PROJECT DESIGN REVIEW — MINUTES OF MEETING
Project: pallet conveyor for an assembly cell, belt pick-up upgrade
Review subject: timing belt drive selection, design power and safety factor
Date: 19 May, 09:00 to 10:30, workshop office
Document ref: DMR-2026-0519-CNV02
Attendees
| Name | Role |
|---|---|
| F. Lindqvist | Lead mechanism designer (chair) |
| B. Novak | Drive commissioning engineer |
| Y. Hassan | Plant maintenance lead |
| A. Moreau | Conveyor design, brackets |
| G. Richter | Procurement |
1. Purpose of the review
The pallet conveyor moves 14 kg pallets at 0.42 m/s, driven by a 0.75 kW motor through a timing belt to a live roller shaft. The first design used an 8M belt, 30 mm wide, on two pulleys of 30 teeth. The complaint arrived after a month in service: the belt jumped teeth on cold mornings, the motor bracket cracked welds twice, and the maintenance log showed the belt tensioner backed off every two weeks. The review was called to find out whether the belt was merely under-sized, or whether the whole selection method had been wrong from the drawing table.
2. Review comment CN-01: the design power was lower than the real shaft power
B. Novak started with a spreadsheet that the design office had never seen. The conveyor runs four hours a day under varying load, with an average of three pallets on the live rollers plus a surge when two pallets butt. The datasheet motor is 0.75 kW, but the motor runs at 0.55 kW average with peaks to 0.68 kW, because the rollers and the belt themselves consume part of the motor output. The original design had taken the motor nameplate 0.75 kW, multiplied by a service factor of 1.4, and selected a belt on 1.05 kW of design power. The correct design power should add the transmission loss and the starting transient: the motor can pull 2.6 times rated torque for 0.4 seconds at start, which over 4.2 seconds of ramp gives an equivalent RMS power of 1.24 kW before any service factor. The review wrote the corrected number on the board: design power 1.24 kW times a service factor of 1.6 for daily starts and light shock, giving 1.98 kW, which is nearly double the belt was chosen on.
3. Review comment CN-02: belt size and the tooth jump mechanism
Y. Hassan described what a jumped belt looks like at the conveyor: the pallet stops dead in the middle of the line, the drive shaft keeps turning, and the belt teeth are found nibbled on the driver side. Tooth jump is not a fatigue failure, it is a capacity failure. The belt was skipping because the transmitted load exceeded the belt’s ability to hold teeth in mesh on the small pulley. A. Moreau asked what controls that ability, and F. Lindqvist answered: the number of teeth in mesh with the smaller pulley, the belt width, and the belt pitch. On the 8M-30 belt with a 30-tooth driver, only 15 teeth are in mesh, and the catalogue capacity at 1.24 kW real power was found to be 1.38 kW, so the margin was four percent, which is no margin at all once tension drifts.
The tension drift was the second part of the story. The tensioner backed off because the belt stretched as its span heated in service, and nobody re-checked the tension. A stretched belt reduces the wrap pressure on the teeth, and a belt at 60 percent of its design tension loses roughly a third of its transmissible torque. The maintenance log confirmed the jumps mostly happened after the warm-up stretch, which pointed the review at both the sizing and the tensioning practice.
4. Review comment CN-03: what the correct choice looks like
B. Novak pulled the catalogue page for the next family up. Keeping the 30-tooth driver but moving from 8M to an equivalent wider belt, or to a 14M pitch, changes the capacity. The review compared three candidates: the fitted 8M-30 belt at 1.38 kW capacity, an 8M-50 at 2.30 kW, and a 14M-30 at 2.85 kW. The width increase is the cheap fix, because the pulley faces are machined and widening only changes the belt and the flanges, while moving to 14M pitch requires new pulleys and a new bracket. The numbers put the 8M-50 belt, capacity 2.30 kW against a 1.98 kW design power, at a safety factor of 1.16. That is still thin for a belt that is expected to run four years; the review preferred a belt with the capacity about 1.6 to 1.8 times the corrected design power.
5. Review comment CN-04: the cracked weld was a tension problem, not a weld problem
A. Moreau brought up the second complaint, the cracked bracket. The maintenance log blamed the welder. The review read the crack position: both cracks ran from the bottom of the bracket flange next to the tensioner pivot, which is where the belt tension reaction loads the bracket in bending. The original design tension for the 8M-30 belt at 1.24 kW is about 640 N of static tension, but the team measured the fitted tension at commissioning and found the fitter had set it at 1,050 N to stop the slipping, nearly 65 percent over, because the belt was undersized and kept jumping. That over-tension loaded the bracket seam at about 1,500 N of alternating bending force, and a 3 mm fillet weld at the pivot boss was the weakest point in that load path. The fix was not a bigger weld, it was the right belt and the right tension, and then the bracket loads fall back to their design values. B. Novak wrote the numbers against the weld: with the 8M-50 belt at correct tension, the bracket reaction drops to about 780 N, inside the weld capability without any rebuild.
6. Review comment CN-05: tensioning practice and the spring-loaded idler
Y. Hassan asked what keeps the belt tension from drifting again. The old tensioner was a screw-adjustable idler that nobody touched between services. The review decided on a spring-loaded idler on the slack span, sized so the belt is held at the set tension without the fitter guessing by feel. The numbers were agreed on the board: belt static tension for the 8M-50 at the corrected design power is 480 N, the idler spring is sized to apply 180 N of radial preload on the slack side, and the spring travel covers 6 mm of belt stretch before the tension drops below 90 percent of set. The maintenance lead accepted the arrangement on one condition, that the tension check becomes a monthly item and the reading is written in the log, because a belt with a perfect selection and a forgotten tensioner is a belt that jumps again.
7. Review comment CN-06: procurement and the two belt widths
G. Richter compared supply. The 8M-50 belt costs 62 euros versus 38 euros for the fitted 8M-30, and the flanges need widening, quoted at 1.5 hours of machining per pulley. The alternative, the 14M-30 set, costs 210 euros and needs new pulleys at 8 days lead. Against the actual cost of the failure, two machine stoppages, one bracket re-weld, and a maintenance visit that the plant manager had already put at 640 euros, the 8M-50 upgrade at under 100 euros total looked like pocket change. The review chose the wide belt and the existing pulleys, and closed the procurement line with a note that the cheapest belt is the one that never needs replacing.
8. Disagreement that had to be resolved: service factor 1.6 or 2.0
The genuine dispute was the service factor. B. Novak argued for 1.6 because the conveyor has light shock and daily starts, and the catalogue table for this duty class gives 1.5 to 1.7. Y. Hassan argued for 2.0 because the plant runs the conveyor near its peak at shift handover, when two pallets butt and the motor surges, and because the belt runs cold in winter. F. Lindqvist settled the argument with the actual numbers: at a design power of 1.24 kW, a factor of 1.6 gives 1.98 kW and an 8M-50 belt holds it at a 1.16 safety factor; a factor of 2.0 gives 2.48 kW and pushes the 8M-50 belt to a 0.93 ratio, which fails. So the factor could not be raised without going to the 14M set or a 75 mm wide belt. The review split the difference by choosing the 8M-75 belt, capacity 3.45 kW, which holds both 1.98 and 2.48 kW with margin, at a price the procurement desk could live with. The vote passed, and the 14M option was dropped.
9. Risks and mitigation
Three risks were logged. Risk 1: the wider belt needs the pulley flanges machined and could be delivered with a damaged face; mitigation is a face-width inspection on receipt and a trial run of five hundred cycles before the conveyor re-enters production. Risk 2: the spring-loaded idler could be set with the wrong preload in the field; mitigation is a jig pin that fixes the idler arm at the correct 180 N compression during assembly, so the fitter cannot guess. Risk 3: the stretch of a new belt is largest in the first week, so the spring travel could run out before the first service; mitigation is an initial re-tension check after seventy-two hours, logged in the maintenance sheet.
10. Actions and owners
| ID | Action | Owner | Deadline |
|---|---|---|---|
| ACT-1 | Order 8M-75 timing belts, two spares | G. Richter | 26 May |
| ACT-2 | Machine pulley flanges to 75 mm face width | A. Moreau | 2 June |
| ACT-3 | Build spring-loaded idler with 180 N jig pin | B. Novak | 6 June |
| ACT-4 | Re-check tension after 72 hours and log it monthly | Y. Hassan | from 10 June |
| ACT-5 | Rewrite selection sheet with corrected design power method | F. Lindqvist | 12 June |
11. Decision summary
Adopt the 8M-75 timing belt on the existing pulleys with widened flanges. Base the selection on the corrected design power of 1.24 kW RMS plus a 1.6 service factor. Fit a spring-loaded idler with a fixed preload jig. Re-check tension after seventy-two hours and monthly thereafter. The old belt and its drifting tensioner are removed from service.
12. Post-meeting addendum, 20 June
The 8M-75 belt ran a full month without a jump. The belt temperature reading stays about 4 degrees C above ambient, the idler spring stays within a millimetre of its set position, and the maintenance log shows the tension check returning the same number three times in a row, which the plant lead described as suspiciously quiet. The original belt gave the shop its most expensive free lesson, that a safety factor starts at the motor shaft, not at the nameplate, and that the cheapest part of a drive is the one sized twice.
Glossary
Design power: the actual power the belt must transmit, after losses and start transients, before the service factor. Service factor: a multiplier that covers duty class, starts and shock. Teeth in mesh: the number of belt teeth engaging the smaller pulley, which sets the transmissible torque. Static tension: the belt tension set at assembly, which governs tooth grip and bearing loads. Slack span: the belt side with lower tension between the two pulleys.
