Rail Fatigue Life Forecasting Using Big Data Analysis Techniques

Start: January 2017 - End: August 2018

Principal Investigators/Researcher

Dr. Allan M Zarembski (professor) and John Cronin (Graduate Student), University of Delaware

Project Description

Railroad rail represents one of the largest infrastructure costs for railway systems, and is often the largest single maintenance of way expense item. Rail fatigue is one of the primary causes of rail failure, occurring in all modes of rail transit from heavy axle load freight to rail transit. The current approach to forecasting the fatigue life of rail is a cumulative defect analysis using Weibull equations, which allows for the determination of the rate of defect growth and prediction of defect rates based on cumulative traffic levels (defined in terms of Millions of Gross Tons of Traffic or MGT).

While the Weibull equations have been effective in projecting the growth rate of rail defects, they are insensitive to such key parameters as axle load, speed, curvature, and rail maintenance activities such as rail grinding. As such, traditional Weibull forecasts are based on the assumption of homogeneous conditions throughout the entire analysis period and specified location. This has been a very limiting assumption.

With the availability of large volumes of data, it is now possible to extend the Weibull analysis, using new “Big Data” analysis techniques, to allow for more accurate and effective forecasting of rail life. This research activity makes use of a large volume of rail defect data, representing over 30,000 miles of railroad track over a period of almost 10 years, to extend (and possibly replace) the existing Weibull forecasting models to overcome many of the deficiencies of the current model and to allow for more accurate rail life forecasting. These analyses are expected to bring out underlying relationships in the role of infrastructure on the development of rail defects, without requiring an assumption on the nature of the relationship beforehand.

The updated rail forecasting equation(s) then can be used to better predict the rate of rail defect development, allowing more accurate management of the expensive rail replacement process. It should also become possible to gauge the change in the rate of defects due to infrastructure changes, allowing better control over future maintenance costs.