Big Data – it’s more than a buzzword. Big data represents a challenge and an opportunity that every industry is looking into. The railroad industry is no exception. How to manage, harness and exploit big data has become a regular part of railroad conferences around the world. In 2015, Dr. Allan Zarembski, Director of Railroad Engineering and Safety at the University of Delaware, organized the first annual Big Data in Railroad Maintenance Planning at the University of Delaware to specifically explore big data in the railroad industry, from practical applications and research programs to advances in data analytics. Dr. Zarembski recently spoke with Interface Journal’s Managing Editor Jeff Tuzik about the upcoming conference and the role of big data in the railroad industry.
Building on the success of last year’s conference, the University of Delaware’s 2016 Big Data in Railroad Maintenance Planning conference features an expanded roster of presenters, including railroad industry operators, suppliers and researchers. “We put together the program by looking at three guiding themes as they relate to big data,” Zarembski said. Those themes are:
- Current industry needs, and near-term goals
- Applications and real-world examples of leveraging data
- Theory and technique: a look at the state of the art in data processing and analysis
Part of the reason the railroad industry is so interested in leveraging big data is that there is a potential for incredible value-added for relatively little expenditure. “In a lot of cases, we’re talking about using data that’s already there,” Zarembski said. Rather than collecting new data from new sources, deep data analysis can provide new information from extant data.
Every railroad collects track defect, rail profile measurement, track geometry and many other disparate data throughout the system. And while each point of data has value in and of itself, there’s also value at a deeper level. Illustrating this point, Zarembski recently delivered a presentation on the relationship between track defect data and track geometry data at the Association of Railway Engineering and Maintenance-of-Way Association’s (AREMA) annual meeting. “Integrating multiple data sets and making sure data cross references isn’t easy,” he said, “but once you start mining the data, you find a lot of hidden information.”
This year’s program also reflects an expanded international presence. Big-data-related topics are relevant around the world; “There’s cutting-edge research going on globally,” Zarembski said. In addition to being an international subject, Big Data is an inter-industry and interdisciplinary study, too. Whatever the industry, there are engineering, operations and maintenance implications when it comes to leveraging big data, Zarembski said.
But while the theory and techniques at the heart of data management and analysis are widely applicable, Zarembski added that none of these tools are simply plug-and-play. “The tools have to be tempered with an understanding of the [railroad] industry, which means you need to get big data experts together with industry experts in order to move forward.” There also has to be a long-term commitment to the research. It can take months just to get a baseline for correlating disparate data before you can move,” he said. Data quality and reliability is also an issue that researchers struggle with, and one that often requires close collaboration with the data supplier to resolve.
Collaboration between researchers and the railroad industry have already been fruitful, Zarembski said, but these collaborations have only scratched the surface of deep data analysis. There’s a lot of research in various big data applications going on right now, and there is no shortage of data in the railroad industry, he added. That means there is a great, and growing research opportunity for railroads that get involved. Zarembski noted that railroads have a good relationship with universities in part because of their commitment to acting as a confidential buffer for the data they are given access to. “We don’t give out raw data – just the aggregate. That’s a cornerstone of our policy, and it allows our research to continue.” Industry suppliers are also collaborating; Zarembski noted that Georgetown Rail, Ensco, and Mermec, for example, have taken advantage of these research opportunities and will present some of their findings and techniques at Big Data 2016.
Zarembski expects the conference to continue to expand as industry/research collaboration expands. The conference provides an opportunity for researches to see that the railroad industry is an interesting environment for data analysis, and for the industry to see that their data can do a lot more work for them than it currently does, he said. Given the potential and growing significance of big data analysis in the industry, Interface Journal looks forward to covering this, and future iterations of Big Data in Railroad Maintenance Planning.
The Big Data in Railroad Maintenance Programming conference will be held December 15 – 16, 2016 at the University of Delaware campus in Newark, Delaware.