Periodic Reporting for period 2 - GoSAFE RAIL (GoSAFE RAIL – Global Safety Management Framework for RAIL Operations)
Reporting period: 2018-04-01 to 2019-09-30
There is approximately 215,400 km of rail lines in the EU which represent a significant asset. Many of the rail networks in Eastern Europe and in parts of Western Europe were developed more than 150 years ago. Failure of a single asset results in potential fatalities, large replacement costs, the loss of service for sometimes extended periods and reputational damage. The safety level of much of the EU rail network is significantly lower than modern highway infrastructure. Replacement costs for civil engineering infrastructure items such as rail track, bridges and tunnels are prohibitive. Given current economic constraints and the challenges of climate change and population growth it is vital that we maintain safety level and develop optimal ways to manage our rail network and maximise the use of all resources. At present Infrastructure Managers make safety critical investment decisions based on poor data and an over-reliance on visual assessment. As a consequence their estimates of risk are therefore highly questionable and failures of significant assets and derailment caused by obstructions are happening with increasingly regularity. As the European rail infrastructure network ages, investment becomes more challenging. As a result reliability and safety are reduced, whilst user perception of these is negative and the policy move to increased use of rail transport is unsuccessful.
Why is it important for society?
The European Rail Agency (2013) reported that the total number of passenger fatalities on the European rail network was 196, making rail the mode of travel with the lowest number of fatalities. Despite the very encouraging safety record for rail, a number of high profile failures of rail infrastructure have occurred in recent years, with the incidence appearing to increase in response to climate challenges and ageing networks amongst other factors. In addition a significant number of non-passenger fatalities occur each year where the general public interacts with rail infrastructure, with 1284 people being killed at level crossings during the same time period. The focus of the GoSAFE rail project is to provide a near-eradiation of sudden infrastructure failures, provide warning systems for obstructions or intruders on the network and using a sophisticated micro-simulation model allow the impact of safety decisions on network capacity to be determined.
What are the overall objectives?
The GOSAFE RAIL project will deliver and demonstrate a safety framework with the following key objectives:
1. Near eradication of sudden unexpected failure of critical infrastructure.
2. Develop a global safety framework to consider risk assessment across a range of infrastructure assets. The framework is fully compatible with the European Railway Agency Safety Management Systems (SMS) wheel methodology.
3. Increase traffic smoothness through a better understanding of the state of infrastructure across the network and using risk based decision making in concert with traffic flow models.
4. Demonstrate a range of techniques to detect obstructions on tracks including humans, animals, vehicles and inert obstructions such as rock falls, landslides and trees on the track.
- A study on railway safety indicators was undertaken that allowed Global Safety Key Performance Indicators (GSKPIs) to be identified.
- An object detection system based on pattern recognition software using high-resolution digital images was developed by UZ and deployed on the Croatian Rail (HZ) network. This could detect obstructions including people, wildlife, landslides and rock falls on the track.
- A landslide detection system was developed by NGI. Based on micro-seismic (sub-terraneous geophone network) an early warning system was deployed on a 360m section of the Norwegian Rail Network susceptible to ice/rock falls. The system proved capable of also identifying animals on the track.
- An approach in which monitoring data was used to provide a more refined risk assessment for railway infrastructure objects was developed and demonstrated using monitoring data from an Irish Rail bridge.
- A data converter to import Infrastructure Managers files to the OpenTrack format for both infrastructure and timetable information has been created to automatically generate the required input data for the algorithm based on Kronecker Algebra.
- The algorithm based on Kronecker Algebra has been tested successfully on the railway line from Zagreb to Rijeka to evaluate the capability of the algorithm to predict delays of all trains at their final destination.
Decision Support Tool
- A Global Safety Management Framework was developed that integrates risk assessment across asset categories including; slopes and retaining walls, level crossings and bridges, tracks and tunnels and network flow model outputs.
- A report detailing the typical data sources available, key open-data repositories and providing guidance on big-data information for informed decision making was presented.
- A methodology for integrating data into both asset management and safety was developed. The framework prescribes an integrated approach for quality-oriented operational railway planning and describes how different types of data can be collected, analyzed and used in artificial intelligence applications.
There has been unprecedented growth in data available in recent years. Data related technologies have also evolved significantly in recent years, allowing vast amounts of data to be easily brought together and accessed and visualized to reveal new insights and enhance decision making providing an opportunity for an evolutionary step-forward in risk based decision making.
An artificial neural network was used to predict delay propagation across the Irish Rail network.
A advanced micro-simulation model based on Kronecker Algebra was applied to Zagreb-Rijeka train line in Zagreb. The algorithm provided a deadlock free timetable with all trains performing at their technical running time. A particular feature of the model is that traction energy can be optimised in the annual timetable.