Drilling into Arctic Ice

Haughton Crater, Devon Island, Nunavut, Canada

The Drill Hill test site for the IceBreaker drill. Testing facilities are in the tent on the left, lavatory facilities in the tent on the right. The flowers in the foreground are Arctic poppies. Credit: Henry Bortman

Drill Hill is a circular mound of impact breccia about a kilometer (0.6 miles) in diameter, one of many such mounds that litter the floor of Haughton Crater. It’s a large pile of rock rubble, initially created 39 million years ago when a massive bolide slammed into the Earth, and is now intermixed with ice.

It’s this mixture of underground ice, rock and dirt, present year-round, that makes Drill Hill a good place to test drills designed for future missions to Mars.

In 2008, NASA’s Phoenix mission, which landed in the martian polar north, scraped its way down to dirty ice lying mere inches below the surface.

This past summer, a team headed by Brian Glass, of NASA Ames Research Center (ARC) in Moffett Field, California, made the trek to Drill Hill, to test a drill under development for a proposed follow-up mission to Phoenix. That mission, known as IceBreaker, will be able to drill up to a meter (3 feet) into the ice-cemented ground below the martian surface, in search of frozen evidence for life.

IceBreaker was previously tested in the Antarctic by a team headed by Chris McKay of ARC.

During the Drill Hill test, Glass’s focus was on fine-tuning a drilling system that could reliably serve up samples from the frozen martian underground, bringing them to the surface for analysis.

“Reliably” is the key word here. When drilling into ice, it’s easy for the drill to get stuck. Drilling generates heat, which melts the ice. When the drill stops, the water refreezes, trapping the drill bit in place.

If that happens on Earth, somebody grabs another drill, or a pickaxe, and whacks away at the surrounding ice until the trapped drill bit breaks free. But there won’t be anybody with a backup drill or a pickaxe on Mars. Even mission engineers back on Earth won’t be much help. Given the limited communication between Mars and Earth, by the time they find out about the problem, it will likely be too late to do anything about it. The mission will be over. So a drilling system for Mars will have to be smart enough to avoid getting stuck, and able to free itself if it does.

When I first rode out to Drill Hill to observe the drill in action, IceBreaker was stuck in the ice.

The IceBreaker drill and its human handlers (left to right): drill-operations engineer Bolek Mellerowicz, software engineer Sarah Thompson and project PI Brian Glass. Credit: Henry Bortman

This had happened before with the CRUX drill that Glass and his colleagues tested at Drill Hill back in July 2009. Then, it hadn’t been a serious problem. Putting the drill into percussive mode – hammering as well as turning – created enough force to break the bit free from the surrounding ice. But the CRUX drill had a thicker shaft and a more powerful motor than IceBreaker.

That was good for getting it unstuck, but made it too heavy and power-hungry to be approved for a trip to Mars on a Phoenix-sized lander. The goal of Glass’s project is to develop a flight-ready drill, so the IceBreaker drill was designed to be lightweight and to be able smash and cut its way through rock and ice using less power than a 60-watt light bulb.

Unfortunately, the new design may have gone too far in the lightweight and low-power direction. After a couple of hours of furious effort, IceBreaker’s drill string – the long metal shaft that connects the motor to the bit – did finally rip free. But it left behind, entombed in the Arctic permafrost, both the bit and an expensive adapter for which there was no spare.

After that, things went more smoothly.

The first order of business was to get the drill functional again. That job fell to Bolek Mellerowicz, an engineer with Honeybee Robotics, the company that designed and built the IceBreaker drill.

Then, over the course of the next week, Glass’s team proceeded to force the drill into various other failure modes, albeit in a more-controlled manner than the original stuck-in-the-ice failure.

There were half a dozen such failures that the IceBreaker project had committed to experiencing, and recovering from, in its quest to have the drill considered robust enough for a mission to Mars. Failures like auger-binding, choking, jamming, and corkscrewing.

The differences among these failure modes are subtle, but they all amount to more or less the same thing: the drill has gotten stuck. If it’s going to survive to drill another day, it needs to get unstuck.

The wrinkle, as previously pointed out, is that if a drill gets stuck on Mars, it’s going to have to recover without human assistance. That’s an unusual requirement for a drilling operation.

“Even the ‘automated’ drills in the oil and gas industry are usually tele-operated. They’re joy-sticked, like a video game. Even if they’re on the sea floor, there’s somebody 30 miles onshore in a control room somewhere monitoring it and tele-operating it robotically,” Glass said, “in real time.” But “take the human out of the loop,” he added, “oil and gas doesn’t know how to do that. No one really does that.”

Honeybee Robotics engineer Bolek Mellerowicz reassembles the IceBreaker drill string after the drill surrendered its bit to the Arctic permafrost. Credit: Henry Bortman

So the long-term goal of IceBreaker testing on Drill Hill is two-fold. First, to observe, or in some cases to induce, various failure modes, and learn to recognize the diagnostic signals that indicate they’re occurring. Sudden spikes in power draw, for example, or changes in the drill’s rate of downward progress. And second, to try out various software algorithms for backing away from a problem before it becomes critical, “before it actually ends up breaking the drill shaft, or jamming and getting stuck, or burning out a motor,” Glass said. Before, that is, it kills a mission.

“What we are doing here is trying to break the hardware in ways that will allow us to create reliable software,” Glass said. That job, creating reliable software, software that can recover autonomously from impending disaster, falls to Sarah Thompson, a software engineer who works with Glass at ARC. “It’s only by demonstrating that it’s going to function on its own autonomously that the overall system will be deemed sufficiently mature to propose as a flight,” Glass said.

Okay, that makes sense. But you might be wondering why people need to fly thousands of miles to an uninhabited desert island in the Arctic to watch a drill fail. After all, Honeybee Robotics does have a test chamber in its headquarters in Pasadena, California, where it can simulate martian temperature and atmospheric-pressure conditions even more precisely than what’s possible on Drill Hill.

The limitation of laboratory testing, Glass explained, is that any problems that crop up are “going to be stuff we expect to happen.” In the lab, for example, the engineers know in advance what sequence of layers they’ll be drilling into. “It’s too easy,” Glass said. “We don’t put in unanticipated problems. Because if we could anticipate them, they wouldn’t be unanticipated.”

Glass is perhaps referring here to problems like getting stuck in the ice and losing a drill bit. Which, for the record, was a form of auger-binding. Just not a form of auger-binding that anyone anticipated.

This story has been translated into Spanish.