RotorLab logo RotorLabDocs

Maintenance: reliability and warnings

This page is part of Maintenance. It covers what your own records say about parts and repairs, and where maintenance warnings show up.

How long parts last here#

On the Components tab, the How long parts last here card shows your fleet's own component experience, kind by kind: a part taken off because it failed or was damaged is an ended life at the hours or flights it had accrued, and a part repaired and refitted after that starts a new life; a part still running, a part pulled for a suspected failure the shop has not confirmed, on the shelf, or taken off for any other reason (scheduled replacement, an upgrade, moved to another airframe) counts as still going at the age it reached, because it was working when it came off. This is why removing a part asks for a reason every time.

Per kind, the table shows Failures, Survived (incl. retired) with the number retired for other reasons in parentheses, and A quarter failed by, Half failed by and Three quarters failed by. A figure the data never reaches shows as "beyond the data" rather than a guess.

A kind appears in the table only once it has at least five failures with recorded hours or flights to stand on, in the card's own words: a median of three points is a rumor. Under the table, each kind that is short says what it has, for example "An estimate needs 5 failures of a kind. Not enough yet for motor: 8 parts, 2 failures." Failures with no recorded hours or flights are named separately there, because they cannot place a failure in time. The estimate is advisory and never a limit; the limits that govern a part stay the ones you set on it.

Is the limit in the right place?#

Below it, Is the limit in the right place? fits each kind of part that has at least five lives with flown hours to a Weibull life curve, from the same failures and the same rule about what counts as one: a part repaired and refitted after a failure counts two lives. Each kind that is short is listed under the table with how many of its lives have flown hours, from how many parts, and how many failures it has. Per kind it shows:

  • Evidence: failures in lives, and the hours flown by the parts of the kind, for example "420 h flown by 8 parts". Damaged parts are counted apart as a rate per 1000 h, because no replacement age prevents damage, and the card counts the parts awaiting a shop finding and those with no fault found.
  • Typical life: the age by which about two thirds of these parts have failed, with a 90% range, "with the wear pattern held at the shape shown". With no failures yet it gives only a floor ("at least 420 h"), at 90% confidence at the assumed shape, because that is all no failures can tell you.
  • Wear pattern: whether the kind wears out with age, fails at random regardless of age, or fails young, with its shape and how much rests on your records: fitted (10 or more failures), partly assumed (3 to 9), mostly assumed (1 or 2) or assumed, with no failures to check it. Under it the card gives the likely range of the shape and, until the label reads fitted, the engineering assumption for the kind (motors and packs wear out; electronics fail at random) and what your records alone say.
  • Before your limit: about what share of these parts fail before the limit most of them carry, and "likely A% to B%, allowing for the shape".
  • Replace by: the age by which one part in ten has failed, when the kind wears out, with "likely A h to B h", and a note when the records cannot yet rule out that age does not matter. For a kind that fails at random it says early replacement does not help, which is itself worth knowing before spending on spares.
  • By how they failed: the failures per failure mode, in your words. A mode with five or more failures gets its own shape and typical life, with the other modes counted as lasting.

Beneath the table, each kind with enough history is checked on later parts: the fit is made on what was known on a split date, with the failures by then and the parts still running counted at the age they had then, and both your limit and the age those earlier lives suggest are applied to the lives going after it. Each line gives failures caught of the total, parts replaced while still working with at least the hours of life left on them, and the share failed by that age as predicted and as seen ("not yet seen that far" when no later part reached it), so you can weigh the two rather than being handed a winner. It needs six failures of a kind; until then it says how many it has.

On the aircraft page, a fitted part of a kind with at least five failures shows a Fleet record line under its life bar: the hours it likely has left as a range, the chance of it failing in the next 30 days at that aircraft's flying rate over the last 90 days (the same rate the maintenance program uses), and the failures the estimate stands on. The range never runs past a limit you set; when the limit is the nearer end, the line says it is held at your limit.

Due and overdue warnings elsewhere#

The By aircraft tab summarizes each airframe: hours, flights, open work, grounding orders, parts due, and component count.

On the Ops Dashboard, the Airworthiness tile goes red when any aircraft is grounded and names each grounded aircraft with its reasons, then each aircraft held at condition unknown with the alerts to review, then lists the open orders and the parts at or near a limit, with a link back to Maintenance. It goes amber when an aircraft is held at condition unknown, orders are overdue, parts are at a limit or an aircraft is ready with limits. The dashboard's KPI strip shows aircraft grounded, with the grounding and open work orders beneath it.

A life limit ranks by when it is projected to fall due first, at the aircraft's own flying rate over the last 90 days. Without a rate, the limit used most ranks first.

What each reading means#

An aircraft reads no go for a grounding defect, a deferral that has run out, open grounding work, grounding work lowered or withdrawn without a release, a grounding Fleet Health finding (severe vibration, motor imbalance, a crash on the last flight) with no work open for it yet, a part past its life limit, a part the register holds unserviceable, repairable or scrapped still shown on the aircraft, an overdue program task (a task never completed reads overdue after its first interval), a mandatory service bulletin past its due date, a major defect that has not been deferred, a status of In maintenance ("In work, not released"), or an expired registration.

It reads condition unknown for an occurrence that names the aircraft after a hard landing, damage, a component failure, a battery thermal event or a flyaway, or of major or critical severity, until a defect raised from the occurrence or a released maintenance record answers it, and for a Fleet Health alert nobody has reviewed and no work answers: an alert that raises no work, such as the compass tracking throttle, until a named person links the record that answers it or dismisses it with a reason (see Review an alert).

It reads Ready with limits for a deferred defect, and Ready, advisory for urgent work, an advisory service bulletin past its date, an urgent Fleet Health finding with no work open for it, a program task due (including a task never completed within its first interval), a part near its life limit, a life-limited part whose flying before registration is not counted, or no maintenance program applied. Worst first, the readings are no go, condition unknown, ready with limits, ready with an advisory and go.

Every surface reads one airworthiness, with its reasons, and it is read from the records, not from whether a sweep has run. The aircraft's own page, the Aircraft page, this page, the Ops Dashboard, Dispatch and the maintenance log give the same answer on the same day.

Data used and the occurrence a record answers#

A maintenance record has a Data used field: the manual section, task card, service bulletin or directive paragraph the work followed. It is part of what the countersignature covers. Every record on the list shows Data used, reading Not named when it is blank, and so does the maintenance log PDF. The suggested release statement says the work was performed "in accordance with the data named under Data used".

A blank Data used never blocks a release, because a description of the work is enough on its own; while it is blank and Release is offered, a warning above Release says the record and its PDF will print Not named and asks you to name the data or say the work is described above. A record that answers an occurrence shows Answers occurrence with the occurrence's reference on the list and on the PDF.

A person who may release to service also sees Answers occurrence on the record sheet. It lists only the occurrence holds on this aircraft, each with its day, category, severity and reference, never the summary. Releasing the record lifts that hold, and the save confirms that the occurrence it answers no longer holds the aircraft.

The other way to answer a hold is a defect raised from the occurrence by an organization admin or a member with Safety records: manage or Maintenance: manage. That defect is at least major: a minor one naming an occurrence that holds the aircraft is refused, whoever raises it, and the aircraft reads No go until a released record rectifies the defect or it is deferred. Lower it after the inspection, by rectifying it. A Safety records viewer, and anyone else, cannot answer a hold.

While an occurrence holds its aircraft, its category, severity and aircraft, and whether it involves an aircraft at all, can be changed in a way that ends or moves the hold only by the same people; anyone else is refused with the field named, and the person who reported it can still amend the narrative.