Custom Pet Carrier Wheel Systems

Pet carrier production desk · Updated 2026-10-07 · 15 min read

For a custom pet tote or backpack carrier, strap geometry is the first thing to fix - width, padding and load-lifter position decide how the bag carries.

Diameter first: the one number that decides everything else

Wheel diameter is the specification from which the rest of the system follows, and it is almost always the specification chosen last. That inversion is why so many wheeled pet carriers get dragged rather than rolled, and it costs nothing to correct if it is corrected before the pattern is cut.

The physics is straightforward and brutal in its implications. The peak horizontal force needed to start a loaded wheel over a step of height h is approximately N x sqrt(2rh - h^2) / (r - h), where N is the vertical load carried by the axle and r is the wheel radius. Take a realistic case: a 12 kg animal in a 2.5 kg bag gives about 142 N on the axle when the handler is not lifting. Set the obstacle at an 8 mm threshold strip, which is roughly what an airport expansion joint or a lifted paving slab presents.

Run it across the diameters actually offered by wheel suppliers. A 45 mm wheel demands about 168 N of peak horizontal force — more than the vertical load itself, which is why handlers lift instead of pushing. A 55 mm wheel needs about 141 N. A 65 mm wheel needs about 124 N. An 80 mm wheel manages the same obstacle at about 107 N. Moving from 45 mm to 65 mm removes roughly a quarter of the peak push force; moving to 80 mm removes over a third, and it is the single change most likely to turn a wheeled product from nominal into genuine.

What stops every brand simply taking the largest wheel offered is what the wheel costs elsewhere. Diameter is consumed twice: once in the height added to the base, which eats directly into internal height for the animal and can break an airline under-seat envelope, and once in the base depth demanded by the standoff needed to keep the shell clear of the ground. An 80 mm wheel plus its housing typically adds 95-105 mm below the base panel; a 55 mm wheel adds 70-78 mm. That 25-30 mm difference is often the difference between a compliant cabin product and one that only works on the ground.

So the derivation is: list the obstacles in the target channel first (kerb height, expansion joint height, the threshold into the target retailer's fitting area), choose the smallest diameter that clears them with acceptable force, and only then check whether the resulting base height fits the selling envelope. Working in that order costs one afternoon. Working in the other order costs a tooling change.

Reading the target surface before choosing anything

Three channels dominate the decision, and they do not want the same wheel. Airport and station travel is dominated by smooth hard flooring punctuated by expansion joints and threshold strips, which favours a mid-diameter hard tread in the 55-65 mm band. Hotel, clinic, and retail environments add carpet pile and carpet-to-tile transitions, which push toward a larger diameter and a softer compound because rolling resistance on carpet scales badly with small wheels. Festival, park, and camping channels introduce gravel, grass, and uneven ground, where nothing under 65 mm performs at all and an 80 mm pneumatic-look EVA tyre is worth its height penalty.

A brand usually sells into two of the three. Where that is true, specify for the rougher channel and accept that the product is a little quieter and heavier than strictly necessary on the smooth one. Carriers are abandoned in car parks and across grass, not on marble.

Two further channels deserve a note because they are usually folded into the general brief and should not be. Clinic use means short indoor journeys over smooth floors with one threshold at the door, which favours a smaller wheel at a higher durometer, and it also means frequent cleaning, which rules out open bearings entirely. Event and exhibition use means many units moved together across mixed surfaces, favouring the largest diameter the product's internal height permits, and it is the one channel where fleet replacement rather than repair is the normal response to a failure — which changes how much bearing quality is genuinely worth paying for.

Tread compound and durometer: grip, noise, and floor marking

Durometer — hardness measured on the Shore A scale — is the second number, and it is the one most often copied from a supplier's stock list without reference to the floor. The consequences show up as complaints rather than as failures, which is why they survive into production.

Below about 70 Shore A the tread is genuinely soft. Rolling resistance rises because the contact patch deforms continuously, the wheel feels sluggish and heavy, and under prolonged static load the contact patch takes a permanent set — flat-spotting. We have seen carriers delivered to a brand after eight weeks in a container emerge with a measurable flat every 300 mm of tread circumference, which reads as a defective product and is in fact a durometer error.

Above about 90 Shore A the tread becomes hard enough to chatter. On a smooth hard floor the wheel buzzes audibly, transmits every joint into the base panel and therefore into the animal, and begins to mark vinyl and sealed timber. It rolls beautifully and it sounds cheap, which is the worst combination available.

The working band for a pet carrier is 78 to 85 Shore A, and the reason it is a band rather than a number is temperature. Compounds stiffen in cold conditions, so a wheel specified at 80 A for a Nordic brand behaves like an 88 A wheel in a January car park and starts to chatter just when grip matters. Specify durometer at the temperature it will be used, not at 23 degrees in a test room.

Material choice follows the durometer rather than replacing it. Cast polyurethane gives the best wear life and the widest durometer range, which is why it is used on inline skate wheels and why it should be default here. TPE is cheaper and quieter but abrades roughly twice as fast. Solid rubber grips best on wet stone and marks vinyl most. Nylon is a poor choice anywhere in this product: it is hard, loud, and unforgiving on every joint it meets.

Flat-spotting, storage, and the shipping position question

Flat-spotting deserves its own note because it is a logistics defect disguised as a manufacturing one. A wheel left loaded against the same contact point for weeks at elevated container temperature will deform, and the deformity is recoverable only partially and only with time.

The controls are simple. Do not ship or store wheeled units standing on their wheels; ship them lying on a side panel, or stood on a dedicated foot if the geometry allows. Specify a durometer at or above 78 A where a poly rather than a TPE tread is used. And where a programme will definitely sit through a summer in transit, request and pay for elevated-temperature loaded storage testing as part of the sample round — it costs a few hundred dollars and it has repeatedly saved a shipment.

Bearings, axles, and the load path into the baseboard

A wheel revolves around an axle, and the axle does nothing useful unless something imposes a limit on where its load arrives. The most common structural defect in a wheeled pet carrier is not a failed wheel; it is a failed wheel attachment, because the load was delivered into the fabric of the base and the base simply did not have an opinion about it.

Axle diameter should be 6 mm for wheels up to 55 mm and 8 mm above that, in stainless or zinc-plated steel, with flat washers either side of the hub and either threaded-into or backed-by a plate. Where an 8 mm wheel uses the standard 608-size sealed bearing — 8 mm bore, 22 mm outside diameter, 7 mm wide — the whole arrangement is cheap, standardised, replaceable, and rated well above anything this product will ever see. That standardisation matters more than it appears: a bearing that can be bought anywhere keeps the product repairable long after the brand has moved on.

Sealed bearings cost a little more than bushings and are worth it above about 55 mm diameter, because the combination of higher speed and greater load is exactly where a plain bushing begins to wear oval. Below 55 mm, on a smooth floor, at walking pace, a polymer bushing is genuinely adequate and saves real money at volume. Do not specify bushings on anything going outdoors, because grit enters the bushing, and grit plus polymer is a grinding paste.

The load path is then the remaining question, and the answer should be drawn rather than described. The axleplate needs to bear on the baseboard — the same structural panel discussed in our writing on baseboard materials — through a bolted or riveted connection, with the fabric trapped between rather than loaded itself. Where wheels are mounted to a moulded external baseplate, that plate needs to be fastened into the board, not merely into the surrounding shell panels, with at least four fixings and ideally six.

Our default requirement is a static test at 3x rated load for 60 seconds with no permanent deformation of the plate and no tear-out, followed by a threshold-impact rig. A wheel system that passes a static load test but fails after 400 impacts against a 15 mm step will still be returned by a customer, and it is the dynamic test that predicts that.

Fixed axle versus swivel caster: the manoeuvrability trade

Two architectures cover the category, and the argument between them is not ideological so much as arithmetic in component count and behaviour.

A fixed axle — two wheels on a common through axle, inline-skate style — has four moving parts if the bearings are counted and effectively none if they are not. It tracks dead straight, which is exactly what a handler wants in a corridor and exactly what she does not want when turning into a lift. To change direction she lifts slightly and pivots, which is fine with a 10 kg load and tiring with a 20 kg one. It is also the cheaper build and statistically the most reliable, because there is no swivel race to wear and no second axis to develop play.

A swivel caster assembly adds a vertical race, a fork, a bearing set, and a second fixing plate — six-plus additional components per unit, multiplied by four. It gives omnidirectional movement, which transforms usability in a crowded terminal queue, and it introduces two characteristic failures: shimmy at speed when the caster trail is too short, and race wear producing wobble after some hundreds of hours. Swivel casters also want a larger total diameter than their fixed equivalent because the swivel offset effectively reduces the radius of the contact patch during rotation.

Our recommendation for medium and large formats is a hybrid: two fixed wheels at the rear taking the majority of load and tracking the unit straight, plus one or two smaller swivel casters at the front providing steerage. Load splits roughly 65/35, which keeps the tracking behaviour of a fixed axle while allowing the handler to turn without lifting. For small formats under 7 kg of animal, two fixed wheels are entirely sufficient and the added components are not justified.

Why twin wheels appear in this category at all

Twin-wheel casters — two narrow wheels sharing an axle inside a single fork — solve a specific problem: they turn on the spot without scrubbing, because the inner and outer wheel travel different distances without dragging. That is genuinely useful on carpet, where a single wide wheel scuffs badly.

They pay for it in debris. Twin wheels trap hair, grit, and thread between the pair, and once packed the assembly stops rotating freely. In a product used by animals who shed, that is a maintainability problem a specification should acknowledge openly rather than discover in reviews. Where twin wheels are used, specify a fork geometry open enough to clear debris and say in the care instructions that the wheel gaps need brushing.

Tip stability is a number, not a feeling

A wheeled carrier that tips over is a lodging complaint rather than an inconvenience, and it is entirely predictable from two dimensions a brand can measure before sampling: the wheel track width and the height of the combined centre of gravity.

The static tip angle is approximately arctan of half the track divided by the height of the centre of gravity above the axle line. Put real numbers on it. A mid-size wheeled carrier with a 220 mm track and a combined centre of gravity 180 mm up carries a tip angle near 31 degrees. Widen the track to 300 mm and it rises to about 40 degrees. Lower the centre of gravity to 150 mm by moving the baseboard-mounted hardware below the axle line, and the 220 mm case rises to about 36 degrees.

The practical target we design to is a tip angle at or above 38 degrees for any wheeled pet carrier sold above 10 kg of animal, achieved through track width rather than ballast. Weight placed low in the bag helps and is why we push brands toward putting the accessory load in a base-level pocket rather than in a lid pocket — which links directly to the question of how load is distributed through the carrier.

Two further constraints bind at the same time. Track width may not exceed the narrowest door the product has to pass through in normal use, which in practice means 320 mm for a domestic doorway with a unit swung slightly. And the base footprint must remain inside the airline carry-on envelope where the product is sold as cabin-capable, which is a hard limit rather than a preference.

Wheel system specification table

The matrix below sets out the five configurations we specify most often, with figures measured on our own rolling rig at 4 km/h against a 142 N axle load. Static ratings are per wheel; noise is measured as a pass-by figure at one metre on polished tile. Use it to shortlist two options, then prove them on samples rather than choosing from paper.

Two rows in that table carry a warning worth stating in words rather than numbers. The 45 mm inline wheel is the cheapest and reads well on a specification sheet; measured against an 8 mm threshold it demands more horizontal force than any handler will apply voluntarily, and it is the configuration most likely to come back to us for re-specification. The 80 mm EVA wheel is the quietest and the most capable off-road, and it adds enough base height to remove a size from a range aimed at cabin use — choose it knowingly.

Note also that noise does not fall monotonically with softness across materials. The 50 mm pressed-steel row is louder than the harder 45 mm PU row because the hub transmits structure-borne sound into the base panel, which then acts as a sounding board. Hub construction, not tread, is often the dominant term.

Noise, marking, and the hotel lobby test

Noise in a wheeled carrier has three sources, and only one of them is the tread. There is tread-generated noise, which scales with durometer and surface roughness as expected; there is structure-borne noise transmitted through the hub and bearings into the base panel, which then radiates from the largest flat surface on the product; and there is debris noise, which is grit in the bearing or a trapped stone in the housing and which sounds nothing like the other two.

The second deserves the most attention because it is invisible in a component datasheet. A wheel system that is quiet in isolation can be loud once bolted to a stiff base panel with an air cavity beneath it. The remedy is decoupling — a polymer washer or an elastomer grommet between the axle plate and the board breaks the transmission path and typically removes 3-5 dB, which is a halving of perceived loudness for a component costing cents.

Floor marking is tested rather than guessed. Our protocol drags a fully loaded unit back and forth across a light vinyl tile at 40 degrees for thirty metres, then inspects for transfer. Two causes dominate: a durometer above 90 Shore A on any compound, and carbon-black-loaded rubber, which migrates a grey mark onto light vinyl and pale timber regardless of hardness. Both are avoidable at specification stage and neither is fixable afterwards.

The field test we ask brand teams to run themselves is the lobby test, and it takes two minutes. Roll the fully loaded sample across a quiet hard floor at walking pace and listen from three metres away, then across a carpet-to-tile transition. If the approach is audible above normal conversation in either case, the configuration needs decoupling regardless of what the datasheet says. Product reviews describe this experience in exactly these terms, so testing it in the same terms is the cheapest possible development step.

Detachable wheels and the trolley question

Some programmes want wheels that come off, usually so the carrier can be checked in as soft baggage or cleaned without risking the bearings. Detachable assemblies are feasible and our colleagues treat the mechanical detail separately in the note on removable wheels and detachable trolleys.

Two cautions apply to every detachable design we have evaluated. The release mechanism becomes a structural component, so it needs to be load-rated like one and proof-tested at multiples of static load, not just cycled for ease of use. And any detachable wheel is a wheel the customer will eventually lose, so the release should be tool-assisted or deliberately stiff rather than a satisfying push-button that works itself loose.

Field failure modes and how they are inspected at AQL 2.5

Wheel systems fail in a short, predictable list, and every item on it is detectable at inspection rather than discovered in the field. Writing the list into the approved inspection sheet is what converts it from a hope into a control.

Axle tear-out. The plate pulls through the base fabric, usually after repeated impact rather than static load. Prevented by fastening into the baseboard and verified by pull test at 3x rated load.

Wobble. Lateral play at the rim. Tolerance should be set at 0.8 mm measured at the tread edge under a 20 N lateral load, tested on 100 percent of units because it is fast and because it is the defect a customer notices first.

Bearing seizure. Almost always water or grit ingress. Where the product may go outdoors in rain, specify a sealed bearing and add a twenty-minute shower test to the sample round.

Tread de-bonding. The tyre separating from the hub on bonded assemblies. Accelerated by heat, so include an elevated-temperature loaded run for any programme shipping through a hot season.

Flat-spotting. Permanent set after static storage, addressed above by compound choice and shipping orientation.

Marking. Detectable only by the drag test, which is why it belongs in sampling rather than in bulk inspection.

At release we work to AQL 2.5 general inspection level II, and four defects are classed critical for a wheeled product: any lateral play exceeding 0.8 mm at the rim; any crack in a baseplate or axle plate; any wheel that does not rotate freely through a full revolution by hand; any missing fastening. Finish quality is assessed separately, since plating quality on exposed wheel hardware is dealt with in our writing on salt-spray and plating tests for carrier hardware, and it is a standard reference such as those published by ASTM International that defines the salt-fog exposure method we work to.

Programme terms stay constant whichever configuration is chosen. MOQ applies at 500 per colourway and per wheel specification — meaning a programme wanting two different wheel sets effectively holds two colourways' worth of commitment, which is worth knowing before briefing — with samples in 6-10 working days and bulk production 35-50 days after approval. Any product marketed for cabin use should additionally be checked against the carriage framework published by IATA before wheel diameter is locked, since base height, not bag height, is the constraint that usually fails.