Failure Analysis
Diagnostic checklist
Most of the evidence on a failed coupling is destroyed by the first person to clean it. Four observations, taken in the right order and before anything is wiped or slackened, separate the four failure modes and tell you which article to read next.
- Author
- Priyansh Thummar, Editor
- Dates
- Published · Last updated
- Reading time
- 4 minutes
1. Order matters more than thoroughness
Most of the evidence on a failed coupling is destroyed by the first person to clean it. Grease gets wiped away, hubs get degreased, bolts get slackened to get the guard off — and by the time an engineer looks at it, every question worth asking has become unanswerable.
A partial inspection done in the right sequence produces a diagnosis. An exhaustive one done after a cleanup produces a list of observations that cannot be attributed to anything.
2. Before anything is touched
- Photograph both hubs, in place, before cleaning, with enough light to see the flank surfaces.
- Sample the grease — from the sleeve, into a clean container, not from what has been thrown around the guard.
- Take alignment readings as found, before any foot or flange bolt is slackened. Slackening one bolt on a machine with a soft foot changes the geometry you were trying to measure.
- Note the running conditions at failure. This exists in an operator's memory for about a day.
3. The four observations
- Do the two meshes match? Both carry the same torque, see the same grease, ran the same hours. A difference cannot be duty, lubricant or age — it is geometric.
- Where does the contact patch sit? Central and polished, driven to one end, spread across the full flank, or present on both flanks.
- What condition is the grease in? Homogeneous, separated, hardened, milky, or carrying bright flakes.
- Is there debris at the bore or keyway? As distinct from at the teeth.
4. What the combination means
| Meshes | Patch | Grease | Most likely mode | Read next |
|---|---|---|---|---|
| Unequal | Driven to one tooth end | Debris consistent with wear rate | Misalignment | Misalignment damage |
| Equal | Central but degraded | Separated, hardened or contaminated | Lubrication failure | Lubrication failure |
| Equal | Spread across the full length | Debris, sometimes bright flakes | Overload — check duty and service factor | Service factor guide |
| Either | Both flanks worn | Backlash grown faster than flank loss | Torque reversal | Service factor guide |
| Either | Any | Red-brown powder at the bore or keyway | Fretting — a separate investigation | Fretting corrosion |
Where the table sends you: misalignment damage, lubrication failure, fretting corrosion, the service factor guide for the overload and reversal cases, or tooth wear for the full pattern set.
5. When it is not the coupling
The step most often skipped, and the reason the same coupling fails twice. Establish what the coupling was reporting on.
A coupling records the condition of the train around it — alignment, foundation, pipework, bearings, torsional behaviour, lubrication practice. Most of what gets written down as coupling failure is the coupling faithfully recording something else.
If the diagnosis ends at the part number, the replacement runs in the same conditions and wears the same way, on the same schedule.
6. What to write down
So the next person has something the first inspection did not: the hours run since installation and since the last re-greasing; the grease designation actually used, not the one specified; the alignment readings as found and as left; which hub was worse and by how much; the condition of the grease sample; whether the seals were passing; and what changed on the machine in the months before the failure.
The last one is the question that most often produces the answer, and it is never in the maintenance record.
Frequently asked
- What should I check first on a failed gear coupling?
- Nothing, in the sense of touching it. The first action is to record the state you found it in: photograph both hubs before cleaning, sample the grease before it is wiped, and take alignment readings before any bolt is slackened. Almost every diagnostic question later in the investigation is answered by evidence that a well-meaning cleanup destroys in the first ten minutes.
- How do I tell which coupling failure mode I am looking at?
- Four observations settle it in most cases. Do the two meshes match each other? Where does the contact patch sit on the tooth? What condition is the grease in? And is there debris at the bore or keyway as well as at the teeth? Asymmetry between meshes means misalignment; degraded grease with symmetric wear means lubrication; a patch spread across the full tooth length on both hubs means overload; red-brown powder at the fit means fretting, which is a separate investigation.
- Should I replace a failed coupling with the same size and type?
- Only once you know why the first one failed. If the cause was misalignment, a foundation problem or pipe strain, the replacement runs in the same geometry and wears the same way — the failure will simply repeat on a schedule. If the cause was duty the coupling was never sized for, the same size is the wrong answer. Diagnose first; the replacement decision follows from the diagnosis rather than from the part number.
- Is a coupling failure always the coupling's fault?
- Rarely. A coupling is the component that reports on the condition of the train around it — alignment, foundation, pipework, bearings, torsional behaviour and lubrication practice. Most of what is recorded as coupling failure is the coupling faithfully recording something else. That is the useful framing: the worn hardware is evidence about the machine, not just a part to replace.
References
- ANSI/AGMA 1010-F14, Appearance of Gear Teeth — Terminology of Wear and Failure, for the vocabulary a failure report should be written in.
- API 686, Recommended Practice for Machinery Installation and Installation Design (2nd edition, 2009), Chapter 7 — Shaft Alignment, for the as-found alignment practice in §2.
- Editorial note: this checklist is a framework for ordering an investigation, not a substitute for examining your own coupling, and it publishes no acceptance limits — backlash, tooth thickness loss and alignment tolerance all depend on size, speed and duty. A downloadable version and worked cases from returned units are to be added from Super Mech Industries service records and confirmed at technical review.
Related articles
Failure Analysis
Tooth wear
The full pattern set and what each rules in and out.
Failure Analysis
Misalignment damage
When the two meshes disagree, the fault is geometric and local.
Failure Analysis
Lubrication failure
Four ways the film is lost, and why it usually started at a seal.