When a bearing fails well short of the life expected of it, four explanations present themselves almost immediately: the bearing was wrong for the duty, it was installed incorrectly, it was not maintained as specified, or it operated in conditions nobody accounted for. Each points at a different party, and each is capable of producing damage that superficially resembles the others. What keeps an attribution defensible is deciding in advance what each explanation would have to leave behind, then reporting what was found and what was looked for and was not.
Rating life is a yardstick, not a verdict
Almost every early-failure dispute starts with a life calculation, and it is routinely over-read. ISO 281 gives basic rating life as a statistical quantity: the life reached or exceeded by ninety percent of a large population of apparently identical bearings under the same conditions. A single unit failing below the rating is not, by itself, evidence of a defect.
It remains indispensable, because it converts a vague sense that a failure was premature into a comparison that can be tested, and because it exposes the input assumptions, which is where most disagreements actually live.
What the modified rating life models
ISO 281 also provides a modified rating life applying a reliability factor and a life modification factor that depends, among other things, on lubricant film condition expressed as a viscosity ratio and on a contamination factor for the cleanliness of the operating environment. That structure is useful forensically because it makes the two most common field explanations quantitative rather than rhetorical.
A bearing running with a thin film in a contaminated environment has a substantially shorter modified life than the same bearing running clean. Entered honestly, the calculation frequently shows the observed life to be consistent with the actual conditions and inconsistent only with the conditions somebody assumed.
The design question: selection against duty
Selection is tested against the loads and speeds the machine really imposes, including start-up transients, reversals, thermal growth and static loading while stationary, which is checked separately against the basic static load rating under ISO 76. A bearing correctly rated for continuous running can be under-rated for shock or for a long idle period under load.
The relevant record is the OEM design calculation and the duty specification it relied on. Where the machine was later re-rated, re-driven or given a heavier tool or impeller without the bearing arrangement being revisited, the design question migrates toward whoever made that change.
The installation question: fits, clearance and mounting
Mounting errors are the most frequently found cause in field practice and the easiest to demonstrate metrologically. Shaft and housing seats are measured against the tolerance grades the OEM specified under ISO 286; ring geometry is checked against ISO 492 tolerance classes and residual clearance against the ISO 5753 clearance groups.
An interference fit that is too tight consumes internal clearance and preloads the bearing beyond design; too loose and the ring creeps on its seat, fretting and generating debris. Mounting force routed through the rolling elements leaves indentation damage before the machine has turned a revolution. A substituted bearing complicates this further, because clearance group, cage material and internal geometry can differ from the original in ways invisible without measurement.
The maintenance question: lubricant, interval and cleanliness
The maintenance case rests on three records and one sample. The records are the specified lubricant, the specified relubrication interval, and what was actually performed. The sample is the lubricant recovered from the failed bearing, examined for wear-debris morphology, moisture, additive depletion, contamination and, critically, for whether it matches the product the specification called for.
Grease incompatibility deserves particular attention because it is invisible to inspection. Mixing thickener systems can collapse the base oil out of a grease and starve a bearing that appears, on any maintenance log, to have been properly serviced.
The environment question: stray current and the drive
Where the bearing sits in or adjacent to an inverter-driven motor, the electrical environment becomes a candidate in its own right. Fast-switching drives produce common-mode voltage that can appear as shaft voltage and discharge through the lubricant film, eroding the raceway with a characteristic spacing and arc-pit morphology under SEM.
This is a specification question as much as a materials one. Insulated or hybrid ceramic bearings, shaft grounding rings, shielded motor cable and proper bonding are the recognised countermeasures, and NEMA MG 1 addresses inverter-fed motors. Whether those measures were specified and maintained is documentary evidence sitting alongside the physical finding.
Concurrent causes are the normal case
These four candidates are not mutually exclusive and rarely operate alone. Contamination shortens the life of a bearing already edge-loaded by misalignment; a marginal lubricant film magnifies the effect of an indentation that a clean, well-lubricated contact would have tolerated for years.
The useful output is therefore not a single culprit but an ordered account: what initiated, what accelerated, and what merely followed. Sequencing is what apportionment arguments are built on, and it is recoverable from the physical evidence far more often than a sole cause is.
What survives cross-examination
An attribution is tested on its inputs. The predictable challenges are that the load spectrum was nominal rather than measured, that the contamination and film factors were chosen to produce the desired answer, that the fits were taken from drawings rather than measured, and that the maintenance history came from interview rather than records.
Work that states each input, its source, and how sensitive the conclusion is to it can be argued with but not dismissed. A calculated life presented without its assumptions is not an attribution at all.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.