Pool Return Bench

A pool of identical bearings, one part number, no serial on the outer ring. Each unit is read at receipt, worn through service, then read again from one view when it comes back. The pool's own records say which units are out on which aircraft, so each return is compared with that short list. The verdict is a match, a conflict or a rescan, and the threshold comes from the pool's yearly comparison count.

Pool · 2,400 units of one part number, 48 drawn

unit, number inside = service cycles claimed record candidate set ▲ unit actually in hand

Click a unit to put it on the bench. The candidates are the units the pool records as out on the same aircraft. Shelf units are checked against the shelf.

Bench · outer ring, 8.6 mm × 8.6 mm view, three lights

ENROLLED RECORD RETURN READ drag to orbit, scroll to zoom

Slope map at enrolment

Slope map of the return read

Reads across service cycles

Run service cycles to build the record. Try wear of 10 µm with re-enrolment off, then on.

Verdict

Best score· Match threshold· Margin over threshold· Usable area (needs 60%)· Compared· Since last enrolment·

Candidates, score in σ above other units

Error budget for this pool

Units in pool2,400

Reads per unit per year3

Units on that aircraft8

Wrong matches allowed a year1.0

reads a year

comparisons a year with no pool record

comparisons a year with it

false-match rate to hold, threshold

Try one

Part on the bench

The claimed unit Another unit from the pool A unit never enrolled

Service cycle

Material removed per cycle3.0 µm

Re-enrol after every match

Return read

Grease or dirt film0.20

Scoring or corrosion in view0%

One lamp only, no slope solve

Surface physics

Feature depth ÷ width (κ)0.10

Share of pattern from the tool lot0.50

Roughness 2.5 µm RMS, 10 µm per pixel, 78 tiles per view. κ sets how fast features wear away: low κ means flat features that go in a few microns. Changing the lot share re-enrols the whole pool.