An Overlay for the Headlines, Nothing for the Tracks

A device the world has run for decades, Nine years on Indian files, Six tenders, Not one unit on an Indian train

Editorial note: We invite the stakeholders mentioned here to share their perspectives and versions of events—we assure to give them space fairly. These stories have been carefully curated by senior and highly respected retired railway officers who are committed to avoiding sensationalism.

Two safety systems. Two fates. One is on every poster, in every speech, fed a budget that will one day run into tens of thousands of crore. The other was specified nine years ago, funded, tendered six times, and is still not on a single train. The first is a collision overlay. The second listens to the track itself. Guess which one the system loves.

What the overlay does, and what it cannot

Begin with what #Kavach is, and is not. It is an automatic train protection system that sits on top of the existing signal interlocking—a layer, not a new nervous system. It reads the signals already set and acts on them: it catches a driver who passes a signal at danger, it catches over-speeding, it can halt two trains closing head-on or nose-to-tail on the same line.

Now the limits, because the limits never reach the poster. Being only a layer over the existing signalling, it inherits that system’s blind spots and adds few new senses of its own. It does not read the track. It cannot feel a developing geometry fault or a rail fracturing under load, so it cannot stop a derailment. It is built for trains meeting on the same line; a coach flung across the adjacent track, a side-on or oblique collision, is not a scenario it is designed to see. And the version now being bought in bulk—4.0—is still finishing its passenger field trials even as thousands of units are ordered against it. The procurement is running ahead of the proof.

And the road beneath the wheel is where most of the dying happens. The national auditor put it on record: “roughly seven in ten accidents in the audited years were derailments, the leading cause track maintenance and parameters drifting past limits. The largest category of accident sits entirely outside the system the railway celebrates—a layer over the signals for the wound the railway rarely suffers, and nothing for the wound it suffers most.”

The device that listens to the track

There is a system for exactly that wound. As a train runs, its #axle-boxes record the rhythm of the rail. A fault shows up first as an abnormality in that rhythm, long before it shows up as a heap of coaches in a field. Measure the rhythm continuously, flag the abnormal stretch, pull it for inspection, and you have caught a derailment before it was born. Not experimental—railways abroad have run it for decades.

India knew this. In 2017 the standards body went to Europe to study it, published a domestic specification the next year, and set aside funds for forty units. The process began.

That was nine years ago. Not one unit is on the rails.

Nine years. Six tenders. Zero.

Read the procurement record and a sick comedy unfolds. Six tenders since 2019. The first four collapsed without a contract. One was discharged after the eligibility bar was redrawn. One died before it even opened, when the Board decided a research body should not be running tenders at all. One was killed a month before its own closing date for “administrative reasons” that were never explained. One drew the largest field the device had ever seen, ran a six-month evaluation through two rounds of committee, and was then aborted with no result.

Six tenders. Pre-bid conferences. Committees convened, bids opened, firms cleared and then un-cleared. Every time, the same outcome: nothing reaches the track.

The pattern points one way

A run of bad luck looks random. This does not; the six lean the same way.

The lone indigenous developer disclosed its credentials at a pre-bid meeting. The eligibility bar moved afterwards, and moved it out. Relaxations it asked for were refused. In a later round it was failed for a missing document while others missing the same document were waved through; the moment it objected in writing, the whole tender was discharged. And when an award finally came, it came on a bid propped up not by a commercial purchase order from any railway, as the tender demanded, but by a foreign research grant for a short academic project. The technical reviewers flagged the substitution and asked the committee to examine it. The committee declared the grant a “contract” and examined nothing. The clause that existed precisely to reject such a document went missing from the committee’s own minutes. Then a sixth tender appeared within weeks, its eligibility quietly widened in a direction that suits the same favoured profile.

Each step, read alone, is a paperwork quibble. Read together, they are a current. And currents have a source.

The overlay’s own house is not in order

Here is the irony that should burn. The celebrated system is caught in the very pattern that strangles the cheap one. This publication has reported, across a three-part series, that the conditional approvals granted to the Kavach makers carried a hard clause—finish mandatory testing within 180 days, or the approval lapses on its own. The window closed. The testing, by those accounts, was not finished. Yet the names stayed on the approved list, and fresh tenders for thousands of units were rushed out—by the reporting, timed to lock in eligibility before the approvals could die.

One order ran to ten thousand units, placed on a maker that could build only a fraction of that by the deadline. Against an earlier order for a few thousand, a few hundred were supplied and barely a score actually installed—the sheds, reportedly, turning the rest back over hardware and software faults.

Create the urgency. Invoke safety. Float the big tender. Bend the terms to the bidder who cannot deliver. The favour flows. (See this publication’s Kavach series: “Race against time to extend undue favours”; “Railway Management giving undue favours to the companies”; “Allegations of Rushed Tenders to Benefit a Few Companies”.)

Follow the priority, follow the money

The contested award for the track-listening device is worth about one and a half crore rupees—for the primary safety tool aimed at the biggest killer on the network. And it cannot clear a tender committee in nine years.

The overlay, meanwhile, already carries a sanctioned line near four thousand crore, with the real programme aimed at forty-odd thousand route kilometres and a bill that will climb into the tens of thousands of crore. Barely deployed, live on a sliver of a sixty-eight-thousand-kilometre network, absent from the very crash sites that put it in the headlines—and the version being installed is still in trials—while the profits booked by protagonists are already eyebrows. Officials credit it for falling accident numbers—a claim its thin footprint cannot yet carry. A system that photographs well gets the flagship treatment and the indulgence over lapsed approvals. A system that simply works waits in a file, tender after tender, for nine years.

The dull cure, again

The fix here is unglamorous, which is why it is ignored. Deploy the cheap thing that reads the track. Run the tender straight, on the conditions as written, and let the lowest compliant indigenous bid win or lose on merit. Stop confusing the system that looks like safety with the system that does safety. Put the auditor’s finding back at the centre: most of the dying is on the track, and only tools that watch the track will stop it.

Until then the overlay gleams on the slides. The forty units sit in a file. And somewhere a rail is fatiguing, quietly, with no axle-box listening, waiting for a curve to be unkind.

#Sources: CAG Performance Audit, “Derailments in Indian Railways” (2022); Ministry of Railways / PIB releases and budget statements on Kavach allocation, route-kilometre coverage and accident trends (2024–2026); RDSO Kavach 4.0 approval, vendor field-trial and SIL-4 certification records (2024–2026); ICF and South Central Railway #UABAMS tender eligibility documents—disclosed evaluation records; #Railwhispers Kavach series, Parts I–III (Nov 2025).