Menu Close
Close

PILOT: Vienna / Linz (AT); USE CASE: Predictive maintenance and asset management of roads.

CONTEXT AND
CHARACTERITICS OF PILOT

This pilot is located on the Austrian motorway network which spans approximately 2,265 kilometres and includes 5,862 bridges and 168 tunnels, all managed under the federal administration of the national road operator ASFINAG. In 2003, the budget allocated to maintenance (717 million Euros) was 43% higher than the amount invested in new construction (499 million Euros), highlighting the significant costs associated with maintaining the infrastructure. 

Currently, there is a huge need for regular KPI collection (measurements of physical parameters related to infrastructure and maintenance) as are performed rarely (usually every 4 years) and after larger interventions (e.g., rebuilding, new constructions). The working area has been defined as follows: 8,5 km long road section near city of Vienna and a bridge near city of Linz.

Scope (What will be analysed/assessed/evaluated by MITHOS? What is ultimately expected to be obtained?)

AIT and JKU will collect both historical (from ASFINAG) and real-time road ground data dynamically collected both from in-vehicle and embedded sensors (Cameras, Lidars, GNSS (Galileo)/IMU, CAN-BUS, etc.). This information will be processed on federated data module, ultimately creating valid data set (time series). Furthermore, an AIT sensor equipped bicycle called BikeStar will be used to obtain the needed data for cycle lane assessment. The common, critical and most relevant data will allow a real-time monitoring and will be tested to be used for specific dynamic quick assessment applications based on AI applications both for cycle and road infrastructure properties of sudden and urgent events/incidents/defects that will shorten reaction times to mitigate critical incidents and sudden damaged areas, optimize road maintenance processes and cost, increase life cycles and reduce congestions times as well as safety risks. On the other hand, all data will be then combined on DSS by innovative AI-based methodologies to predict maintenance measures: precise predictions of road defects and timely planning of road maintenance, extensions of infrastructure life cycles and a higher robustness/resilience of the infrastructure. In addition, IFPEN will use the above data from predictive maintenance actions as an input for the DSS to design the optimal EV charger locations. In other words, this tool will allow redesigning the EV charging infrastructure considering the expected future road modifications. Finally, to ensure the technical and social robustness of the DSS AI-based system, AIT and JKU will carry out a comprehensive approach involving stakeholders to clearly communicate the capabilities and limitations of the system, integrate high quality historical data (ASFINAG) from the complete motorway network and the iterative validation of the model’s performance across various conditions and tests campaigns to ensure reliability and accuracy. IFPEN will investigate the implications of predicted road maintenance operations on possible disruptions in the optimal trip planning for EVs on long-distance journeys. Road closures or road capacity reduction due to road works may potentially alter the optimal trip planning for EV users, and therefore the charging infrastructure for EVs could be tailored for increased travel efficiency and for reducing such disruptions.

Data / tools needed

(i) Image-based data (safety, traffic behaviour); (i) Ground-truth data (infrastructure condition); (iii) In-vehicle data (vehicle dynamics) – JKU research vehicle.

MITHOS validation in use case

  1. KPI Efficiency: time lost per vehicle per km. Baseline: Average speed per section and per vehicle with current individual transport. Target: 30% reduction of travel time;
  2. KPI operating costs: transport operation costs. Baseline: Annual operation/maintenance cost on the current road network. Target: 20% reduction of transport operation costs
  3. KPI fossil fuel consumption. Baseline: Existing material use and work-zone activities. Target: 20% reduction of fossil fuel consumption dur to less usage of road materials and less numbers of work zones/ repairs
  4. KPI infrastructure failure probability. Baseline: Existing life-cycle robustness (years of operation per infrastructure type). Target: Extension of the life cycle by 30% (reduction of failures)
  5. KPI emissions (GHG and other pollutants). Baseline: Average GHG emissions per km (current year). Target: 30% reduction of GHG per km, using the predictive maintenance system
  6. KPI nº of accidents. Baseline: Current crash numbers per km. Target: Reduction of 50% of crashes per km, considering AI-based quick assessment, and 50% reduction of conflict rates (also on work zones).

Stakeholders and end users involved in the use case

ASFINAG will support the implementation of the Austrian pilot. A letter of support has been signed, detailing their support: definition of requirements and user needs, workshop involvement (as well as dissemination efforts).