An automated storage and retrieval system is not a piece of equipment you maintain the way you maintain a forklift. It is a multi-million-pound single point of failure, wired into every downstream process in the building. When the cranes stop, picking stops. When picking stops, the trucks wait. There is no manual fallback for a facility built around one.
That makes the maintenance programme the thing actually standing between a normal Tuesday and a missed cut-off. Most AS/RS operators know this. Fewer have a preventive schedule that was genuinely built to cover it, rather than assembled from whatever the commissioning engineer had time to write down before they left site.
Why this keeps getting harder to ignore
Automated storage keeps growing as a share of the market it sits in — STIQ's 2026 research on the system-integrator sector puts the global SI market at $34bn this year, on a 10% CAGR toward $49bn by 2030. More of that spend is landing in facilities where an AS/RS or a shuttle system is the backbone, not an add-on, and STIQ's goods-to-person research for 2026 tracks the same shift toward denser, more automation-dependent fulfilment.
More systems in the ground means more maintenance teams inheriting a programme they didn't design, on equipment where the manufacturer's service manual is the only real source of truth. The gap between "we have a PM schedule" and "our PM schedule actually matches what the OEM specifies" is where the expensive failures live.
What actually breaks
Strip an AS/RS down to the parts that fail, and there are four subsystems worth separating, because each has a different failure signature and a different inspection rhythm:
- Cranes and shuttles — bearing wear, wheel and drive-belt fatigue, lift-cable wear on mini-load systems. These are the moving parts under the most continuous cycle load, and the ones most likely to fail without warning if inspection intervals slip.
- Rails and guidance — misalignment and rail fatigue from thermal cycling and repeated load transfer. Slow to develop, expensive to ignore, and one of the few AS/RS failure modes that vibration monitoring catches well before it becomes a stoppage.
- Controls and sensors — encoder drift, positioning-sensor faults, PLC and drive-fault codes. These often present as intermittent faults long before a hard failure, which makes them the easiest category to catch early and the easiest to miss if nobody is watching for the pattern.
- Power and safety systems — battery health on shuttle-based systems, e-stop and interlock circuits. Lower failure frequency, but the category regulators and insurers care about most, which is its own reason not to let it drift.
What a real preventive cadence looks like
The specifics belong to the OEM manual for your system, not a generic blog post — cycle intensity, duty class and manufacturer tolerances all move the numbers. But as a working baseline, most AS/RS preventive programmes settle into a rhythm of quarterly inspection of drive components, monthly battery and health checks on shuttle-based systems, and annual replacement of high-wear parts (wheels, in particular) on lines running high cycle counts.
The point of a schedule like this is not the paperwork. It is catching a bearing that is starting to run hot, or a rail that has drifted half a millimetre out of tolerance, while it is still a scheduled repair — not an emergency one, with a stranded shuttle blocking every pallet position behind it.
Where most programmes actually fail — and it isn't the schedule
In our experience, the AS/RS preventive programmes that fail don't fail because nobody wrote one down. They fail because building the schedule from the OEM manual in the first place takes months nobody has — hundreds of pages, service intervals scattered across sections, procedures cross-referenced against part numbers that need transcribing into whatever system the maintenance team actually uses.
So the schedule that gets built is the one someone had time to build under deadline: the intervals that were easy to find, not necessarily the ones the manufacturer actually specified. Six months later, the six-monthly gearbox service buried on page 148 never made it in, and nobody notices until the warranty claim gets challenged for exactly that reason.
How we approach it
This is the specific problem Reliabilytics was built to remove. Upload the OEM manual for a crane, shuttle or mini-load system and the AI reads it in full — every service interval, procedure, torque spec and safety-critical check — and turns it into a live PM schedule linked to the asset, not a spreadsheet someone has to remember to update.
From there, IoT condition monitoring — either dedicated sensors or your existing PLC and SCADA telemetry — watches for the vibration and temperature drift that precedes a bearing or rail failure, and raises a work order before the failure does. Every completed task is timestamped with labour, parts and evidence, which is the record that actually holds up when an OEM warranty assessor or a compliance inspector asks to see it.
None of that replaces engineering judgement. It replaces the months of manual transcription that usually stand between "we bought a CMMS" and "our AS/RS maintenance programme actually matches the manual" — which is normally the real reason the schedule was incomplete in the first place.