Guide · Signalling Asset Monitoring

Axle Counter Faults and Reset Procedures

Axle counters detect whether a section of track is clear by counting wheels in and out rather than passing a current through the rails, and they are now the default train-detection technology on much of the network. They are also the asset most likely to drop a section to occupied and need a reset — and every reset is a controlled, safety-critical procedure. This guide covers how an axle counter works, the fault modes that cause count discrepancies, the family of reset and restoration procedures, and where condition monitoring adds value by trending resets and disturbances without ever touching the vital function.

What is an axle counter?

An axle counter is a train-detection system that proves whether a defined section of track is clear or occupied by counting axles rather than by sensing a train electrically through the rails the way a track circuit does. It has two parts: one or more wheel sensors — also called detection points — mounted at the boundaries of the section, and an evaluation unit (the evaluator) that does the counting and logic. A wheel sensor is clamped to the rail and detects each wheel that passes over it; the evaluator counts the axles that enter the section and the axles that leave it.

The principle is simple and robust: when the count in equals the count out, the net is zero and the section is proved clear; while the two differ, the section is held occupied. Because counting does not depend on the electrical condition of the rails, axle counters are largely immune to the ballast-resistance, rusty-railhead, and poor-shunt problems that affect track circuits, which is a large part of why they have become so widely used — including at level crossings, where they increasingly provide the train detection that drives the warning sequence.

How axle counting actually works

Each wheel sensor contains two inductive coils set a short distance apart. As a wheel flange passes, it disturbs the field of each coil in turn, and the evaluator reads the two pulses. From the order in which the coils are triggered it derives direction, and from the time between them it can derive speed and confirm a genuine wheel rather than electrical noise. The evaluator converts those coil signals into clean axle pulses and keeps a running, directional tally for every section it serves.

The whole arrangement is built and certified to a high safety integrity level and is fundamentally fail-safe: whenever the evaluator cannot positively prove a section clear — a missing pulse, a count it cannot reconcile, a loss of communication with a head, a power disturbance — it holds the section occupied and, depending on the cause, marks it disturbed and requiring a reset. That bias toward the safe side is exactly why axle counters need a well-understood reset regime: the system would always rather hold a clear section occupied than risk releasing an occupied one.

Common fault modes

Most axle counter faults surface the same way — as a count discrepancy, where the in and out totals no longer reconcile and the section is held occupied with no train present. The underlying causes, though, are varied, and distinguishing them is the heart of fast fault-finding.

Fault mode	What it looks like / root cause
Count discrepancy	Count in and count out fail to reconcile; section held occupied with no train present — the common symptom of most underlying faults
Wheel rock / standing wheel	A wheel stops directly over a sensor and rocks or creeps, generating spurious extra counts and leaving the previous section falsely occupied
Sensor head misalignment	Tamping, rail wear, rail replacement, or a shifted mounting clamp changes the air gap and weakens the detection pulse, eventually causing missed or unreliable counts
Cable / connection fault	A break or degraded connection between the trackside head and the evaluator interrupts the signal and is read as a system fault
Electromagnetic interference (EMC)	Traction return current, eddy-current or electromagnetic rail brakes, and overhead line equipment induce noise that can mask or mimic wheel pulses
Power supply disturbance	A supply dip or interruption drops the evaluator into a disturbed state, holding affected sections occupied until reset
Reduced pulse amplitude / damping	Snow, ice, metallic debris, or an ageing sensor reduces detection-pulse amplitude toward its threshold — a gradual drift that precedes outright failure

The important operational distinction is between a hard fault, such as a cable break or a failed head that stops the system detecting at all, and a disturbance that leaves the section recoverable by reset, such as a wheel rock or a brief power dip. The first needs a technician and a repair; the second needs a controlled reset. Knowing which you are dealing with — and how often each is happening — is what separates a quick restoration from a long, blind investigation.

Resets and restoration

A reset is how a disturbed section is returned to a usable state, and it is always a controlled, safety-critical procedure: the reset itself does not prove the line clear, so it is wrapped in rules that confirm nothing remains in the section before, or as, it is restored. Several reset types exist, chosen for the situation and the operator's standards.

Conditional and unconditional resets

A conditional reset only restores the section if the evaluator's own stored record supports it — typically only where the last counted movement was outward, implying the train has left. An unconditional (or direct) reset forces the section to clear regardless of the stored count, which carries the most risk and is therefore the most tightly controlled: it is used only when the section has been positively confirmed clear by other means, under strict authority. Neither reset, on its own, proves the line clear — the surrounding procedure does.

Preparatory (sweep) reset

A preparatory reset, often called a sweep or counting reset, arms the section to recover on the next movement rather than clearing it immediately. After the request is accepted, the section stays occupied until a sweep train passes completely through it; if the axles counted in match those counted out, the section restores to clear. A speed or aspect restriction is normally applied to that first movement so the driver can stop short of any obstruction. It is a common, low-risk way to return a section to service after a disturbance without forcing a clear state.

Co-operative, local and remote resets

A co-operative reset requires a technician on site and the signaller in the control centre to act together — for example each holding their respective reset control at the same time for a set period — so that someone with eyes on the section and someone with control of the signals both agree before it is reset. Resets may also be performed locally at the equipment via a key-operated switch, or remotely from a control panel or signalling VDU. The right combination depends on the operator's standards and the criticality of the location.

What condition monitoring can — and cannot — do

Axle counters are vital train-detection systems, and condition monitoring sits firmly alongside the vital evaluator, never inside it. A monitoring platform does not perform train detection and does not initiate resets; resets stay entirely within the signaller's and technician's controlled procedures. What monitoring does is read the evaluator's non-vital status and diagnostic outputs — section occupied/clear indications, system-fault and reset-required relays, and any diagnostic data the evaluator exposes — and turn them into operational intelligence.

In practice that means a timestamped, sequence-of-events record of every disturbance and every reset, so a section that failed at 02:14 and was swept clear at 02:51 is documented rather than reconstructed from memory. It means trending reset frequency to find the detection points that are quietly degrading. And where the evaluator exposes analogue diagnostics such as detection-pulse amplitude or damping over a maintenance interface, it means watching those values drift toward their thresholds so a head can be re-adjusted on a planned visit rather than after it drops a section in service. The value is visibility and prediction — without touching the safety function.

What to monitor in practice

For a useful axle counter monitoring program, capture the following from the evaluator's non-vital outputs and diagnostic interfaces, enabling only what each installation actually exposes:

Signal	Purpose
Section occupied / clear status	Operational picture and the basis for correlating events across sections
System-fault / disturbance indication	Distinguish a hard fault from a recoverable disturbance
Reset-required and reset-performed events	Timestamped record of every reset for trends and post-incident analysis
Reset frequency per section (derived)	Rank detection points by how often they need attention — the leading fault indicator
Detection-pulse amplitude / damping (if exposed)	Trend sensor health and catch misalignment or ageing before failure
Power supply / UPS health at the location	Tie disturbances back to supply dips (see the battery and UPS guide)
Cabinet door / tamper	Security and a change-of-state log for the location case

Power is worth singling out: a meaningful share of axle counter disturbances trace back to the supply, so monitoring the location's battery and UPS alongside the evaluator's status often explains a cluster of resets at a stroke. That overlap is covered in our guide to wayside battery and UPS monitoring.

Why early detection matters

An axle counter that drops a section to occupied removes capacity from the line until it is reset, and because the reset is a controlled procedure it can take time to authorise and carry out — particularly out of hours or where a co-operative or sweep reset is required. A single nuisance disturbance is an inconvenience; a detection point that disturbs repeatedly is a recurring delay generator and, eventually, a hard failure. Safety and assurance frameworks such as the CENELEC standards for railway signalling (EN 50126 for RAMS and EN 50129 for safety- related electronic systems) push operators toward demonstrable, condition-based upkeep rather than purely periodic inspection.

Trending resets and disturbances, and watching sensor diagnostics drift, is the most direct way to meet that intent for axle counters without going anywhere near the vital function. Across a fleet of sections the effect compounds: catching the handful of detection points responsible for most of the resets, and fixing them on a planned visit, removes a disproportionate share of the delay and the late-night call-outs they cause.

Frequently asked questions