When a facility automates, the budget and the attention go where the technology is newest: the AS/RS cranes, the shuttle fleet, the robots. The conveyor and sortation network physically connecting all of it — the thing every pallet and every tote actually travels across to get anywhere — tends to get treated as inert infrastructure rather than a system with its own failure modes.
It isn't inert. It's under continuous cycle load for the entire operating day, it has more individual wear points than almost any other subsystem in the building, and when it stops, it doesn't just idle one lane — it can take out the throughput of every automated asset feeding into or out of it.
The backbone carries more traffic than it used to
STIQ's 2026 goods-to-person research tracks the industry moving toward denser, more automation-dependent fulfilment — more SKUs, more zones, more automated assets per square metre than five years ago. None of that density is possible without a conveyor and sortation network sized to move it, which means the backbone in a modern automated facility is carrying more total throughput, and absorbing more total wear, than the equivalent system did a few product generations ago.
That growth in load rarely comes with a matching growth in maintenance attention, because the conveyor line doesn't announce itself the way a stalled shuttle does. It just runs, until it doesn't.
What actually fails
Strip a conveyor and sortation line down to where the wear concentrates, and a handful of failure points account for most of the unplanned stops:
- Belt tracking and misalignment — uneven tension or a drifting idler pulls the belt off-centre gradually, long before it becomes a hard stop.
- Splice failures — the joint where a belt is spliced is the highest-stress point on the loop, and the most common place a belt actually fails outright.
- Motor and gearbox bearing wear — continuous duty cycling degrades bearings steadily, and a bearing running hot is one of the earliest and most reliable signs of an approaching failure.
- Roller and shaft seizure on non-belted sortation — the mechanical equivalent of bearing wear, on the diverter and roller mechanisms that do the actual sorting.
Why it gets skipped anyway
Conveyor and sortation PM often runs on whatever generic interval made it into the maintenance system at commissioning, rather than the specific service schedule in the manufacturer's manual for that exact model and duty class. Nobody double-checked it against the manual because the manual is long, the line was running fine, and there wasn't an obvious reason to go looking — right up until a splice lets go at 2am and takes the whole sortation network down with it, which is a very different outage than a single stalled shuttle contained to one lane.
How we approach it
Condition monitoring for a conveyor and sortation network doesn't need new sensor hardware in most facilities — motor current draw, vibration and drive temperature are already sitting in the PLC and SCADA telemetry the system produces. We watch those signals for the drift that precedes a bearing seizure or a belt-tracking failure and raise a work order while it's still a scheduled repair, not a 2am callout.
That sits on the same foundation as everything else here: a PM schedule built from the actual OEM manual for that specific conveyor and sortation line — the real service intervals for that model and duty class, not a generic default nobody checked against the manufacturer's spec.