AGV and AMR Caster Specifications: What You Can Fix Before RFQ and What the Supplier Must Prove

Rolling resistance: what to ask the supplier to test

Rolling resistance is a system property, not a wheel-material constant. Load, tread shape, wheel diameter, bearing type and floor properties act together, so a single coefficient quoted without those conditions tells you little 9 .

On a stem caster the wheel must swivel before it rolls, which means the resistance the drive motor overcomes includes swivel torque, bearing drag torque and trail geometry losses 24 . For 3–5 inch casters, swivel torque can represent 30–50% of total rolling resistance 24 .

That is why a quote listing only wheel-material friction understates the load the drive motor sees. General caster data puts the coefficient at 0.01–0.08 lb push per lb load, with breakaway force 1.5–3x steady-state 1 .

Larger diameter and harder materials on smooth floors reduce push force 1 . For AGV caster wheels, polyurethane 92–95A on a forged core is the material called out for low rolling resistance and tight concentricity 3 .

If caster rolling resistance drifts, drive-wheel torque compensates until it saturates and the platform slows or stalls on expected routes 3 . No source in this set gives a rolling-resistance coefficient or force at a stated AGV/AMR rated load and speed with a specific tread hardness.

So put that test on the supplier: ask for push force at your rated load, your route speed, and your floor, with the swivel contribution reported separately.

Per-caster load rating: pick a method and stack the derates

Published ratings are lab values — smooth floor, about 70°F, 3 mph, no impact — and every condition that moves your application away from that reference cuts the load the caster can carry 22 . The derates stack 22 .

Speed above the 3 mph reference reduces dynamic capacity, with one source reporting roughly 25% per mph above 3 mph for standard polyurethane 22 .

Cracked or uneven floors de-rate by 20–30% from shock loading, continuous 24/7 operation by 30–40% against an 8 hr/day reference, swivel operation under load by 15–20%, and side pushing by 25–40% 22 . Use the dynamic rating, not the static one, for any rolling application 22 .

The sources give two different formulas for the base figure, and they do not agree. One is total loaded cart weight divided by 3 plus a 20% safety factor, checked against the wheel material ceiling 19 .

The other is total weight plus a 10–20% safety buffer divided by the number of casters 20 . Both are reported methods, not a single agreed formula, so your RFQ should name the one you used and the derates you applied on top.

The wheel material sets the ceiling the per-caster figure is checked against 19 .

Odometry geometry: match diameter and hold one overall height

An AGV platform mixes driven wheels under motor control with passive caster wheels that free-roll, and the two have different requirements on the same floor 3 . Caster wheel diameter must be spec-matched to drive wheel diameter to keep platform height and sensor mounting geometry correct 3 .

When caster diameters are off-spec, the platform cants and the LIDAR and camera mounts go out of calibration 3 .

The material split follows the function: drive wheels want high friction and tight diameter tolerance, called out as polyurethane 85–92A on a machined hub, while caster wheels want low rolling resistance and tight concentricity, called out as polyurethane 92–95A on a forged core 3 .

Low vibration matters because LIDAR, cameras and IMU sensors are vibration-sensitive, and the wheel material sets the vibration spectrum 3 . All four positions on one cart need the same overall height; mixing sizes makes the cart rock 25 .

The dimensions to measure for the geometry check are wheel diameter, wheel width, overall height, mounting type, plate dimensions, bolt hole pattern and diameter, stem diameter and length, grip ring dimensions, and swivel radius 29 , 30 .

No source in this set gives numeric concentricity, runout, caster offset, swivel friction or mounting-height tolerance values, so those tolerances have to be quoted by the supplier against your sensor stack.

ESD construction: specify the path, not just the tread

Mobile equipment on ESD control flooring needs dissipative or conductive casters or wheels to make electrical contact with the floor, and the components must be electrically connected 4 . A conductive tread on an insulating bearing breaks that path, so the construction has to be specified as a chain.

One documented conductive caster construction uses a conductive polyurethane tread, a conductive bearing pack and an aluminum housing, giving a continuous path from floor through caster to cart frame 28 . Continuity therefore has to run wheel to fork to stem to frame, and the mounting interface must seat with no gap 19 .

For bolt-hole and stem-mounted casters, the mounting structure must hold the top washer, round plate or mounting shoulder firmly against the mounting surface at the load-bearing interface 19 . Conductive casters are classed at 10^3–10^5 ohms resistance 12 .

That is a caster impedance class, not the same measurement as the ANSI/ESD STM97.1 resistance-to-ground limit of less than 1.0x10^9 ohm for a flooring/footwear system. Which also requires body voltage under 100 V in a walking test 4 .

The two address different measurement points, so do not treat one as proof of the other. Dry conditions can increase static build-up and affect ESD caster effectiveness 26 .

No source in this set gives a resistivity range for the wheel, bearing and bracket individually, or cites IEC 61340 or an equivalent ESD test standard for casters. So demand the supplier's own resistance measurements at each interface.

Mounting interface and envelope: check against the chassis you already have

Plate versus stem, bolt pattern, overall height, wheel diameter and width, and swivel radius all have to be checked against the existing mounting holes and surrounding jigs before you order. Re-check equipment height after selection, because the caster sets it.

• ✓Confirm mounting type — plate, stem or bolt-hole — against the chassis interface.
• ✓Measure wheel diameter and wheel width; both are always needed.
• ✓Measure overall height from floor to top of mounting.
• ✓For plate mounts, record plate length × width, bolt hole pattern and bolt hole diameter.
• ✓For stem mounts, record stem diameter and stem length; for grip rings, ring diameter and stem diameter.
• ✓Confirm swivel radius — the space from the swivel vertical centerline needed for a full 360° turn.
• ✓Check that all four positions share the same overall height so the cart does not rock.

Wheel material and environment: match the floor, then evidence the claims

Material choice follows the floor. Polyurethane protects hardwood, vinyl, epoxy, tile and concrete, rolls quietly and resists oils, chemicals and moisture; its noted limitation is ultra-smooth polished tile, where a softer TPR wheel may be needed for grip 14 .

Smooth concrete floors perform best with hard polyurethane, while uneven or outdoor floors call for rubber or pneumatic, and epoxy or delicate surfaces call for non-marking wheels 33 . Hard casters such as nylon or steel should not be used on finished or delicate floors because they scratch and indent 14 .

Where oven heat exceeds 200°F, phenolic is the recommended wheel material 6 , and phenolic resin is rated to 280°C 11 . Nylon is called out for chemical resistance and the lowest rolling resistance 6 .

Environmental durability is tested rather than assumed: salt spray per ASTM B117 for 72 hours, chemical corrosion across pH 2–12, and high/low temperature cycles from -30°C to 80°C 23 . Those are test conditions, not a caster rating.

No source in this set gives an IP rating or washdown capability for a specific caster. So if your route includes washdown or a wet zone, ask the supplier to state the rating and the test behind it rather than accepting a material claim.

Commercial terms: normalise the quote before you compare it

Cost is driven by material and construction, load capacity, wheel size and type, bearing type, swivel mechanism, custom features, brand, quantity and market factors 21 , 34 .

Because those variables move together, two quotes are not comparable until they are normalised against the same load, wheel diameter, brake mode, mounting dimensions, tread requirement, quantity and documentation request 7 .

A lower unit price can cost more if the caster forces frame changes, raises push resistance, marks the floor or misses the required locking behaviour 7 . Ask what the quotation includes: sample cost, tooling or customization charges, packaging, test reports, lead time, and the order quantity the unit price is based on 7 .

Conductive casters typically range $25–$100 per unit depending on material quality and manufacturing process 36 . On custom work, one supplier states that custom molds remain buyer property and are stored securely, NDAs are available.

And free replacement or rework applies if specs are not met, with optional SGS or Bureau Veritas inspection 35 . No source in this set gives AGV/AMR volume unit pricing, tooling cost, MOQ, lead time or spare availability, so those terms have to be negotiated against your own volume and service plan.

Test evidence and documents to require with the caster

Ask for dynamic load and endurance evidence, not just a catalogue rating. The ICWM protocol rolls the caster at 3 mph on smooth concrete or steel for a minimum of 10 miles under rated load without failure.

And counts bearing seizure, wheel deformation beyond tolerance, kingpin failure and raceway brinelling as failures 8 . ISO 22883 covers industrial casters for hand pushing or power dragging at no more than 1.1 m/s (4 km/h) and includes more than 50,000 fatigue cycles 17 .

The two standards use different reference speeds, and both are lab reference conditions rather than application limits 8 , 17 . Documentation packages can include material certificates, inspection reports and certificates of conformity, with optional third-party inspection through SGS or Bureau Veritas 35 .

Where the application is pharmaceutical or cleanroom, ISO 14644, FDA, USP Class VI and NIST calibration requirements may apply 16 . Brand benchmarking tables list EN 12530, ISO 22883 and anti-static options, and advise requesting dynamic load test reports and verifying anti-static compliance 37 .

No source in this set cites IEC 61340 or an equivalent ESD test standard for casters, so an ESD caster certificate has to be demanded from the supplier as their own test data.

Where the sources disagree

These are the two places where the sources state different figures for the same item. The load formulas are two reported methods, not a single agreed rule, so the RFQ has to pick one.

The ESD figures are not the same measurement: one is a caster impedance class, the other is a flooring/footwear system resistance-to-ground limit, and they address different measurement points 4 , 12 .

What the sources do not establish

• No rolling-resistance coefficient or force at a stated AGV/AMR rated load and speed with a specific tread hardness.
• No caster-specific load/speed derating curve at an AGV/AMR duty cycle.
• No numeric concentricity, runout, caster offset, swivel friction or mounting-height tolerance values.
• No resistivity range for wheel, bearing and bracket individually, and no IEC 61340 or equivalent ESD test standard for casters.
• No AGV/AMR-specific mounting envelope or clearance values.
• No IP rating or washdown capability for a specific caster.
• No AGV/AMR volume unit price, tooling cost, MOQ, lead time or spare/replacement availability.
• No ESD test standard certificate or inspection report for a caster.