Line 02

Track maintenance optimisation — Headway Track

Ranks inspection, tamping, grinding and component renewal by risk and operational criticality, within the budget and the windows the operation actually has.

For whom
Concessionaires, new entrants and operators with their own network
Timeline
Pilot on a delimited perimeter before any scope expansion
Commercial model
Implementation + annual licence + variable component on measured gain

Context

Track maintenance decisions rarely fail for lack of prediction. They fail from an excess of candidates and a shortage of criteria: more segments ask for intervention than budget, crew and window can serve, and the queue is ordered by apparent urgency, history and internal negotiation.

The product does not exist to predict failure. It exists to improve allocation under real constraints. Degradation forecasting is an input; the deliverable is the prioritised sequence, with a suggested window, a rationale for each recommendation and the cost of not acting.

Every recommendation carries a confidence level and a calculation trail. Below the threshold the system states that it does not recommend — explicit abstention is a product requirement, not a limitation. The client’s engineer must be able to validate or reject each line, and that decision is recorded in the audit trail.

What you get

What you get

01

Risk and criticality ranking

Segments ordered by predicted degradation combined with the operational effect of failure — not by asset condition in isolation.

02

Degradation forecast by segment

Models calibrated per network and load regime, fed by geometry, testing, incidents and intervention history.

03

Intervention plan with window

A sequence of tamping, grinding, welding and component renewal compatible with the real circulation grid and available crews.

04

Budget scenario simulation

What changes in the plan — and in the risk taken — when the budget rises, falls or is brought forward. The curve, not a single number.

05

Explainability and audit trail

Rationale per recommendation, model version, input data and the decision recorded by the client, retrievable at any time.

When this applies

Situations in which this line of work is the right starting point.

  • The annual plan is built from history and negotiation rather than comparable criteria across segments.
  • Track kilometres under speed restriction do not fall as budget is applied.
  • The maintenance window is the scarce resource and the service queue is not ordered by effect on circulation.
  • The same decision must be defended before auditors, a regulator or a board.
Discuss your case

FAQ

Frequently asked questions

Does this replace mandatory inspections?

No. Inspections and tests required by regulation remain mandatory and are a data source for the model. The platform prioritises and sequences decisions; it does not replace regulatory obligations or issue technical certificates.

Do we need new sensors?

Not as a rule. The work starts from existing instrumentation — inspection cars, non-destructive testing, incidents, work orders. New instrumentation enters through partners, when the data gap is genuinely limiting.

Does it work on metre gauge?

Models are calibrated for the network they operate on — gauge, load regime, climate and maintenance practice. Metre gauge with mixed traffic is the starting case, not the exception.

Next step

How much of your maintenance budget is well allocated?

The conversation starts with the problem, not the platform: which decision is made today without enough data, and what the order of magnitude would be if it improved. Thirty minutes, no corporate deck.