The axle counter also has disadvantages. It will not detect a broken rail (which track circuits will sometimes but not always detect)
From a safety case perspective that can be covered by more frequent and thorough track inspections and sometimes an increased rate of rail renewal. Also specific products for the purpose of break detection can and are being developed, perhaps homing in more accurately on the precise position of damage than a track circuit's simple 'break somewhere' within a defined section. Even if electrically based, freed from the track circuits finely calibrated requirement to detect train axles at any point, such alternative systems might be much better optimised for the specific rail break detection task. There's a wide market for such systems, not just axle counter fitted main line railways but also long ore haulage lines in various remote parts of the world where the prospect of thousands of tonnes tipped off the rails in the middle of hard to access mountains or desert and blocking the supply route is a powerful incentive.
Many track circuits, in electrified territory especially, are termed 'single rail'. For these example only one of the rails is series bonded. A break on the other rail will certainly be bypassed by parallel traction bonding so will not be detected at all.
Nor will it detect the track circuit clip that is traditionally put down on the adjacent track if a derailed train fouls it - drivers have to contact the signaller by phone or radio instead.
In modern control centres where full cab secure or GSM-R radio comms is provided the message can be passed verbally just as quickly, even by a driver or guard who may be trapped. Modern signalling then allows the signaller to replace any or all of the signals in an area and a radio mayday to all traffic can also be sent.
Probably the worst problem is that if the axle counter loses power, or during engineering work, or sometimes even if a train stops on the evaluator, it could lose count and declare the section occupied when it is clear with even a small probability of declaring it clear when it is occupied. After these events the axle counter has to be reset manually after physically checking that there are no trains in the section. Obviously this is a major hassle especially if a wide-area problem affects many axle counters.
This is no doubt their 'Achilles heal' but the miscount issue appears primarily to be a problem in permissive platforms at major stations where train detection is often split up into multiple separately indicated short sections so signallers have a better idea of free space available to accept multiple trains for splitting and joining, run-round and shunting operations. In this scenario it is very likely that a wheel may be stopped right above a sensor, and if the system cannot determine which direction the wheel moves away subsequently, that is when the evaluator fails safe and miscounts 'right side' leaving a section occupied when it should have been clear. Reset procedures today usually require a 'sweep train' to be authorised to pass through under verbal authority after which normal operations can be resumed. To get around this, the recent Nottingham area resignalling, which employed axle counters widely for the vast majority of the area concerned, substituted track circuits through the platforms themselves at Nottingham station. The relatively small numbers of short track circuits, confined to a well drained major station area, are likely to be fairly easy to maintain and to keep in adjustment compared to the numerous longer examples distributed along trunk stretches of main lines that were typical of first generation power signalling. Once set up and working, axle counters require significantly less routine testing and adjustment than their predecessors. Perhaps GWML resignalling will employ a similar strategy.
--- old post above --- --- new post below ---
I see there are often delays attributed to the newly installed counters on the GWML upgrade. So at the moment they sound like something else to go wrong and require fixing
The question is are the total signalling attributed delays improving or not? Many previous so called 'signalling failures' may actually have been down to the track circuits anyway, and the replacement systems are probably still on the leading edge of their new system reliability curve, whilst tweaks in installation and maintenance methods are being optimised and any rogue products from a large new batch are being weeded out.