Carnegie Mellon University researchers are aiming to make gas pipelines safer, by designing a modular robot that can crawl through to map, analyze, and even repair the pipework.
"We don't even know where all the old pipes are," Howie Choset, professor of computer science at Carnegie Mellon and director of the project, says of the problem of tracking the estimated 1.2 million miles of distribution and 900,000 miles of service lines which lead to 75 million homes and five million commercial users in the US alone. That's a problem, but there's an even bigger one: many of the pipes are old and in danger of leakage.
Digging up all the pipework to investigate whether it needs to be replaced isn't an option, so the Advanced Research Projects Agency - Energy (ARPA-E) is funding projects that aim to both detect potential failure and repair it in-place: the Rapid Encapsulation of Pipelines Avoiding Intensive Replacement (REPAIR) program, a somewhat tortured backronym.
For the REPAIR program, CMU researchers have developed a robot capable of traveling through the pipes themselves in order to map where they lead without having to dig them up. As it maps the pipework, it also inspects its walls for damage using a precise laser sensor — running at around nine miles of pipe for every eight hours, the researchers say. These data are evaluated by machine learning for areas of concern, while engineers on site can use augmented reality to peer below the ground without digging.
If damage is detected, it can be marked on a map — but it can also be repaired in-place. Each robot can haul around 60 pounds of payload, using two motorized wheels on the underside and one stabilizing wheel on the top, including a resin which can be applied in a spiral to repair and reinforce the pipe's walls in-situ.
"I think the resin is the real star of the show," Choset says of the material, which was developed at the University of Illinois. Initially, the material is the consistency of soft-serve ice cream — but once applied to the pipe wall it hardens in seconds.
The concept has already been proven in a test-bed pipe system, with a 200-foot range the team hopes to extend to 6,500 feet in future designs — along with a shrunken version suitable for pipes smaller than 12 inches. "We believe we can also reduce the size to a 2-inch diameter in the near future," research scientist Lu Li says, "[but] smaller is harder."
More information on the project is available on the CMU School of Computer Science website.