Data-driven maintenance analysis of tramway network: towards digital transformation of Gothenburg tramways
One of the proven applications of digitalization in public transport in recent times is the improvement of maintenance strategies using data-driven methods. A robust maintenance strategy ensures the availability of assets to perform its designated operations maximizing the owner’s revenue at minimum...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/455571 |
| Acceso en línea: | https://hdl.handle.net/2117/455571 |
| Access Level: | acceso abierto |
| Palabra clave: | Tramways Railroads--Maintenance and repair Tram Maintenance Track Switch Contact line Inspection Work order Tramvies Ferrocarrils--Manteniment i reparació Àrees temàtiques de la UPC::Enginyeria civil::Infraestructures i modelització dels transports::Transport urbà |
| Sumario: | One of the proven applications of digitalization in public transport in recent times is the improvement of maintenance strategies using data-driven methods. A robust maintenance strategy ensures the availability of assets to perform its designated operations maximizing the owner’s revenue at minimum costs. This thesis carries out a data-driven analysis of the tram system in Gothenburg with an objective to generate insights on the current maintenance procedures and asset performance which would aid the tram operator in their journey towards digital transformation. The analysis focuses on critical fixed infrastructure assets such as track, switches and catenary and was carried out following CRISP-DM, one of the most common frameworks for data mining. The project analyzed three different data sets – monthly track switching operations, operating restrictions resulting from faulty infrastructure assets and unscheduled maintenance events and historical maintenance records of the fixed assets. Few performance indicators were measured from the switching data such as vehicle passage error rates and rate of manual switching operations. The analysis on operating restrictions focused on identifying the reasons for restriction, its duration, the occurrence of restrictions over time and the associated cost impacts. The most significant part of the analysis was carried out on the past maintenance records available as inspections and work orders. Maintenance performance indicators based on the time incurred to perform such activities were measured. The primary causes of a failure for each asset category were identified. Further, a comparative analysis of inspections done against the standard requirements was also carried out. The analysis found satisfactory performance of switching operations. Regarding track restrictions, a pattern on the number of restrictions over time was observed. The analysis of inspections and work orders pointed out underperformance by maintenance teams and evident shortcomings in data collection. The performance indicators of maintenance teams measured may be used as a benchmark for better monitoring and control. However, they should be subject to scrutiny owing to questionable data quality. Future research should explore the feasibility of employing real-time predictive analytics for maintenance in tram systems based on machine learning. |
|---|