects	Local heating and possible arcing	Thermal patterns, resistance comparisons, operating history
Bushing or insulation defects	Partial discharge or dielectric breakdown	Bushing tests, discharge measurements and other corroborating evidence

How to Investigate a Changed Transformer Hum

Step 1: Establish comparable operating conditions

A noise comparison is useful only when its operating context is known. Record voltage, frequency, tap position, load, ambient temperature and cooling state. Keep the measurement position, distance, instrument settings and background-noise conditions consistent.

Factory and site sound readings need compatible measurement definitions. Sound pressure and sound power describe different quantities. A reading taken beside a wall with fans running cannot directly establish deterioration against a factory result obtained under different conditions.

Step 2: Identify what changed first

The event preceding the sound change helps narrow the investigation. Check whether it followed a short circuit, load transfer, tap operation, supply disturbance, maintenance activity or installation modification.

The same audible symptom can require different tests depending on that history. A change after an external fault raises mechanical questions. A change accompanying lower frequency raises excitation questions. A change after connection or support work raises installation questions. These are investigation paths, not diagnoses.

Step 3: Require independent corroboration

A noise hypothesis becomes stronger when electrical, thermal or mechanical evidence supports it. Combine acoustic observations with appropriate temperature trends, power-quality records, protection events and diagnostic tests.

Liquid-filled and dry-type transformers require different diagnostic combinations. Dissolved gas analysis can reveal thermal or discharge activity in insulating liquid, but interpretation needs gas patterns, generation rates and history. Dry-type assessment relies more on insulation condition, contamination, cooling, thermal observations and partial-discharge measurements where appropriate.

Serious fault indications require isolation and preservation of evidence. Following protection operation, pressure release, abnormal gas indications, visible deformation or damaged bushings, follow the approved response procedure. Repeated trial energization can worsen damage and obscure the original event.

Where Routine Inspections Miss Developing Faults

Nameplate loading does not describe every hot spot

Total current below the rating does not prove adequate thermal margin. Harmonics increase winding eddy-current and other additional losses. Unequal current sharing can overload one winding or connection while aggregate capacity appears acceptable.

Harmonic assessment needs the spectrum and equipment design, rather than one THD number. Use actual waveforms, loss information, cooling conditions and manufacturer-supported thermal assessment. A universal derating percentage cannot represent every transformer and load combination.

Rectifier systems can deteriorate while still producing DC

Continued DC output does not prove that a multipulse system remains correctly configured. Mismatched secondary voltages, impedance differences or unequal bridge loading can concentrate heat in one winding group.

Single-bridge operation can change the original harmonic and thermal duty. Verify bridge status, current distribution and the actual operating arrangement when investigating new noise or heating. Transformer-only acceptance tests do not establish complete rectifier-system performance.

Redundant equipment needs verification under transfer conditions

Long periods of light loading can conceal problems that emerge during emergency transfer. Cooling capacity, connections and protection coordination may face their first demanding operating condition when another supply path fails.

A running fan or pump does not prove effective cooling. Check direction, flow where applicable, controls, alarms and standby changeover. Forced-cooling capacity remains valid only under the conditions supporting that rating.

A normal test result answers a limited question

Infrared imaging identifies accessible surface-temperature patterns, not every internal hot spot. Interpretation depends on loading, emissivity and observation conditions. Compare like-for-like measurements before attributing a temperature difference to damage.

FRA depends on repeatable connections and a suitable baseline. A changed trace can support investigation of mechanical or winding changes, but one local difference cannot independently locate or quantify every defect.

Internal inspection and corrective work require an approved outage procedure. Isolate every source and backfeed path, verify absence of voltage, discharge and ground as required, and prevent unintended energization. Internal reclamping requires design knowledge and diagnostic justification.

Frequently Asked Questions

Is transformer humming normal?

A steady hum can be normal. Core magnetostriction occurs whenever the transformer is energized. Investigate changes against a comparable baseline and supporting operating data.

Why does a transformer hum with no load?

The energized core still carries alternating magnetic flux. Magnetostriction and core losses remain even when the secondary supplies no load.

Does a louder hum mean overload?

A louder hum alone cannot establish overload. Excessive V/Hz, cooling equipment, resonance, mounting changes and electrical loading can each affect sound.

Can a transformer fail below its rated load?

Yes, local stress can exceed acceptable limits before total loading reaches the rating. Harmonics, poor cooling, uneven current sharing and defective connections can create dangerous hot spots.

Does normal output voltage prove the transformer is healthy?

Normal voltage does not exclude developing damage. Insulation deterioration, local heating and mechanical displacement may exist before voltage regulation becomes visibly abnormal.

Can winding resistance rule out short-circuit damage?

Normal resistance cannot rule out winding deformation. Assess significant fault exposure using event records and complementary tests appropriate to the equipment.