Network Rail needed to modernize its data analytics solution for asset management and maintenance. The company’s Intelligent Infrastructure program chose Microsoft Azure to help build several new solutions, bringing together multiple data streams for asset management. One of the tools gives maintenance engineers better insight into the 20,000 miles of railway track and countless other assets that make up the network. Today, employees use the solution, called insight, to understand data 50% faster than before and improve efficiency, passenger experiences, and safety—all while saving costs.
Network Rail oversees a dizzying amount of infrastructure across England, Scotland, and Wales: 20,000 miles of railway track; 30,000 bridges, tunnels, and viaducts, many of them dating back to the Victorian era; and a vast network of signals, intersections, and stations. Greater still are the number of people and businesses who rely on the railway, from commuters traveling into the city for work to the freight trains that carry essential goods around the country. The railway is an always moving, constantly improving ecosystem that needs around-the-clock maintenance and monitoring.
Martin Mason, Senior Program Manager of Intelligent Infrastructure at Network Rail, describes the organization as one on the cusp of major change. “We are at a crossroads here at the railway,” he says. A drive to innovate, and to harness the paradigm-shifting potential of data, comes from increased demand on the railway and a growing awareness that entrenched ways of working were not as effective as they could be. “In the past, we were very data-rich and intelligence-poor,” Mason says, describing how the nearly half a petabyte of data that Network Rail collects weekly was kept siloed and hard to interpret. For instance, multiple measurement trains run through the network daily, collecting data every 200 millimeters. “We often printed out that data at the end of the day, which could give us a baseline, but it was very difficult to visualize and build the necessary degradation curves,” Mason explains. Patrick White, Track Maintenance Engineer at Network Rail, describes how employees like himself would review separate systems in isolation. “You had to have multiple screens up, trying to do a proper root cause analysis,” he remembers.
Collaborating with Microsoft to bring data streams together and expedite processes
Network Rail’s Intelligent Infrastructure program worked to optimize processes by creating a Microsoft Azure–based solution, called “insight,” that captures the wealth of data produced by railway assets. It helps employees predict and plan maintenance, in turn saving time, improving employee and customer experiences, and making the railway a safer, more efficient operation. “Intelligent Infrastructure is a collection of applications which leverage some of the very best data that Network Rail collects, and some of the very best data that we’re able to buy in the open market, like weather data,” explains Jacqueline Young, Program Director of Intelligent Infrastructure at Network Rail. “It’s a suite of products to make better, more-informed decisions for maintenance and for asset management.”
Today, railway engineers use insight, built using a combination of Microsoft Azure Data Factory, Power BI, Azure Databricks, and Azure Machine Learning, to explore years’ worth of data to understand how assets are performing and predict when they might need to be replaced. “Before insight, we were unable to manage that volume of data and to produce the outputs that we do today,” recalls Mason. “It’s been a game-changer.”
Building a complex, predictive view of the railway network, and all the thousands of assets within it, requires the interplay of countless data sources. In the past, bringing together data from, for example, track geometry and ultrasonic rail testing in a single view was a time-consuming and difficult process. “We called this kind of work ‘separate projects,’ and they could take up to five months to come to a conclusion,” says Mason. “insight represents the first time that that level of detail and cross-pollination of data sources has been widely available in near real time to our users nationwide.”
To create insight, Network Rail worked with Microsoft architects and engineers to perfect the solution. “The Microsoft technical teams we worked with have been absolutely fantastic,” emphasizes Young. “Any issues we faced were met with professionalism and a ‘rolling up our sleeves’ attitude that has been really refreshing.” Network Rail took an agile approach to developing the solution and used shared environments to conserve costs. “We’ve delivered significant efficiencies by sharing environments, and we’ve saved a considerable amount of money too,” says Young. Using Azure Databricks also helped expedite the project, notes Mason. “Azure Databricks changed the game in terms of writing code, the speed of processing, and then the ability to manage the size of the datasets that we’ve got,” he explains.
Prioritizing safety with predictive capabilities
For everything Network Rail does, safety is top of mind. “I am really proud of how we have harnessed Microsoft technology to make the railway a safer place for our workers and our passengers,” says Young. The predictive capabilities of insight help improve safety by preventing unexpected failures that disrupt travel and force maintenance employees to solve issues on the fly. “We are now able to project deterioration years into the future,” explains Mason. The implications of that future-looking perspective are simple but powerful. Young elaborates: “Our Azure-based solution, insight, can detect potential failures earlier than ever before—in some cases, up to a year before the fault is likely to happen. This gives teams the opportunity to plan the right work at the right time and the right place and avoid delays for customers.” This foresight can mean the difference between an unexpected failure requiring maintenance late at night and a planned maintenance operation at a convenient time of day for employees and customers. Improved insights also mean cost savings. Says Young, “By detecting and fixing issues earlier, we avoid the need to implement emergency speed restrictions, which can cost around £250,000 for busy sections of the network. And, world-leading cyclic top (repetitive pattern of dips in the rail track) detection gives teams the power to prioritize the use of expensive on-track machines that can cost £10,000 a shift.”
White also affirms that the ability to understand a fault in greater detail before sending in a maintenance team often results in more expedient, accurate repairs. “By presenting all the necessary information in one place, and layering machine learning on top, we are better able to understand details and the scope of repairs, meaning we can prepare maintenance teams adequately,” he says.
Conserving resources and understanding data 50% faster
Along with boosting safety, creating efficiencies was another driving force behind insight. “The technology we have built with Microsoft Azure and Power BI allows our engineers to digest and understand the data 50% quicker than they could before,” explains Mason. “With it, we have achieved massive efficiencies for the workforce.” Those efficiency gains translate into more time for high-impact work. “Now, users can spend more time doing the things that matter,” says Mason. “Rather than trying to manage paper and spreadsheets, they can go out and maintain the railway.”
Being better able to plan maintenance work helps Network Rail conserve resources too. “When we created this system, we were thinking of how to make cost-effective decisions for the entire system, rather than fixing individual faults as they arise,” recalls Young. Employees use insight to plan the use of resources like ballast and rail to improve outcomes and be proactive, rather than reactive, with repairs. “It’s about reacting to the bigger picture rather than individual incidents,” says Young. The system also helps cut down on unnecessary work. “We now have the insights to see that a section of track is in really good condition, even if it is due to be replaced, and we can decide not to replace it immediately, which reduces the overall asset cost over its life,” adds Young.
Embracing a data-driven future for the railway
As thrilling as it has been to develop Intelligent Infrastructure, which Young says has the potential to “revolutionize the railway,” it is the response from its users that stands out most. “For me, hearing users say they want to adopt and even enhance our tools going forward is massive,” says Mason. “It’s about continuous improvement.” While Network Rail is still early on its journey, the business already has plans to extend insight’s functionality moving forward. “I am really excited for the future,” concludes Mason. “We are going to be embracing new types of data collection and start to build out a more system-level management system for railway assets.”
Find out more about Network Rail on Twitter, Facebook, and LinkedIn.
“I am really proud of how we have harnessed Microsoft technology to make the railway a safer place for our workers and our passengers.”
Jacqueline Young, Program Director, Intelligent Infrastructure, Network Rail
Follow Microsoft