You have to read
the data.
TCAT is assessed in two parts: knowledge items, and graded diagnostic tasks on synthetic measurement cases with exact, known ground truth. Each part is passed on its own cut score, so nobody certifies on recall alone.
Knowledge, and what you do with it.
Knowing that a bearing outer-race defect shows as a repeating impact rate is one skill. Looking at a spectrum from a machine you have never seen and saying what is wrong with it is a different one. TCAT assesses both, separately.
- Knowledge items
- Multiple-choice and graphic questions against the published syllabus for your category. Marked automatically.
- Graded diagnostic tasks
- Measurement cases you have to interpret and act on. Marked automatically against a known ground truth, with partial credit per step.
- The pass rule
- Each component must be passed independently. A high knowledge score does not carry a failed diagnostic component, and the reverse is equally true.
- Emphasis by category
- Categories I and II lean on data interpretation. Categories III and IV add analyser set-up and written diagnosis on top of it.
What a diagnostic task actually asks you to do.
You are given a case, not a question stem: a machine, its running speed, its context — and its measurements. Then you make the call an analyst would have to make.
You are given the case
A spectrum and a time waveform, the machine and what it drives, the running speed, and the context you would normally have on site.
You state the fault class
What is wrong with this machine — not "something at 2x", but the fault the evidence supports.
You locate it
Where in the train the fault is, which is where a plausible-looking wrong answer usually comes apart.
You put a severity on it
How bad it is, in a severity band. Numeric answers are marked against a tolerance band, because demanding an exact figure tests arithmetic rather than diagnosis.
You recommend an action
What should happen next — and that answer has to be consistent with the severity you just called.
You are marked per step
Partial credit per element, and more than one valid path is accepted where a real analyst could legitimately get there differently.
Some cases are confounders on purpose — an artifact that looks like a fault, or a healthy machine that looks alarming. Reporting a fault that is not there is a real failure mode in this job, and an assessment that never tests for it is not measuring competence.
Why there is no percentage on this page.
TCAT publishes its pass rule as a rule, not a number: each component must be passed on its own cut score. We deliberately do not print a fixed percentage, because no fixed percentage is a standard requirement — the convention that a vibration analyst exam passes at 70% is a convention, disputed in the literature, and quoting it as though it were a standard would be false.
The per-component rule is the part that matters, and it is TIERA's own standard. It exists to protect exactly the competence TCAT is trying to certify: an assessment that can be passed on multiple-choice recall alone certifies the wrong thing.
The number of items, the time allowed, the cut score applied to each component and the resit policy are set alongside the syllabus. They will be published in full before enrolment opens — every candidate should know the rules before they pay, not after.
Every case is synthetic, and every case says so.
This is the kind of thing a training provider usually buries. We would rather put it on the page.
TCAT's exercise and exam cases are generated to a known specification by TIERA's own signal-processing engine — they are not recordings from real machines. Every case displays its provenance to the candidate, so you always know what you are looking at.
We looked hard at the public benchmark vibration datasets and did not use them. None of the three we examined has terms we could rely on for a paid product that redistributes data to paying candidates: one publishes no licence at all, one has lost its dataset page and sits behind a blanket rights-reserved notice, and one only requests acknowledgement for data it did not author. "Widely used in research" is not a licence.
The honest advantage of generating cases is that the ground truth is exact rather than inferred. A recorded dataset's true condition is asserted by whoever published it; a generated case was built to a specification, so what it contains is known rather than believed. Every case in the exam set has to be able to state the basis of its ground truth — "the label says so" is not a basis.
Where TIERA's own captured measurements can replace a generated case, they will: the library is built so a case can be swapped without rewriting the module or the exam around it. Until that happens, this page says synthetic, because it is.
Rules that are in from the start, not retrofitted.
Teaching and exam cases are separate
The cases you learn on and the cases you are examined on are disjoint sets. You cannot revise the answer key.
No two candidates get the same paper
Exam cases are drawn from equivalence classes, so two candidates see different data testing the same competence.
Results cannot change under you
Ground truth is frozen on first exam use, so a later correction to the case library cannot retroactively alter a result.
TCAT is a TIERA-issued credential aligned with the ISO 18436-2 body of knowledge. It is not an accredited ISO certification, TIERA is not an ISO certification body, and TCAT training hours do not count towards the formal training an accredited ISO 18436 certification requires.
Next: what gets you in.
Assessment is only half of it. Entry to a TCAT category runs on verified experience, not a declaration.