Train detection

One of the main safety requirements of a train control system is that, before a train is given authority to move along a section of line, it has to be proved to be clear of other traffic. Thus, the ability to detect the presence of a train on a particular stretch of track is a key enabler for automatic signalling, and hence modern train control.

There are two types of technology generally used for train detection, a track circuit or an axle counter.

The track circuit continuously proves the absence of a train from a given section of track in a fail-safe manner. It cannot absolutely prove the presence of a train, since any failure mode will give the same indication as if a train is present, but, by proving the absence of a train, a clear track circuit can be used to confirm that it is safe to set a route and permit a train to proceed.

As its name suggests, with an axle counter system track mounted equipment counts axles entering and leaving a track section at each of its extremities. This information is evaluated to determine whether the track section is occupied or clear.

Fundamental design principles

With a track circuit system, a section of railway track is normally electrically defined by the provision of insulated rail joints (IRJ) in the rails. A source of electrical energy is connected, via a series impedance or resistance, across the rails at one end, and a detector is connected across the rails at the other end.

If there is no train within its boundaries, the detector senses the transmitted electrical energy and energises a repeater circuit. This conveys the absence of a train to the signalling system (track circuit clear). The metal axles of a train within the track circuit will cause the rails to be ‘short circuited’ such that the detector no longer sees sufficient electrical energy and it changes state, informing the signalling system (track circuit occupied).

Any electrical short-circuit between the rails, whether caused by a train or not, or any disconnection within the circuit (for example a cable being cut or falling off the rail), will ‘fail’ the track circuit and inform the signalling system that the track circuit is occupied. This means that any fault will cause the system to ‘fail safe’ – a good thing. However, it can also lead to spurious results and unreliability if the track circuit is not maintained or set up correctly. How many times have we heard the announcement “Trains delayed due to a track circuit failure”?

Correct operation of a track circuit also depends upon good electrical contact between a train’s wheels and the rails, together with a continuous low-impedance path between each wheel via the connecting axle on the train.

DC, AC and coded track circuits

Simple as the track circuit may seem – detecting a train is just a question of monitoring a short circuit between the rails – there are various ways of powering and controlling the system, and all have their benefits and weaknesses.

The source of electrical energy may be DC, AC at power frequencies (typically 50Hz), AC at audio frequencies (several thousand Hz) or a series of impulses or complex waveforms as used by coded track circuits. Similarly, the detector may be a simple relay, a simple AC ‘vane’ relay or a more complex receiver tuned to a particular frequency or pattern of signals.

On electrified railways, the track-circuit equipment must also work despite the large return currents passing through the rails from the electric traction systems. Some track circuits, therefore, have to be either AC or DC traction immune, or, in some parts of the network, both at once.

In addition, the two rails on a railway are not perfectly insulated from each other. There is always a leakage path between the two through the rail fixings, the sleepers, the ballast and the ground itself. This is called the ballast resistance. Its value is dependent upon the condition of any insulation, the cleanliness of the ballast, and the prevailing weather conditions. It is inversely proportional to track circuit length, with lower values in wet conditions where there is bad drainage and/or contamination from conductive materials. In simple terms, if the track is flooded, the track circuit will show occupied and the signal controlling the section will remain red.

Wet tunnels can be a particular problem, as the conditions can vary quite significantly, and higher values (the lower the resistance the worse the problem, the better the insulation the higher the resistance) may be obtained in dry/clean conditions or during frosty weather. A reliable track circuit must therefore be able to operate over a wide variation of ballast resistance.

One difficulty with adjusting track circuits is knowing the prevailing value of ballast resistance. If a track circuit fails due to wet weather, it may be possible to remedy the situation by reducing the feed resistance. But it is important that the track circuit is re-tested after it has dried out, otherwise a ‘wrong side failure’ may occur with trains not being detected.

This adjustment and testing has to be carried out manually, putting staff out on the railway and, therefore, at risk.

Rust films and contaminants

The resistance through the train’s wheels and axles is also important, as it is the train which shorts out the track circuit. The presence of a light rust film on the rail head and/or wheel results in a high resistance which may prevent the short circuit, and therefore the train detection, from occurring. Very heavy rust films, from prolonged disuse, can result in many track circuits being incapable of detecting trains, especially lightweight trains as they are not heavy enough to penetrate the layer of rust.

The mechanical strength of light rust films is much reduced by the presence of moisture, when the contaminant tends to be squeezed out from the wheel/rail contact patch. Therefore, lightly rusted rails will only be a problem when dry.

This problem is most severe when conditions combine showers with a drying wind, or after prolonged periods without trains. Care needs to be taken after track relaying, when track circuits must not be restored to full operation until a reasonable surface has been created.

Other contaminants that increase the electrical resistance between the rails and the train’s wheels can cause the same problems. Those associated with falling leaves are generally limited to autumn and confined to particular locations, although even some built-up areas can be affected. Leaves are drawn into the wheel-rail interface by the passage of a train where they are squashed into a pulp. This contaminates both the rail and wheel, causing wheel-slip problems as well as reducing electrical conductivity.

In simple terms, reasonably dry weather with little wind will cause the leaves to fall gradually over a long time period and to be reasonably sap-free when they do fall. But gale conditions will lead to a sudden fall of sap-laden leaves, giving rise to the worst conditions.

Problems with coal dust on the rail head tend to be confined to colliery areas, and so this is not the problem it once was. Sand contamination is not so much due to the seaside but is usually associated with slow-moving locomotives using sanders excessively. In each case, the effect is similar to heavy rust.

Problems can also occur with ballast condition issues associated with carbon-based contaminants, and of course heavy rain causing puddles and floods can short out the track circuits completely.

Train issues

Where a thin film of contaminant insulates the wheel from the rail, this can often be pierced by a rough surface. The older style of tread brakes caused the tyres to be roughened at each brake application, whereas more modern disc-braked trains allow the tyres to be rolled into a very smooth surface condition. Therefore, older tread-braked trains provided better track circuit operation than modern disc-braked trains.

Similarly, the axle weight has an effect, as a heavy load will pierce a film more easily. Again, modern lightweight trains (and not-so modern ones, such as Pacers), designed to keep track wear down to a minimum, have more problems than old-style heavy freight trains.

One positive result from today’s crowded railway, however, is that busy lines have little chance to rust, reducing the problem. However, seldom-used branch lines, particularly those in coastal regions, are particularly at risk.

To assist vehicles to shunt track circuits, a device known as the ‘Track Circuit Assister’ (TCA) is fitted to modern trains to induce an electrical potential between the wheelset and the rail head. Typically, a TCA consists of a control unit and aerial with associated tuning unit, mounted between a pair of wheelsets close to the rails.

Insulation

As has been described, any direct metallic connection between the two rails will be interpreted as a train and will cause the track circuit to fail occupied. Therefore, apart from the insulated rail joints used to electrically separate sections of rail, the reliable operation of track circuits requires the provision of other insulators.

At a set of points, for example, there are many of these cross-rail connections – stretcher bars, point motors and heating elements – all of which need to be fitted with insulators, giving rise to quite complex insulator and bonding arrangements.

Damp concrete or wooden sleepers can behave as an electrochemical secondary cell, which can give rise to residual voltage problems with DC track circuits.

Concrete sleepers incorporate a rubber pad under the rail foot and moulded insulations where the fixings bear on the top of the foot. These increase ballast resistance to levels significantly higher than those obtained with timber sleepers. However, the insulations can erode due to the vibration of passing traffic and, consequently, require inspection and periodical replacement – another maintenance headache.

Obviously, steel sleepers are even more of a potential hazard. They are also insulated, but any degradation of that insulation will result in severe problems.

Bonding

Bonding is the means by which the individual components of the railway track are connected together electrically for track circuit purposes. The term also includes the additional electrical connections necessary for the proper operation of electric traction. In order for a track circuit to fail safe (to show occupied) in the event of a bonding disconnection, it is necessary to bond all elements of the track circuit in series, so that any one failure breaks the circuit.

Insulated rail joints can be expensive, both to install and to maintain, especially on tracks subjected to high speed, high axle-weight traffic or where there is an intensive service. Also, in areas of switches and crossings, it may not be physically possible to arrange total series-bonding of both rails.

One solution is the use of audio-frequency AC track circuits which permits the physical limits of an individual track circuit to be defined by ‘tuned’ short circuits between the rails, rather than by insulators in the rails. The track circuits operate at different audio frequencies and each tuning unit is designed to its own track frequency. It is possible, with careful design, to arrange a short overlap in the centre of the tuned zone where both track circuits are effectively shunted.

However, it is not always an ideal solution for complex switching and crossing layouts and, because of the additional complication of significant rail impedance with parallel bonding, audio-frequency track circuits are often unsuitable unless the layout is quite simple.

Track circuits and electric traction

Track circuit arrangements in electrified areas are constrained by the need to ensure safe and reliable operation of both signalling and traction systems. This means that the track circuit must be immune to both false operation and damage by the flow of traction currents through the rails.

It also causes complications because, while the signalling track circuit is broken up into sections by insulated rail joints, the traction current return needs a continuous electrical connection back to the substation.

This explains the need for impedance bonds. These are devices that present a low-impedance to traction current and a higher impedance to track circuit current. In simple terms, they allow traction current to pass along the rail, but stop track circuit current in order to create track circuit sections.

Although track circuits will normally be inherently immune to false operation (wrong side failure) from the presence of traction currents flowing in the rails, any imbalance can create a signal that looks like a track circuit feed. The traction currents can also be of a magnitude sufficient to cause damage to equipment, or a right-side failure of the track circuit.

In DC-electrified areas, the relatively low supply voltage results in high currents returning to the sub-stations via the running rails. In order to minimise voltage drop in the DC-traction supply, all running rails are used for the return of traction currents wherever possible, and therefore double-rail track circuits are used.

In switches and crossings, however, it is not usually possible to bond the track in double-rail form, therefore single-rail track circuits have to be installed. Traditionally, in DC-electrified areas, the track circuits were all AC, using phase-sensitive vane relays, so they could be distinguished from the underlying DC. Nowadays, jointless modulated audio-frequency track circuits are used, reducing the number of insulated rail joints and impedance bonds required in plain-line DC areas.

In 25kV AC electrified areas, traction currents are generally lower than in DC systems and, in most cases, single rail traction return is sufficient for electrification purposes. However, on occasion, increased traffic levels and alternative feeding arrangements may increase the need for both running rails to be used for traction return.

Once again, all track circuits in AC electrified areas were traditionally operated with DC current, although feed and relay components were specifically modified to provide protection from damage and immunity to interference.

In combined AC and DC traction current areas, the choice of track circuits has to be limited to those that are immune to both and do not use frequencies (including harmonics) contained in the traction supply. With the variable voltage and frequency traction packages on modern trains, there can be pretty much any frequency in the rail (or close to it). Together with traction pack and power supply failure modes, the harmonic issues can get extremely complicated.

Modern traction units employing active control methods (such as three-phase drives) can actively generate currents at other frequencies and superimpose them onto the supply. This problem can be designed out, but it’s not easy to avoid critical frequencies and some interference is possible.

Axle counters

In many countries of the world, modern axle counting systems have long since taken over from track circuits, as axle counters suffer from none of the above issues. For example, rails can be under water and trains can still run. Axle counters are now the preferred method of train detection for all new schemes in Great Britain, with systems supplied by both Thales and Frauscher.

One particular advantage of axle counters over track circuits is that they can be overlaid on another detection system (whether track circuits or another axle counter system) for upgrade purposes – whereas usually only one-track circuit can be installed on a section of rail at a time.

Axle counters are not without their problems, however. An axle-counter section cannot be made ‘occupied’ by the use of a track-circuit operating clip to protect a train, nor will an axle-counter system detect a broken rail (although a track circuit will also not detect all broken rails, especially in single-rail track-circuit areas).

When an axle-counter system fails, for example due to a power supply problem, it loses track of how many axles have passed through it. Therefore, for safety, it will always recover and ‘come back on line’ showing the section of line occupied. The section then needs to be proved clear of a train before the axle counters can be restored and reset, which can take some time.

The introduction of GSM-R for emergency communications has provided an acceptable replacement for the non-use of track circuit operating clips.

Signalling engineers were always nervous of track circuits being relied on for detecting broken rails as they were never designed for this purpose. Improved rail integrity and ultrasonic testing has provided a far better method of detecting rail problems and the number of broken rails across the network has reduced dramatically.

Another problem with axle counters is that a right-side failure can occur when a wheel stops directly above the axle counter inductive sensor, known as ‘wheel rock’. The previous section will remain occupied with no train present and the time-consuming process of reset and restore has to be carried out. That can cause difficulties at a busy station, where the problem can also occur for multiple short trains stopped along the same platform. For these reasons, Thameslink has elected to retain track circuits on the core section where there are multiple split sections along the platforms.

Remote condition monitoring

Track circuits will still be used for many years to come, such is the huge job in resignalling the network, so clever asset management and maintenance techniques will be required. One initiative that has helped reliability is remote condition monitoring (RCM). By monitoring the track circuit current, potential failure modes can be predicted and interventions planned before failure. It is not something that is easy to automate, but there have been improvements in track circuit reliability with potentially more to come.

One recent innovation on London’s Victoria line is RCM that allows the new jointless track circuits to be inspected in real time from remote locations, improving reliability. Prior to its implementation, track circuits had to be checked by hand using digital multi-meters, which was a time-consuming task and not conducive to finding faults before they occurred. The new RCM is anticipated to reduce the lost customer hours by 39,000 per annum.

A similar scheme is being implemented in Singapore, but built into the jointless track circuit baseplates to provide even greater asset information.

On the other hand, axle-counter systems also have sophisticated built-in remote diagnostics and this is one example of the digital railway delivering results today.

Both methods of train detection, track circuits and axle counters, have their supporters and detractors. However, for now, the world seems to be moving in the direction of axle counters, that is, until something else comes along. 

Thanks to Mark Glover, head of innovation, Siemens Rail Automation UK, for his assistance with this article.