Sourcing Casters and Wheels From China: Load Ratings Are Almost Always Optimistic

There is a caster on the bottom of nearly everything you ship, and almost nobody gives it a second thought until a customer sends you a photo of a shredded wheel and a torn-up floor. I have bought casters for carts, hospital furniture, tool chests, and industrial dollies out of Guangdong and Zhejiang, and the single most useful thing I learned is this: the number printed in the catalogue is not a working load, it is a hope.

The classic trap is the stack-up. A supplier quotes you a 100 kg caster, you build a four-wheel cart, and you assume you have 400 kg of capacity. You do not. Take the rating, cut it for uneven floors, cut it again for the fact that on a flat surface only three of the four wheels actually touch at any moment, and cut it once more for the shock load when someone slams the trolley into a door frame. On a five-inch caster, serious industrial buyers end up derating somewhere between 40 and 60 percent of the catalogue figure before they trust it. If your cart is going to roll on a warehouse floor with expansion joints, budget for that from the start instead of replacing wheels in the field.

What the wheel is made of decides everything

Polyurethane (PU) is the default for anything that touches a finished floor. It is quiet, it does not mark, and it has good chemical resistance. But PU is not one material — it is a family, and the cheap end uses a low-hardness compound that flat-spots under a static load and cracks at the hub after a few thousand kilometres. Ask for the Shore A hardness by number and ask what the bond between the PU tread and the core is, because the most common field failure I have seen is not a worn tyre, it is a tread that simply delaminated and slid off the hub.

Nylon is the opposite trade-off: hard, cheap, low rolling resistance, and brutal on floors. It is fine for rough concrete in a factory and terrible for anything you would not want scratched. If a supplier offers you "PU or nylon, same price", be suspicious — retail-grade PU wheels usually cost meaningfully more than nylon, so "same price" often means a filled, chalky PU that will not survive the year.

Other options worth knowing: rubber wheels for high shock absorption and outdoor use (heavier and prone to marking), cast iron for the heaviest industrial loads (it will destroy a wooden floor), and specialty materials like high-temperature phenolic for ovens and autoclaves. Match the material to the worst environment the product will ever see, not the showroom.

Bearings are where the money quietly goes

A caster is a wheel plus a bearing plus a mounting, and the bearing is where a reputable supplier and a cost-cutter diverge most. Plain bushings — a metal sleeve, no rolling elements — are the cheapest and make a caster that drags and squeaks. Roller bearings roll far better and take more load. Precision ball bearings are the smoothest and the most expensive, and in most furniture applications they are overkill. For industrial use, ask specifically for a double-ball-race swivel head and sealed bearings, because an open bearing in a dusty plant is a five-month part.

Watch the swivel itself, too. A cheap swivel has a loose kingpin with visible play; grab two wheels and rock them and you will feel it. That slop translates into a trolley that wanders and shimmies at speed. Good casters swivel smoothly with almost no vertical play.

Brakes, mounting, and the spec sheet

Brake quality splits cleanly into two tiers. A simple total-lock brake that pins both the wheel and the swivel is what you want on anything that will be parked on a slope or used as a work surface. The cheap alternative locks the wheel but not the swivel, and the cart still spins on the spot. Test the brake at rated load, not empty — a brake that holds 20 kg is meaningless on a 200 kg cart.

On mounting, get the bolt-hole spacing and plate dimensions in writing with a tolerance, then measure the first production pieces. A friend of mine shipped an entire pallet of top plates that were 1.5 mm off pattern; every one of them had to be redrilled. It is the kind of error that costs nothing at the factory and a fortune at the warehouse.

Before you place the order

A few things that have saved me real money. Ask for the load rating test method and the failure mode at 2× rated load, not just a number. Get a sample of every wheel material you are considering and roll it on the actual floor surface you are targeting. Put a small test order through a real abuse cycle — curb drops, wet floor, a week parked under full load to check for flat spots. And decide early whether your supplier stocks replacement wheels and stems, because a caster that cannot be replaced becomes a whole new product.

None of this is exotic. It is just the difference between buying a part number and buying a caster, and the gap between those two is measured in customer returns.