High-Speed Weigh-In-Motion systems infer static axle load from a millisecond of dynamic force. We build the sensors, the signal chain, and the calibration discipline that turn that inference into legally defensible data.
Live spec sheet
System 04 · Quartz-piezo, 4-lane
§ 01 — Principles
Everything else — enclosures, backhaul, dashboards — is downstream of these.
Every wheel-pass is a millisecond of force applied to a sensor embedded in the pavement. The system reconstructs the equivalent static axle load from that transient — accounting for suspension bounce, tire imprint and speed.
Two or more staggered weight sensors, inductive loops for speed and class, and axle detectors are fused per vehicle. Redundancy cancels oscillation noise; loops constrain the geometry the estimator needs.
Initial calibration with pre-weighed trucks gets you to spec. Continuous auto-calibration on steering-axle populations keeps you there between service visits. Everything is versioned and traceable.
§ Live · Sense → transmit
Vehicles cross the in-road sensors at highway speed; each axle-pass is measured, reconstructed into a weight, and streamed to the control room in real time.
In-road sensing
Central platform§ Technology · Sensor stack
Choice is driven by required accuracy class, pavement condition, expected axle count per year, and lifecycle budget — not by sensor fashion.
Thin strip embedded in a slot cut into the wearing course. Charge output linear over a wide temperature range. Dominant choice for new HS-WIM installs.
Steel plate with strain gauges, mounted in a shallow frame flush with pavement. Robust, well understood, common across North American installations.
Hydraulic or strain-gauge cells beneath a rigid frame. Highest accuracy class, largest civil footprint. Reserved for enforcement-grade low-speed WIM.
Immune to EMI, no active roadside electronics in the sensor. Emerging class — attractive for tunnels, long linear arrays, and bridge instrumentation.
§ 02 — Data pipeline
Six stages, each with its own QA. Every record is traceable end-to-end.
Sensing
Pavement-embedded piezo / bending-plate / load-cell array
Conditioning
Charge amplification, temperature comp., 10 kHz ADC
Detection
Per-axle peak extraction, outlier rejection
Classification
Axle count & spacing, EU R1 / FHWA vehicle class
Estimation
GVW & per-axle load, corrected for speed & pavement
Reporting
Signed records, APIs, ANPR match, enforcement queue
§ Capabilities
One instrument stack, scaled across three capability tiers — deploy what the corridor needs today and switch on the rest later.
Inductive loops and axle sensors resolve every axle and inter-axle spacing at highway speed.
Per-vehicle speed, headway and lane occupancy derived from timed loop crossings.
Axle geometry maps each vehicle to EU R1, FHWA or a custom classification scheme.
Dynamic force is reconstructed into a static GVW and per-axle load, corrected for speed.
Likely offenders are flagged live and pushed to a downstream static scale or patrol.
License-plate capture is fused per vehicle and matched against the vehicle registry.
Steering-axle populations drift-correct the system continuously between service visits.
Axle-load distributions feed pavement remaining-life and bridge-rating models.
Per-vehicle records, aggregates and chain-of-evidence logs stream to your systems.
§ Enforcement · Overweight → e-ticket
When measured load exceeds the permitted limit, the system captures the plate, builds an evidence package, auto-generates an e-ticket, and dispatches it to the owner and enforcement — end to end, in under a second.
0
Overweight flags · today
0
E-tickets issued · MTD
0.0%
Fine recovery rate
0 ms
Detect → ticket
Detect
WIM · overweight
Capture
ANPR · plate
Verify
Registry match
Ticket
Auto-generate
Dispatch
Owner + LEA
GVW
44.1 t
Limit
40.0 t
Excess
+10.3%
Electronic Ticket · Auto-generated
WIM-2026-004821
§ Performance · COST-323 classes
Wider classes are cheaper and more forgiving of site conditions; tighter classes demand pavement quality, calibration discipline, and often low-speed geometry.
§ KPIs (System 04, ref. site)
Field values from a four-lane quartz-piezo installation, 24-month rolling window.
Accuracy (2σ GVW)
±7% · Class B+
Repeatability
σ < 2.4% on ref. truck
Detection rate
≥ 99.5% at 15–130 km/h
Classification
≥ 97% by EU R1
Availability
99.4% / 365 d
MTBC drift
> 90 days
§ 03 — Applications
Screen the mainline flow, flag likely offenders to a downstream static scale. Static-scale hit rates rise from ~15% to 60–80%.
Ground-truth axle-load distributions feed pavement remaining-life models and bridge rating updates.
Origin-destination freight profiles, modal-shift evidence, real-time congestion attribution by vehicle class.
Instrument the structure instead of the pavement — non-intrusive to traffic and fast to deploy on legacy assets.
§ Impact · Roads & environment
Enforcing axle limits does more than issue tickets — it protects the pavement, stretches maintenance budgets, cuts emissions from premature road works, and keeps traffic moving with fewer stops, shorter queues, and smoother corridor flow.
0.0 kt
CO₂ avoided · per corridor / yr
0.0 yr
Pavement life extension
0%
Maintenance spend reduced
0.0%
Overloaded HGVs · after 12 mo
Overloaded HGVs on corridor
18.4% → 4.1% in 12 months
Relative pavement damage per axle
The 4th-power law
Damage rises with the 4th power of axle load — a 14 t axle wears the road roughly 30× as fast as a 6 t axle. Catching the heavy tail is where road life is won.
×0.13
6 t
×0.41
8 t
×1.00
10 t
×2.07
12 t
×3.84
14 t
Before vs after WIM enforcement
§ Traffic flow & congestion management
Free-flow weighing removes bottlenecks — traffic keeps moving while overloads are caught in the background.
Pre-selection replaces random static-scale stops with targeted diversion. Lane throughput rises, queue lengths fall, and peak-hour congestion hours shrink across the corridor.
Corridor throughput · vehicles / hr / lane
+12% peak-hour flow
2,010 veh/hr
Peak flow
+8 km/h
Avg speed
−94 %
Lane stops
Congestion hours · peak corridor
4.2 → 1.6 hrs / day
Targeted pre-selection diverts only likely offenders — the mainline stays free-flow while queue spillback at static scales disappears.
Before · static stops
After · WIM pre-selection
280→45 m
Queue length
−18 min
Avg delay
−72 %
Spillback events
12 %
Diversion rate
§ Deployment · Site requirements
Site selection accounts for roughly 70% of long-run system performance. Before we choose a sensor, we characterize the road.
Pavement
Rigid or high-modulus asphalt in sound condition. No rutting > 4 mm. Smooth 30 m approach and 15 m departure.
Geometry
Straight, level lane. Longitudinal grade < 1%, cross-slope < 2%. No braking or acceleration zone within 200 m.
Environment
Temperature-compensated. Freeze-thaw resistant sealing. Drainage away from sensor slot. Salt-tolerant hardware.
Power & backhaul
Redundant supply with UPS. 4G/5G primary, fiber where available. Time-sync to GNSS or PTP for evidence chain.
Maintenance
Annual manual calibration. Sensor life 5–15 yr depending on class. Enclosure servicing on 6-month cadence.
Privacy
ANPR retention windows configurable. Purpose limitation and pseudonymization by default. GDPR-aware pipeline.