Skip to content

247 Application of statistical process control

Published: 24 October 2019

 The aim of the project is to reduce the occurrence of train-disruptive faults in railway switches by achieving the following intermediate objectives:

  1. Identify the correlation between turnaround time, amount of energy and actual faults in the switch during changeover.
  2. Establishing control boundaries using statistical process control for identified correlations between changeover time, amount of energy and actual errors in the switch during conversion.
  3. Check if the control limits had generated alarms before actual fault in the switch occurred.

One method of reducing the occurrence of errors and making maintenance more cost-effective is condition based maintenance, where maintenance measures are planned according to the unit's condition. Previous studies have investigated whether the restructuring time from DS analysis can be used for condition based maintenance. The study identified a probable link between malfunctioning of switches and crossings and the changeover time using statistical process control. To ensure the connection, higher measurement accuracy was needed of data from the changeover of the switch aand crossing. In this project, further study was made possible by providing data with higher measurement accuracy from the Trafikverket and Structon's system POSS. POSS measures the parameters temperature, amount of energy and changeover time and has a higher resolution of data when the amount of energy is sampled every 0.02 seconds during the changeover of the switch and crossing. Based on an iterative process, an approach was developed for analysing data with statistical process control. Analysis resulted in the following shortcomings:

  • POSS data has periods of missing data as well as lack of data during changeover.
  • Incomplete reporting in both 0felia and Bessy and that only one of five incomplete changeovers in DS analysis has led to error report in 0felia.
  • POSS data does not indicate whether a complete changeover has been carried out, which can be supplemented with data from, for example, DS analysis.

The analysis further provided the following recommendations that also answer the project's purpose and interim objectives:

  • It is more advantageous to use the amount of energy per changeover when using statistical process control, since the amount of energy is less auto correlated than the changeover time.
  • Increased opportunity to diagnose faults by jointly analysing changeover time and amount of energy.
  • When coordinating POSS data and DS analysis, the following were identified:
    • POSS data describes the respective switch individual while DS analysis describes the entire coupled switch (two switches depending on each other).
    • POSS and DS analysis need to standardise description of the direction of the changeover.
  • Large differences in the amount of energy needed for switching between normal and deviating main tracks may indicate that there is a need for action.
  • Alarms from established control limits have been identified before actual malfunctioning of the switching device was detected. Implementing these can thus reduce the occurrence of train disruptions.

Stakeholders: Luleå tekniska universitet, Sweco Rail och Trafikverket.

Project leader: Peder Lundkvist, Sweco Rail