Approving a Caster Sample and Inspecting the First Lot: What You Can Specify, and What the PO Must Fix

Which load rating governs the sample approval

A caster does not have one load rating. Sources describe static capacity (load at rest), dynamic capacity (repeated rolling load at a reference speed) and shock capacity (momentary peak before deformation), each governing a different failure mode, and they agree the headline figure is usually the static one 2 , 10 , 21 .

The dynamic rating at the real duty cycle is the number that decides service life, because it is the load the caster carries while moving, including acceleration, deceleration and direction changes 2 , 10 . That gap is large enough to change an approval decision.

A caster rated 1,000 lb often has a dynamic rating of only 600-700 lb at 3 mph 2 . A sample that passes a static check can therefore still fail in service, and the buyer should ask the supplier for the dynamic figure at the actual speed and duty cycle before signing anything.

Rated capacity is also stated for manual, walking-speed use on smooth level floors; powered, towed, high-speed, shock-loaded or sloped and damaged-floor applications need additional margin on top of the rating 22 , 35 . Continuous 24/7 service is a separate case, with one source advising a 25-33% derate because ratings assume intermittent service 25 .

Does the sample physically fit the equipment

A nominal wheel size is not a fit. A real caster assembly is defined by a set of interlocking dimensions, and getting any one of them wrong means the caster will not bolt up, will ride too high or too low, or will bind on the floor 17 .

Measure the sample against the approved drawing on every line below, and record the values rather than a pass mark.

• ✓Wheel diameter and wheel width
• ✓Overall height from mounting surface to floor
• ✓Swivel radius or offset (raceway/kingpin centre to furthest edge of wheel)
• ✓Top plate length and width
• ✓Bolt hole spacing and mounting hole diameter
• ✓Mounting type: plate, stem or bolt hole
• ✓Bore diameter if the wheel runs on an axle

Material and construction: what the sample must prove

Wheel material, tread hardness, bearing type, fork thickness, kingpin construction and surface finish are the attributes that decide whether the rated capacity and duty life are achievable, and they are the attributes a supplier can quietly change between the sample and the production run.

The sources give selection factors and example values rather than a determination for any application, so the sample record should state the quoted material and construction and the production lot should be checked against it. Warranty wording is a useful cross-check on what the supplier is willing to stand behind.

Warranties typically cover manufacturing defects such as delamination, incorrect bonding and dimensional non-conformity, but not under-specified load, wrong environment or installation error 8 . If the sample's material or construction differs from the quote, that is a specification mismatch rather than a warranty matter 8 .

Where the application is hygiene-sensitive, certification can carry the material claim. One supplier states its 2309 caster is NSF certified and that NSF programmes require regular on-site inspections of manufacturing facilities 32 .

That is that supplier's position on that model, not a general requirement.

Brake function: accept or reject on the job, not the look

Brake type is a functional match, not a cosmetic one. A wheel brake stops the wheel rolling only, a directional or swivel lock stops the caster swivelling while still allowing it to roll, and a total lock stops both 12 , 27 , 28 .

A sample with the wrong brake type passes a visual check and fails the application, so confirm the required function first and then verify the sample delivers it. Two hands-on checks belong in the sample approval.

Spin the wheel by hand after installation: it should rotate freely for 1-2 full rotations from a gentle push, and stopping within 0.5 rotation indicates an over-torqued axle nut preloading the bearing 8 .

For brake performance, one manufacturer describes a brake test on a 15-degree slope and 5,000 brake cycles, and a noise target below 55 dB in an anechoic chamber 14 , 26 .

Those are that manufacturer's test parameters; the sources do not establish a required magnitude, so the buyer should ask the supplier to state the brake test it applies and the slope it is performed on.

Acceptance limits the sources do not supply, and how to write them into the PO

The sources describe test methods but not the magnitude that counts as a pass.

Two publishers describe static overload tests at different levels: a static weight of four times target capacity applied in a non-moving, mounted position 16 , against 150% of rated load held for 24 hours with a check for structural deformation or cracking 20 .

Neither cites a governing standard, and neither states a deformation limit that defines failure 16 , 20 . Dynamic testing is described in the same way.

One manufacturer cycles more than 100,000 times at about 5 km/h and observes axle and bracket fatigue 20 , while another defines dynamic capacity as rolling repeatedly at a reference speed without premature bearing or tread failure 2 . Both are reported methods, not a single agreed acceptance test.

The practical response is contractual. Ask for the test method behind any load number, not just the number, and confirm which standard the supplier used: ICWM protocols are widely referenced in North America and the EN 12527 to EN 12533 series applies in Europe 22 .

One supplier states its casters are tested to ANSI ICWM standards to perform at the issued capacity 10 . Write the chosen test magnitude, hold time, speed and the deformation or cracking criterion into the PO, because no source supplies a default.

First-lot inspection plan: what is sampled, checked and recorded

An auditable acceptance decision comes mainly from traceability fields and a clear release sign-off, not from long text descriptions 13 , 31 .

The plan below sets what the buyer specifies at each step and the evidence to keep; the sources do not cite a recognised sampling standard, so the sample size is a PO decision rather than an industry default 29 .

Certification and test evidence to collect with the sample

Certification evidence supports the sample approval but does not replace the physical checks. Sources name third-party test reports such as SGS or TUV, material certification such as FDA or REACH, and actual application case data as the documents to request from a supplier 14 .

For hygiene-sensitive use, NSF certification is named on a specific stainless caster model, with the note that NSF programmes require regular on-site inspections of manufacturing facilities 32 . For conductive or ESD applications, sources state that conductive casters are tested for resistance properties to fall within a safe range 33 .

Two commonly requested documents are not named by any source in this set: RoHS and ISO 9001. If the buyer needs them, they have to be requested explicitly rather than assumed to be covered by a caster certificate.

No source states which certifications are mandatory for any specific caster application or market.

Commercial terms that decide whether a rejection is recoverable

The return position has to be settled before tooling or custom production starts. One distributor states that stock items carry a 30-day return policy in unused resalable condition, while made-to-order and custom items are non-returnable once production starts 7 .

If the order is custom, a rejected sample or first lot may not be returnable at all, so the acceptance limits in the PO are the buyer's only recovery route. Sample and lead times are supplier-specific.

One supplier states samples can be prepared to customer specification typically in 3-5 days, with a standard MOQ of 100 pieces per SKU and bulk lead time of 2-20 days depending on order size and schedule 5 .

Those are that supplier's stated terms, not a market benchmark, and they should be confirmed in the order rather than relied on from a catalogue. Traceability and change control belong in the same clause set.

The incoming inspection record should carry part number and revision, PO or shipment reference, lot traceability, sampling plan name and sample size, checklist items, inspector and timestamp, and release sign-off 13 , 31 . That record is what makes a later rejection enforceable against the supplier.

Where the sources disagree

Two points in this decision are not settled by the evidence. Each row keeps both positions visible so the buyer can pick one and write it into the PO rather than assume a default.

What the sources do not establish

• No source states dimensional tolerances, measurement methods or pass/fail limits for the sample dimensions, so these must be written into the PO.
• No source gives a deformation or fracture acceptance criterion for static, dynamic or impact load tests.
• No source addresses fork thickness, kingpin construction detail, or coating thickness and finish specification.
• No source gives numeric acceptance limits for swivel resistance or torque, rolling resistance, or noise; only a below-55 dB anechoic-chamber target appears in one source.
• No source cites an AQL standard or equivalent recognised sampling standard for incoming inspection, so any sampling table is a PO decision.
• No source addresses tooling or mould cost and ownership, or change-notification requirements.
• No source provides Cpk values or historical defect-rate data for casters.
• No source names RoHS or ISO 9001 as caster compliance evidence.
• No source describes first-article approval record content or retention requirements.
• No source states which certifications are mandatory for any specific caster application or market.