BNSF explains new cameras able to detect rail flaws at 70 mph at ND Capitol exhibit

Asim Ghanchi, general director of technology services for BNSF, explains how a worn rail section can affect railroad safety Tuesday, Jan. 22, at the North Dakota state Capitol. Ghanchi also had a new rail section for comparison. Tom Stromme / Bismarck Tribune
Asim Ghanchi, general director of technology services for BNSF, explains how a worn rail section can affect railroad safety Tuesday, Jan. 22, at the North Dakota state Capitol. Ghanchi also had a new rail section for comparison. Tom Stromme / Bismarck Tribune Tom Stromme / Bismarck Tribune

BISMARCK - BNSF Railway will begin equipping its fleet in March with cameras capable of snapping images at 70 mph with enough detail to show cracks in rails or missing bolts.

The rail company’s Track Health Optical Recognition program, which is coming to the end of a year-long, 3,000-mile pilot program, is just one of several uses of technology showcased by the company at the North Dakota Capitol on Tuesday, Jan. 22.

“North Dakota is a really important part of our network at BNSF,” said spokeswoman Amy McBeth. “So we’re using this opportunity to talk about the technology we have begun to leverage and continue to leverage.”

Asim Ghanchi, BNSF’s general director of technology services, said his team has been devising different ways to use the technology available to them.

“And it’s just going to continue to increase in usage,” McBeth said.

Ghanchi explained the roughly 1,200 people, including 30 data scientists, working in technology services has developed a data analysis system, using artificial intelligence and machine learning, to quickly analyze the millions of images they’re gathering on their equipment each day.

In addition to the images BNSF is taking of its infrastructure, it also has started testing photo technology to identify maintenance issues on their trains.

The Machine Vision System captures 1.5 million images per day, 3,000 images per train, Ghanchi said. Since September, cracks have been identified in 14 wheels using the images.

“That pretty much is a derailment,” he said, pointing to an image of one such cracked wheel.

Ghanchi said, with the artificial intelligence system, the company is being alerted to these cracked wheels within four to eight hours. Starting in March, company crews will start installing hardware trackside that will bring identification time under an hour. He said the goal is to eventually make it instantaneous.

In addition to identifying existing problems, Ghanchi said the artificial intelligence modeling technology also is being used to predict future maintenance.

“Imagine if you were to know while driving on the highway that 30 days from now you were going to have a major part go,” he said.

In the past, Ghanchi said sections of rail were replaced on a set timetable.

In all, McBeth said BNSF has 4,000 sensors along its tracks using thermal, acoustic, visual and force measurements to gather data. Drones and a handful of Rail Detector Vehicles, a couple of which are unmanned that drive along the tracks 24/7, have aided in more than doubling the rail company’s inspection rate.

Finally, BNSF has rolled out Positive Train Control technology on its major routes nationwide.

“This is a game changer,” McBeth said.

Federal regulators are mandating the technology be installed by 2020, so BNSF is ahead of the curve. With PTC, trains are monitored remotely for speed and location. If a switch fails, sending a train in a wrong direction, or if a train is moving too fast, an engineer is notified. And if action isn't taken within 40 seconds, the train is stopped remotely.

"It’s technology to prevent a massive incident," McBeth said.

According to rail safety data, since 2000, the national derailment rate across all rail companies is down 42 percent.

In North Dakota, some of this technology could help prevent another incident, such as the oil train derailment near Casselton in 2013.

The crash involved two BNSF trains. An eastbound oil train collided with a derailed westbound grain train. More than a dozen oil cars caught fire and exploded and about 1,400 Casselton residents were evacuated.

McBeth said the visual monitoring possibly have identified the bad axle such as the one at the root of the Casselton incident. Had PTC been available, it also might have softened the incident by signaling or stopping the second train involved.