Locomotive use decisions supported by data

Discovering and making use of potential savings with the help of Big Data - a practical example

08.04.2019 | The efficient use of production resources is a significant cost factor for any company. This also applies for rail freight companies’ (RFC) locomotives. The line for operational and strategic decisions concerning their use is fine and is generally determined based on experience and individual people’s gut instincts. New technologies such as Big Data, Advanced Analytics and Operational Intelligence offer new approaches to making decisions based on data and data analysis. On the basis of a concrete practical example, a description is given of how substantial cost reduction potentials can be calculated and implemented by means of these technologies.

The aim of the locomotive deployment analysis is the even utilisation of all locomotives and a fleet analysis. In order to discover this, some data must be available.

A usage status is assigned to a locomotive. The locomotive is running – it is “in use”. The locomotive is parked but can be used – it is “available”, or the locomotive is “unavailable” because it has a fault or is in the workshop.

There is also the productivity. If a locomotive is travelling for an order and is thus generating income, it is “Productive”. While travelling to the starting point for the next job, its running is “not productive”. A locomotive with no job which is not running is “parked”. Delivery to and return from the workshop were assigned the status of “shunted”. Fault clearance or repair time fell under the heading of “workshop”.

The times of a locomotive have now been added to combinations of deployment and production status. This was the basis of the data analysis. Added to this were operational process data with time-stamp, object identification (locomotive number), locational information (track, station) and additional information (transport time, workshop).

All of these pieces of information flow into the so-called data lake, where they are first indexed and stored with no structure. Using software and algorithms, complex analysis results are presented in a clear and understandable form. This does not remove the need to be clear about what the question to be answered is. Based on the aim of finding cost-cutting potential, the ratio of productive time to parked time was first analysed over the entire locomotive fleet. To achieve this, individual times were aggregated. Two things are striking:

  • the uneven distribution of productive time
  • across the individual locomotives and
  • the high proportion of parked time.

One conclusion is that a proportion of the locomotives is subject to more intensive use throughout the year and is therefore subject to greater wear. Other locomotives establish themselves as “standing locomotives”. Used less, they are nevertheless sent for maintenance at regular intervals. Another conclusion could be that the company has too many locomotives. With this high proportion of parked time, the productive time could be supplied by fewer locomotives.

However, before “snapshots” are delivered, other operational limiting conditions should be analysed. It was thus necessary to establish how often a peak load occurred in the number of locomotives used during a shift. This answer could also be obtained from the database.

Since the data processing and analysis are drafted as permanent, automated processes, they are able to provide their employees in operational locomotive usage planning with a criterion, in the form of the productive time for each locomotive, on a daily basis in order to give them an equal workload. Objectives:

  • less malfunctions and wear
  • a more efficient use of the locomotives
  • cost reductions

By linking the data analysis to the maintenance system, only one additional assessment is required in order to monitor the success.

Further information about our solution zedas®asset can be found here.