NS’ senior operations leaders challenged Allran’s team to rethink autonomous inspection technology. They wanted something more efficient and less costly to operate than the car-based inspection systems offered by vendors.
“We received a charge to figure out how to do it on a locomotive – so we did,” Allran said.
Work began in the spring of 2018 to develop, test, and perfect the system. They faced the challenge by starting with what they knew.
“We knew the key components used in track geometry systems, and we knew from experience what worked – we called it railroad tough,” said Scott Hailey, track geometry engineer.
NS reached out to some defense industry vendors in search of components available on the commercial market and robust enough for the railroad environment. Two main components make up the autonomous system, Hailey said. One is a system of 2-D laser scanners designed to measure the gauge of track, or the distance between the two rails, ensuring that the rail is stable and not spreading apart. The other is an inertial system composed of gyros and accelerometers. They are attached by cables to the locomotive wheels and record track curvature and elevation, creating a 3-D model of each rail with the pitch and roll of the wheels.
These components are packaged in a ruggedized metal box mounted under the locomotive between the snow plow and the first set of wheels. A computer that powers the system is housed in the locomotive cab inside the electrical locker. Integrated into the machinery is a Global Positioning System that precisely pinpoints potential defects.
As the locomotive moves over the rails, data is analyzed and filtered for exceptions by the on-board computer. The findings are transmitted wirelessly to NS’ track-inspection office in Roanoke for analysis and verification by Allran’s team. Exceptions are forwarded to Engineering Department servers in Atlanta and then passed on to track supervisors who inspect the track and arrange needed maintenance or repairs.
Turning data into actionable information
At computers in Roanoke, members of Allran’s team created the algorithms and software programs that comprise the brains of the system. Using existing track geometry data, they created equations that enable the system to detect, in real-time, conditions that fall outside of certain track geometry parameters.
“We took data off of our geometry cars and reverse-engineered it to come up with the algorithms, or equations, that accomplish the same thing on our system,” Hailey said. To test the calculations, the team connected the system components to an NS geometry car and ran them over track, comparing results with data collected by the car’s inspection equipment.
“Ensuring that our system measured the same values as our geometry car was our golden standard,” Hailey said.
One of the biggest challenges, Hailey said, was deciding how to mount the system on a locomotive. The metal box holding the components is a cube about 2 feet wide, 10 inches tall, and 12 inches long. NS chose the Evolution model because modifications weren’t needed to fit the system underneath the locomotive.
“We took a lot of time measuring and making sure it would not interfere with normal operations of the locomotive,” Hailey said. “The idea was to take a standard locomotive and do as little as possible in changing the locomotive to be able to install the system.”
Achieving NS’ business goals
With the pilot underway, NS is building another autonomous inspection unit to install on a second locomotive. It’s cost-effective: NS engineering calculates that we can equip five to six locomotives with our in-house system at what it would cost to buy one of the car-based vendor systems. In addition, the company might expand the system’s capabilities, such as adding line-scan cameras that take video images of rail and use machine-learning technology to identify defects.
For Allran and his team, the prospects are exciting. It’s extremely rewarding, Allran said, to develop a solution that is helping NS achieve its business goals.
“We’ve got some smart, innovative employees working for us,” he said. “They understand that the company’s leadership wants to get the operating ratio down to 60% – to get expenses down and get the revenues up – and we’re trying to be proactive.
“In our own way,” he added, “the things that all of us are doing internally, like this autonomous track inspection system, go a long way to support our strategic plan.”