Weigh-in-Motion Explained: How Trucks Are Weighed at 1–100 km/h

01 / Explainer

A weighbridge and a weigh-in-motion lane are asked the same question — how heavy is this truck — under completely different conditions. A static platform measures a vehicle that has stopped, settled and gone quiet. A weigh-in-motion system measures a vehicle still travelling, with its suspension working and its load shifting between axles. One is calm physics; the other is a signal buried in motion. Understanding that gap explains both where the accuracy comes from and where it can be lost.

Why a moving truck is a harder measurement

Start with the ordinary weighbridge. A vehicle leaves the traffic stream, waits, creeps onto a platform, stops, and is released. By the time the reading is taken the suspension has settled and the wheels are still, so the force on the deck is close to the true static weight. The cost is not in the scale; it is in the queue. A lane of land earns nothing, engines idle, and someone has to wave vehicles forward. At a few hundred arrivals a day the queue becomes the most expensive part of the operation.

Now keep the vehicle moving. A truck crossing a sensor is bouncing on its suspension, transferring load between axles as it brakes or accelerates, and leaning on a cambered road. The force at any point in the lane is not a clean number — it rises, wobbles and falls as each wheel passes. Take the peak of that signal and you are measuring the vehicle's dynamics as much as its weight. That is the problem weigh-in-motion has to solve: recover a static-equivalent figure from a signal that was never static. A vehicle being checked also has an opinion about being checked — drivers hug an edge, cross at an angle, weave across the plate, or take an S-shaped detour, changing how much of each wheel the sensor actually sees.

What is buried in the lane

The sensing element in a Jiajia WIM system is a wide-flange plate set flush into the carriageway. It is 58 mm thick and is buried directly in the lane with no civil foundation pit to excavate, so the road is cut, the plate seated, and the surface back in service without a concrete structure beneath it. Rated IP68 for water, dust, de-icing salts and constant traffic, it tolerates 150% overload, so a vehicle well above the limit does not end the conversation.

The method is called full-wheel load integration. Rather than sampling a single instant as a wheel rolls over, it gathers the load across the entire contact event and resolves it into one figure. That is what removes the bias caused by eccentric or irregular driving: a vehicle hugging the edge, crossing at an angle, or weaving still presents the whole of each wheel over time, and the integration recovers the load rather than the wobble. It is an unglamorous idea, and it is the reason the rest of the chain is worth building.

The stated operating envelope is 1–100 km/h — the span the method was designed around, from a crawl at a gate to an uninterrupted highway approach. The wide-flange plate sensor is only the first link; what turns a clean wheel signal into an enforceable number is everything downstream.

From one wheel to a gross weight

Once the plate has produced a figure for a wheel, the acquisition unit does a tidy piece of arithmetic: wheels into axles, axles into axle groups, groups into a gross vehicle weight, all stamped with the same time. The arithmetic is simple; the engineering is in making sure each wheel figure was clean. This is where the two accuracy figures a buyer meets start to separate — gross-weight and single-axle accuracy are not the same number, and a supplier should say which is being quoted.

For our systems the stated dynamic accuracy is ±0.5% on gross weight, with the product handbook expressing the result as Class 1 for total weight and Class E for single axles. Two cautions belong in any purchasing note. That figure belongs to a configured system at a stated speed range — sensor, lane geometry, acquisition and software together — not to a component on a bench; a good sensor in a badly prepared lane still gives a bad number.

Separate the honest errors from the deliberate ones. Honest error comes from vehicle dynamics, an off-centre line, an approach that is not flat, or a speed distribution the calibration did not cover. Deliberate error is the S-shaped detour, the rush across the plate, the drag. Our software holds those abnormal passes for review rather than quietly averaging them away. The technical explainer on how dynamic weighing works follows the same wheel in more detail.

Why speed is a design condition

It is tempting to read 1–100 km/h as a tolerance the system puts up with. It is closer to the opposite: the method was built around a moving vehicle, so speed is an input it expects, not noise it tolerates. A crawl at a controlled gate and a free-flowing highway approach present very different signals, and the calibration has to cover the distribution the lane will actually see. A serious specification names that range rather than asking for "highway speed" and hoping.

Lane geometry is the other half. The flatness of the approach, how the vehicle is guided over the plate, and where the sensing sits relative to a bend all set the conditions the calibration has to hold. A plate buried in a well-prepared straight is a different instrument from the same plate buried after a sharp turn. We would rather settle that drawing at enquiry stage than discover it at commissioning.

Then there is what happens after the pass. A weigh-in-motion record is only useful if it survives the network, the shift change and the review. Our systems support 7×24 unattended operation with no operator at the lane, cache records locally when the connection drops, and can bring a group of sites under one central view. Image capture and video linkage bind the weight to a vehicle, and the software flags abnormal passes that should not be trusted. See the weigh-in-motion solution for how those pieces are configured per site.

What to put in a specification

Write down the vehicle mix and the expected speed distribution. Write down the limits you must enforce and what happens when they are exceeded — a notice to the driver, a directed check, or a record for later review. Say whether the lane is attended, how reliable the network is, and how many sites must be supervised together. Those inputs determine the configuration; a percentage quoted without them is not yet a specification.

Then ask the supplier to confirm the conditions behind the accuracy figure, to provide the drawing package for the lane, and to say plainly who carries installation and maintenance. With us, research, manufacturing, installation and maintenance sit under one team in Yantai, Shandong, China, so a distributor is not left chasing three separate suppliers. We are trusted in 70+ countries, and the commercial terms stay plain: a typical lead time of about 60 days from a confirmed configuration, a minimum order of 1 unit so a pilot lane is possible, T/T payment, and a choice of EXW, FCA, FOB, CNF, CIF, DAP or DDP.

One last discipline is worth keeping. We will not name a certification we cannot show, name a customer who has not agreed, or quote a price before we understand the site. A figure without its conditions is a headline; a figure with its conditions is something you can build a lane around.