How to anticipate the aging of tunnels based on data such as geological nature, dimensions, materials or ratings from inspections? This project illustrates the application of predictive maintenance to underground infrastructure.

stakes

By exploiting digital data from 30 years of inspection records, the objective is to understand how the SNCF Réseau tunnel fleet evolves by calculating an aging rate specific to each structure.
This research work paves the way for the prediction of tunnel degradation and meets major technical, financial and operational challenges through predictive maintenance.

ancient heritage

1,400 tunnels

Aging park

high maintenance costs

inspection of tunnels

Every 5 years on average

methodology

To inspect the condition of the tunnels, each structure is cut along its length into 5 m slices. For each, the damages are listed (nature, size, location, etc.) and digitized. An overall score describing the state of degradation, called coast, is then assigned to each slice.

This coastline evolves over the course of inspections, making it possible to assess and quantify the degradation of a tunnel. Its variation depends on several criteria: geological nature, dimensions and materials of the structure.

innovation

Analysis of the aging of tunnels by Machine learning

Development of predictive models based on geological and structural data to anticipate the deterioration of structures over time.

approach

Predictive maintenance aims to anticipate failures and damages by modeling the degradation of systems. The objective is to predict the evolution of a quantity — in this case, the degradation rate — in order to optimize interventions.
This approach reduces maintenance costs while improving reliability and operational planning.

Data preparation and reliability

Formatting and cleaning a database from multiple sources, including manual surveys that may include biases or human errors.

Analysis of influencing factors

Analytical study aimed at identifying and quantifying the parameters that impact the rate of aging of tunnels.

Predictive modeling

Development of machine learning models exploiting the most relevant factors to estimate infrastructure degradation.

Projection and planning

Simulation of the evolution of tunnels over time to anticipate maintenance operations and plan future work.

a tailor-made solution

The solution developed by OSE uses data from inspection reports to generate new indicators useful for the management and maintenance of the tunnel fleet.

Predictive maintenance paves the way for smarter infrastructure management: budget planning, prioritization of interventions, optimization of resources. Do you want to implement this type of approach? Our experts support you at every stage of the project.

Let's talk

How to predict the aging of tunnels?