Machine Part Repairs by the Numbers
Chapter 10 of 10 · 10 min read
Proving the repair and keeping the record
Inspect a repaired part by eye, dye, magnetic particle, ultrasonics and hardness, check its size, run it in, work out what an inspection is worth, and keep the record that improves the next repair.
A repaired part looks finished when it leaves the lathe, but appearance proves little. The weld under the machined surface may hide porosity or a crack, the heat-affected zone may have hardened, the shaft may have moved out of line, and the gear tooth may be slightly the wrong shape. Proving a repair means checking each of those, before the part goes back into the machine and again once it is running, and recording what was found, so that the owner knows what they have and the next person to work on the part knows what was done. Dunham observed that straightening a welded shaft was itself, to a degree, a test of the weld; modern inspection makes that test deliberate.
This chapter sets out the inspections for repaired machine parts, the dimensional and functional checks, how to judge what an inspection is worth, and the repair record, and then reviews the repairer’s season.
Looking for defects
The methods for finding defects in repaired parts are the same ones used for welds generally, chosen for the part and the defect expected.
| Finds | Limits | |
|---|---|---|
| Visual, with magnification | surface porosity, undercut, poor blending, cracks that can be seen | only what is visible |
| Dye penetrant | fine cracks and pores open to the surface | surface only; surface must be clean |
| Magnetic particle | surface and near-surface cracks in ferromagnetic steels | not for aluminium, austenitic stainless or most cast irons' internal defects |
| Ultrasonic | internal lack of fusion, cracks and voids | needs skill, calibration and access |
| Hardness testing | hard zones from fast cooling; soft weld metal | spot readings; needs a smooth surface |
For the harvester shaft, the repairer used magnetic particle inspection of the finished journal and its blended ends, which would show any crack at the surface or just beneath it, and portable hardness readings across the weld, the heat-affected zone and the unaffected shaft, which showed a gradual change with no hard spots. Testing Welded Joints by the Numbers explains these methods, and what a sample of inspections can prove, in its chapter on non-destructive testing and sampling, and the use of hardness readings in its chapter on hardness and the heat-affected zone.
Dimensional checks
A repaired part must also be the right size and shape:
- diameters and fits of journals, bores and seats, measured with micrometers and bore gauges against the bearing maker’s fit;
- run-out of shafts at every journal and seat, as Chapter 6 described;
- surface finish of bearing and seal surfaces, by comparison or a roughness gauge;
- profile, thickness and pitch of rebuilt gear teeth, against a template or a gear tooth vernier and the neighbouring teeth;
- alignment and centre distances of housing bores, by mandrel or measurement.
Each reading is written down, not just judged acceptable, because a measured value tells the next person far more than a tick.
Running in
The final proof is in the machine. A repaired part is fitted with new bearings and seals where needed, the machine is run without load and then under light load before full service, and the repairer watches for the signs of trouble: rising bearing temperature, felt by hand or measured with an infrared thermometer; noise or knocking; vibration; and oil leaks. For gears, the contact pattern after a short run under load shows whether the rebuilt tooth is sharing the load properly. A part that runs cool, quiet and steady for its first hours has passed its most practical test; one that does not is stopped before it does damage.
- Avoid: Run without load firstlisten, feel and look for leaks
- Avoid: Bearing temperatureschecked at intervals; steady, not still rising
- Avoid: Vibration and noisecompared with the machine before failure, where known
- Avoid: Gear contact patternafter a short run under load
- Avoid: Fasteners recheckedafter the first hours of running
- Do: ? Follow-up inspectionat the owner's next service, or after a set number of hours
What an inspection is worth
Inspection costs money, and its value can be estimated in the same way as the repair decision in Chapter 1. This is an illustration. Suppose that a journal build-up has, from experience, about a 5% chance of containing a crack near its surface that would cause it to fail in service, at a cost of about A$10,800 in downtime and a new shaft, as for the harvester shaft. Without inspection, the expected cost of that risk is about 0.05 × 10,800 = A$540. A magnetic particle inspection costing about A$60 that finds about 90% of such cracks reduces the expected cost to about 0.05 × 0.10 × 10,800 = A$54, plus the A$60 inspection, about A$114, a saving of about A$426 for every repair, on average.
The arithmetic usually favours inspection for any part whose failure would be costly, and it shows that inspection is part of the repair, not an extra.
Proving a repair procedure
Some repairs are made often enough that the procedure itself is worth proving once, on a trial piece, before it is used on customers’ parts. The repairer built up a journal on an offcut of the same steel as the harvester shaft, using the written procedure from Chapter 4, machined it, and sectioned it across the build-up. The etched section showed full fusion between beads and into the shaft, no porosity or cracks, and a heat-affected zone whose hardness readings were within the procedure’s aim. That one trial, costing about half a day, turned the procedure from a plan into evidence, and it is cited in the record of every shaft built up with it. Testing Welded Joints by the Numbers explains how procedure qualification works for welds in general; for machine part repairs, a proved procedure and the records of the parts repaired with it are the workshop’s equivalent.
Follow-up
The last check of a repair happens after it has been in service for a while. The repairer asks the owner of each significant repair to report how it is running, and inspects it at the next service where possible: bearing condition and temperature, signs of fretting or wear, and any cracking at the repair. A repair that is still sound after a season has earned its place in the record as a method that works; one that shows early signs of trouble can be replaced before it fails.
The repair record
Weld Repairs That Hold makes the case for a repair dossier in its chapter on the repair dossier, and the same reasoning applies to machine parts. The repairer’s record for each repair holds:
- the failure: what failed, how it was read, and the cause found;
- the decision: repair or replace, the options and costs considered, the owner’s decision and any advice given in writing, such as a recommendation to fit a new part later;
- the procedure: material identification, method, filler, preheat, sequence, cooling and machining;
- the results: inspection, hardness and dimensional readings, straightening, running-in observations;
- the status: permanent or temporary, with any limits on use and the date or hours of the next inspection.
Temporary repairs are labelled on the part itself, where practical, as well as in the record, so that nobody mistakes them for permanent ones a year later.
When to inspect
Inspection is most useful at the points where a defect can still be fixed cheaply. The repairer inspects before welding, to find every crack that must be ground out; after welding and slow cooling, before machining, to find surface defects in the weld while there is still allowance to grind them out and reweld; after machining and straightening, to find anything the machining has uncovered or the press has opened; and after running in, if the part is critical. Each stage catches defects the earlier ones missed. Inspecting only at the end saves little, because a defect found then means undoing the machining as well as the weld.
Who inspects
For most workshop repairs, the repairer’s own trained people do the visual, dye penetrant, magnetic particle and hardness checks, following the method’s instructions and recording the results. For critical parts, or where a customer, insurer or standard requires it, inspection is done by people qualified in the method, often from an inspection company, and their report becomes part of the record. Ultrasonic testing of internal defects nearly always needs a qualified operator. The repairer knows the limits of its own inspection and calls in qualified inspectors when a part’s consequences of failure call for it.
Warranty and the owner’s part
A repair is only as good as the use it is put to. The repairer’s record states what the repair is, what it is suitable for and any limits, such as a temporary repair to be replaced at the end of the season, a maximum load, or an inspection after a set number of hours. The owner is given a copy. A trailer axle overloaded again, a gearbox run without oil, or a temporary repair left in place for years, is not a failure of the repair, and the record shows the difference. Equally, the repairer stands behind a repair that fails within its stated use, investigates the failure honestly, and learns from it.
Learning from the records
Over a season or two, the repair records become a body of evidence. The workshop manager reviews them each year: which repairs lasted, which came back, which methods worked on which parts and materials, how long each kind of repair took against its quote, and what each failure turned out to have been caused by. The review turns individual jobs into the workshop’s own knowledge, so that its procedures, its quotes and its advice to owners improve year by year, and so that knowledge stays with the business when people leave.
The running case: the season reviewed
This is an illustration. At the end of the harvest season, the workshop manager reviewed the season’s repairs.
The harvester shaft ran the season and was replaced with the new shaft in the off-season, as planned; the repaired shaft became the contractor’s spare. The auger gear’s rebuilt tooth ran the season; the new gear and pinion were fitted afterwards. The stitched gearbox housing went back into service and was checked at the next service with no sign of leakage. The trailer had a new rated axle. The keyway repair ran without trouble. None came back. The manager attributed the improvement less to welding skill, which the workshop had always had, than to reading each failure for its cause, calculating the stresses before choosing the method, inspecting each repair, and recording what was done.
Bringing it together
Each chapter of this book applied a few relationships to one kind of repair. Expected cost, with downtime and the chance of failure, decided whether to repair. Beach marks and fracture zones told fatigue from overload and pointed to the cause. Stress rising with the cube of a smaller diameter separated a good sleeve from a bad one. A carbon equivalent and a preheat decided whether a build-up would crack. Key stresses explained worn keyways. Springback set the straightening load. A tooth’s root thickness squared governed its bending stress. Uneven heating cracked cast iron. And the expected value of an inspection showed what proving a repair is worth. The early repairers had the skill and the courage; the numbers add the judgement of which repairs to make, and how.
A failure: the shaft with no history
This is an illustration. Before the workshop kept repair records, a contractor brought in a shaft that had failed at a journal, and a fitter built it up and machined it. It failed again within weeks, and only then did the contractor mention that another workshop had built up the same journal the year before. The new build-up had been laid over the old one, which was of unknown filler and procedure and, sectioning later showed, had contained lack of fusion that the second build-up had covered.
Had there been a record, or a label, the second repairer would have known to machine off the old build-up completely, or to advise a new shaft. The workshop’s records now go with the part where possible, as a tag and a copy for the owner, as well as in the workshop’s files.
Exercise: inspection on a cheaper part
A build-up on a small conveyor shaft has, from experience, a 4% chance of a crack that would cause failure, at a cost of about A$2,000 in downtime and a new shaft. Dye penetrant inspection costs about A$50 and finds about 85% of such cracks. Is the inspection worth it on average?
Without inspection, the expected cost is about 0.04 × 2,000 = A$80. With inspection, it is about 0.04 × 0.15 × 2,000 + 50 = 12 + 50 = A$62. The inspection saves about A$18 on average, a small but positive return.
The margin is narrow, so the decision might go either way for a part of this value. But the inspection also finds cracks before the part is fitted, when they are cheap to fix, and gives the record evidence of the repair’s quality, so most repairers would do it. For parts whose failure costs much more, as the harvester shaft’s would, it is clearly worthwhile.
