WORK ORDER — Maintenance and Improvement Record
Work order no: WO-A-2026-10 Date issued: 2026-09-11 Category: Press fitting and press monitoring
Machine / cell: Bushing press cell PR-3 in the gearbox housing line
Status: Closed Engineer: M. Keller (assembly) Total labour: 7.0 hours Parts cost: EUR 96.00
Reason for work: press fit force profile drifts higher while the measured press-in displacement shortens, caught by the force-displacement monitoring chart
1. Fault description as logged by the shift team
Wednesday, report from the quality office: the bushing press cell PR-3 fits a bronze bush into the gearbox housing bore with a nominal interference fit, target press force 25-32 kN and press-in length 24.0 mm, and the force-displacement chart has been creeping up over the month, peak force now touching 38 kN while the recorded displacement at peak drops to 21.5 mm, and a small share of housings show a ringing vibration at the seating end. The press is a C-frame hydraulic station with a load cell under the platen, a linear encoder on the ram, and a controller that plots force against displacement on every part and rejects parts outside the envelope, and in the last week two housings were rejected on the upper force limit that had run fine for months.
I read the chart trends the way a maintenance engineer reads a treasure map: a force climbing with a displacement that shortens means the interference is growing, the bush or the bore got bigger, the bore shrank, the corner radii changed, or the surface finish roughened and raised the frictional barrel of the press-in curve. The vibration at the seat end and the shortening travel pointed to more than friction, so I went to the cell with the drawing, the incoming part gauges and the press shim kit rather than just the hydraulic service manual.
Evening note: a force-displacement chart is the honest witness, the press records the whole story of the joint every cycle, so when the plot moves, believe the plot and go find what changed in the geometry or the surface, not in the hydraulic set point.
2. On-site inspection record (Day 1)
I split the investigation into incoming geometry and press-side condition, because a rising force with a short travel can be caused on either side of the joint. I gauged a sample of the incoming bushes and housings first: the bush outside diameter came in at the top of its tolerance band on 30% of the sample, up to 30.030 mm against the nominal 30.000 mm, and the housing bore came in at the bottom of its band, down to 30.000 mm, and the two tails together pushed the interference from the nominal 0.030 mm up to 0.055 mm, a rise of more than 80% of the design interference, which alone could lift the press force by thirty to forty percent. I pressed a few parts with a fresh gauge bush of mid-tolerance and the force sat back at 27 kN with the full 24.0 mm travel, which proved the press and the ram were innocent and the work was in the incoming geometry, exactly the harder truth for the quality office to hear because it hides in the tails of two tolerated sizes.
I checked the housing finish as well: the bore had been honed to the drawing Ra 0.8, but a batch showed Ra 0.5-0.6 polish that raised the real contact area at the boundary layer and added to the frictional component, and on the bush side the incoming parts carried a coining burr at the leading corner that was shaving the bore during entry. The vibration at the seat end was the classic stick-slip of an over-interfered fit releasing in jumps at the bottom, and the short travel was simply the geometry bottoming the bush early against the corner radius while the force kept climbing. The chart was telling the truth: nothing in the press, everything in the incoming joint.
Evening note: when both the drawn tolerance tails land on the same side, the designed interference quietly doubles, and of all the drift sources a press can catch, geometry creep is the one that never shows up in the hydraulic oil.
3. Corrective action taken
I did not touch the hydraulic set point, because the chart had put the fault upstream. I flagged the incoming bushes at the top of band to the goods-in inspection with a colour mark and a note, so the batch keeps the interference in the design window, and I asked the supplier to steer the outside diameter back to nominal on the next lot. On the housing side I opened the honing process to hold the bore mid-tolerance and the surface to Ra 0.8, and I changed the press-in schedule to a chamfer-protecting lead-in so the deburred bush selects the bore without shaving it. I re-checked the function separately: the hydraulic pressure and the ram speed were within set point, the platen and the tooling coronet were square, the load cell and the encoder readings matched the reference calibration, so the press itself went back on the line untouched and justified by the test.
I re-ran twenty housings through the cell and the force-desplacement chart came back to the design window: peak force 26-31 kN against the 38 kN drift, and the press-in displacement returned to 23.8-24.1 mm against the 21.5 mm short travel, with the ringing vibration at the seat end gone entirely and the rejection counter back to zero.
| Metric | Before | After | Design window |
|---|---|---|---|
| Peak press force | up to 38 kN | 26-31 kN | 25-32 kN |
| Press-in displacement at peak | 21.5 mm | 23.8-24.1 mm | 24.0 ±0.5 |
| Effective interference | up to 0.055 mm | 0.028-0.038 mm | 0.030 ±0.012 |
| Seat end vibration | stick-slip ringing | absent | absent |
| Rejections on force limit | 2 per week | 0 | 0 |
I logged the interference band into the daily SPC report alongside the chart, so the force-displacement plot gets read against the geometry it is sensitive to, and a rising press-in force stops being a mystery the day it appears rather than a campaign that costs a month of housings.
4. Acceptance result after the repair
The cell ran a full week and 240 housings, every part landed inside the force-desplacement envelope with the peak force between 26 and 31 kN and the travel 23.8 to 24.1 mm, and the SPC chart held flat instead of creeping, so quality signed the re-certification and the goods-in gate now checks the bush diameter as part of the incoming lot. I closed the order with the press released, the SPC alert set to trip on a 15% shift in the mean press force over a rolling twenty parts, and the honing and chamfer instructions updated, so the joint geometry is policed at the source where the chart points.
5. Hours and parts list
- Labour — 7.0 h total: incoming geometry gauging 1.5 h, reference press trials 1.5 h, honing and chamfer process correction 2.0 h, SPC alert and acceptance testing 2.0 h
- Parts and consumed — reference gauge bush set EUR 45.00, gauge bush stock and honing verification rings EUR 28.00, chamfer tooling renewal EUR 23.00; total EUR 96.00
- Impact before repair — peak force to 38 kN, travel 21.5 mm, 2 weekly rejections; impact after — 26-31 kN, 23.8-24.1 mm, zero rejections over 240
6. Retrospective notes for the next engineer
When the press chart drifts, do not reach for the pump first. The force-desplacement plot is the witness of the whole joint, and a climbing force with a shortening travel rehearsed the geometry story before I ever opened the tooling: two tolerance tails on the same side had silently raised the interference from 0.030 to 0.055 mm, and a polishing finish plus a coining burr finished the friction. The press was innocent, and testing with a mid-tolerance reference bush proved it in one short run, which is the cheapest diagnostic a press cell owns. Keep the chart on the panel, read it to the SPC mean, and treat a rising press-in force as a geometry alarm, because an interference fit is a designed fighting pair, and the moment the fight gets louder than the drawing promised, the drawing or the parts changed, not the hydraulics.
Attachments: force-desplacement charts before and after, interference calculation sheet, SPC report on twenty and two hundred forty parts, goods-in diameter gauging record (filed with WO-A-2026-10).
Glossary — interference fit: the press fit where the shaft is larger than the bore; force-desplacement envelope: the accepted window on the press chart; stick-slip: the release-and-grab motion of a high-friction interface; coining: the burnishing of a sharp corner during entry; SPC: statistical process control of the measured trend.
Field note — how to read the three zones of a press force-desplacement curve
The paperwork below the trade shows why the force-desplacement chart kept the escalation honest, and the reading is a three-zone habit that every engineer on a press cell should carry. The first zone is the entry: starting at zero displacement, the force climbs quickly as the bush begins to engage the lead-in chamfer and the first millimetre of interference, and a long, soft entry ramp usually means a generous chamfer or a lightly oversize bore, while a steep, short entry ramp suggests a tight lead-in or a sharp bush corner that is shaving instead of guiding. The second zone is the steady barrel: for the middle of the travel the force holds a plateau whose height is set by the interference and the friction pair, and this plateau is the fingerprint of the joint, a flat plateau at the design force means a clean fit, while a rising barrel and a wavy plateau signal galling, a rough finish or an interference growing along the length. The third zone is the seat: at the bottom of the travel the force climbs steeply again as the bush meets the shoulder or the corner radius, and the press-in length recorded at that climb is the displacement check that caught this order, because a short travel with a tall seat peak means the bush is bottoming early against oversized interference, not seating late. Park those three readings, entry ramp, barrel plateau, seat peak, beside every part, and a chart that drifts in any one zone names its own likely cause before the tooling is ever opened, which is exactly the discipline that turned this month-long mystery into a twenty-part job.
Closing operating note for the press cell: pair the displacement guard with the force guard and never run one alone. The force-only limit would have rejected the over-interfered housings on the peak value, which is useful, but it would have said nothing about why, while the displacement reading catches the short-travel signature that names the interference itself, and the two together trip the alarm the day the tails collapse in, weeks before the rejects would have surfaced by fit inspection alone. Keep the reference bush mid-tolerance in the tooling drawer as the cell’s own gauge, run one trial bushing first of every shift, and the chart will stay honest, because a press that watches both its force and its travel is a press that never has to wonder which end of the joint betrayed it.
