Few things make engineers as proud as seeing their creations shrug off a failure and keep delivering. That's exactly how the designers and operators of Neptune Canada—the world's largest remotely operated undersea observatory—must feel. Since going live in December 2009, Neptune has weathered several insults, including a dangerous encounter with a trawler, but it has still produced a near-continuous stream of live data from over 125 instruments at depths of nearly 2400 meters, including deep-sea video cameras, sonars, seismometers, and robotic crawlers.
At the end of last year, Neptune Canada had managed to bounce back from its biggest technical troubles yet, but now it faces a budget crunch that could put it on life support as early as next month. And that's happening just as the observatory's colleagues in the U.S. Pacific Northwest seem to finally be overcoming budget constraints that held up a sister observatory.
The engineers operating and maintaining Neptune Canada attribute its endurance, in large part, to the redundancy and intelligence built into its design. The observatory consists of an 800-kilometer loop of cable that runs out from a shore station in Port Alberni, B.C., delivering data via optical fiber and DC power to six branching units. Spur lines off those branching units supply power and communications to Neptune's instruments.
No single incident validated this design more than a September electrical fault that could have shuttered the entire system for months. At 14:10 coordinated universal time on 20 September 2011, Neptune's 10-kilovolt DC power supply sensed a fault condition and automatically went off-line, taking the communications feed down with it. A technician trying to restart the system was able to bring it to just 180 volts at the system's standard operating current of 1.4 amperes, but those measurements enabled the team to pinpoint the fault. Figuring in the cable's resistance of 1 ohm per kilometer, they calculated that the fault was 90 km down the line, in the vicinity of the network's first branching unit going counterclockwise, Folger Passage, 87 meters deep on the continental shelf.
Getting a cable ship out to fix the fault would take two months and cost roughly US $1 million. But getting the rest of the network back online took just 10 hours, thanks to the loop design and the low-power switching capability built into the branching units. "The system design was not only clever but also actually worked as designed," says Benoît Pirenne, Neptune Canada's associate director for IT.
Pirenne and his team reconfigured the stricken system with remote help from Paris-based Alcatel-Lucent, which designed and built the loop. Low-power pulses from either end of the cable instructed the branching units to close off their spur lines. Further commands then closed off connections to the Folger Passage branching unit itself, which isolated the fault. Finally, the team fed power and data into and around the loop clockwise, bypassing Folger entirely.
The entire system was feeding data again, with the exception of those instruments attached to the spur at Folger Passage. They would turn on again two months later when that cable ship and a remotely operated vehicle replaced the faulty branching unit.
Ironically, about a week before Neptune Canada demonstrated the value of its loop architecture, a group at the University of Washington led by oceanographer John Delaney had installed the first pieces of another remote observatory—one that eschews the loop design that saved the Canadian system. The $76.6 million Regional Scale Nodes (RSN) project is the first major cabled observatory to move forward under the U.S. National Science Foundation's Ocean Observatories Initiative.
When completed in 2014, the RSN (originally also called Neptune) will boast eight nodes strung along 900 km of cable running out from the Oregon coast as two distinct strands. Neptune Canada's Pirenne says that design may not bounce back so elegantly from deep-sea faults. "All through [the September incident] we couldn't but think what's going to happen to our American friends," he recalls.
Undaunted, Delaney argues that the RSN design may prove to be more reliable, because the dynamically changing electrical loads experienced in loop configurations can put stress on components.
Neptune Canada was awaiting Alcatel's postmortem on Folger's faulty branching unit at press time, as well as something more crucial: a funding commitment from British Columbia's cash-strapped government. A five-year joint federal and provincial funding deal is up for renewal in March. Without it, Pirenne and his colleagues could be out of a job by this summer. He says it would be "a big waste of taxpayer dollars if the government would let us down and let a system like that rot in the water."
Delaney, by contrast, has a 25-year commitment of support from the NSF. Of course, budget exigencies could weaken the NSF's resolve. In December, the agency pulled the plug on two telescopes that had been in the works for over a decade.
This article originally appeared in print as “A Tale of Two Neptunes.”
This caption for this article was updated on 03 February 2012.