The night lead swept fines off the walkway for the third time and called it housekeeping. By morning the carrying edge had climbed the pulley flange hard enough to chew the cover. Maintenance ordered a splice kit. Nobody had logged chute centering, idler freeze, or skirt gap.
Rip-detection briefs own longitudinal tears and stop authority. This page owns belt tracking as a continuous process variable—the walk that precedes rips, seized idlers, and spilled tonnes.
What actually steers a belt
Loading geometry. Off-center chute discharge, worn deflectors, and impact beds that no longer seat invent a permanent side load. Training idlers cannot outvote a crooked feed forever.
Idler health. A frozen trough or return roll becomes a brake on one side. Grease routes and roll spin checks beat “we will change them at turnaround.”
Skirt and sealing. Skirt rubber that pinches one side hard, or a gap that vents dust only on the opposite edge, teaches the belt a preferred lean.
Structure and crowning. Squared frames, pulley crowns within OEM rules, and splices that do not run thick on one shoulder.
Shift rules that hold
- Walk the discharge and loading zone before blaming “old belt.” Photograph chute wear plates against the last MoC sketch.
- Spin-check critical rolls on a named cadence; a silent roll is a tracking cause, not a spare-parts footnote.
- Adjust training hardware one station at a time and wait for the belt to settle—chasing with three wrenches at once invents oscillation.
- Tie spill events and edge-fray work orders to tracking cause codes, not only to “clean-up.”
This is not pneumatic airlock leakage, not tramp-metal overband dumps, and not rip-camera CapEx alone. It is keeping the belt in the middle of the structure.
If the walkway stays dirty and the edge stays shiny, the belt is already voting. Listen before you buy another splice.