On the USNS Apache in late October 2015, the CURV-21 proved a clumsy tool for searching in the sunless deep three miles down for something the size and shape of a coffee can. In short time, the CURV’s navigation system stopped working entirely. Its narrow cone of visibility, further limited by the short range of the vehicle’s spotlight, rendered the search time-consuming, and ultimately, in vain.
Without navigation, the techs had to rely on dead reckoning to maintain their bearings. Using the main wreckage as a starting point, they drove the CURV in a radial pattern, always returning to the half-buried hull to reconfirm their orientation. As the vehicle crept through the city of junk, it used two sonars—high and low frequency for high and low resolution—to scan for anything metallic. There was a lot of metal down there, and most of it was shipping containers. Everything that the CURV detected had to be investigated.
Obstacles would suddenly emerge from the darkness, blocking their way, demanding quick course changes to avoid a collision.
The team often got disoriented in the murky dark. They’d drive the CURV far out from the main wreckage, or so they thought, and then come back upon it sooner than expected. It was clear to Tom Roth-Roffy that they didn’t know where they were. His worst fear was that they could breeze right by the VDR without knowing it. The whole endeavor seemed futile. There had to be a better way.
After days at sea, the waves began to kick up, rocking the Apache enough to make people feel ill. Morale aboard dropped. The tech team wanted to go home. Debilitated by seasickness and dopey from Dramamine, Tom refused to quit and called the NTSB headquarters to tell his superiors as much. Headquarters agreed to enlist the help of their own research and engineering specialist—Dennis Kryer—in the search. He had a gift for finding things.
Kryer asked the Apache’s tech team a lot of questions. He wanted the mass of the wreckage so that he could build a computer model of how the ship went down. He carefully studied the splash El Faro had created on the seafloor to determine its speed and direction when it hit, and he calculated the buoyancy characteristics of the hull and the house to predict their drift pattern. He used this information to come up with coordinates for the VDR’s location. “As it turns out,” Tom says, “he was pretty close.”
Kryer recommended that they spend another few days extending the search much farther to the north of the hull. After a quick trip to Puerto Rico to refuel, the Apache returned to El Faro’s position, took one last look, and finally located the bridge 450 meters off the port bow.
If the VDR was still attached, that’s where it would be. But the NTSB’s $1 million contract with the navy was up. The Apache team carefully hoisted the CURV on its three-mile-long fiber-optic umbilical cord back onto the deck and headed back to Virginia.
Had Captain Davidson been pressured by TOTE? Was he missing critical weather information? Was he suicidal? Without the VDR, investigators could only speculate.
But that wasn’t Tom Roth-Roffy’s style. If there was a way to get the VDR, even if it took extreme measures, the NTSB would do it.
With the support of Congress, the NTSB received an additional $1.5 million to pursue a second mission in the Caribbean in April 2016.
Tom Roth-Roffy stayed behind, leaving the job to NTSB’s Eric Stolzenberg.
Stolzenberg began his career as an engineer in the merchant marine. He shipped around the world on supertankers and eventually got a land-based job as a naval architect—he’d studied the field as an undergrad at New York Maritime College—before joining the NTSB in 2008. Now in his early forties, he was generally younger and brasher than his NTSB counterparts. Tall with a head of thick brown hair, Eric played Will Smith to Tom’s straight-shooting Tommy Lee Jones.
Stolzenberg watched the El Faro investigation from the sidelines for a couple of weeks, eager to jump in. He loved solving mysteries that required piecing together obscure clues. He was also drawn to the incident because, he told me, hurricanes fascinated him. He first got hooked in 1985 when Hurricane Gloria made a direct hit to Long Island, where he’d spent summers with his family.
When the wreck of El Faro was located in November, Stolzenberg was designated the lead on the NTSB’s naval architecture team. He immediately began acquiring any extant El Faro blueprints, as well as her loading diagrams and photos. He was sure that El Faro’s design played a major role in her undoing.
Stolzenberg’s quest was to determine the sea conditions in which El Faro sank and use that information to create a dynamic model of the events leading up to her demise. He stepped onto El Yunque in December and homed in on her downflooding points. “A ship has an enclosed buoyancy,” Stolzenberg tells me. “When all the hatches are watertight, a ship remains buoyant. You cannot sink it. Physics dictate it can’t go to the bottom. You’ve got a balloon in there keeping it afloat. Poke enough holes in that balloon and the inevitable occurs.”
On El Yunque, he could see how those vents, at a certain angle of heel, would get submerged.
One big question: Did this design meet current codes, and if not, why?
When El Faro was converted from roll on/roll off to a container ship in 2006, the regulators—the US Coast Guard and ABS—acquiesced to TOTE’s repeated requests that the work not be deemed a major conversion. Accordingly, the regulators did not redo El Faro’s damage stability index. “And that might be an oversight,” Stolzenberg says.
A damage stability analysis determines how many holes you need to poke into a ship to make it sink. The calculation is based on probability—where a ship is most likely to get hit—and dynamic modeling.
Stolzenberg says that determining a ship’s damage stability is both science and art. But at least it gives you a sense of a given vessel’s vulnerabilities. This would be critical information for a ship’s captain when he’s thinking about heading into a large storm.
Just like John Glanfield, Stolzenberg saw the vents for what they were: ways for water to compromise El Faro’s buoyancy. The ship’s designers saw that, too, and equipped her with fire dampers—big steel louvers that could be closed to snuff out a fire. The dampers weren’t weathertight, but if closed, could slow the ingress of seawater during a storm. Crew on El Faro regularly tested the operation of these dampers during fire drills but never considered using them to protect the vessel in high seas.
If a loose car floating in three-hold had hit the fire pump causing a breach in the hull, the ship would have flooded very quickly.
Stolzenberg thought a lot about these various scenarios and wanted to get a closer look at the wreck. He also wanted that VDR.
In April 2016, he boarded the R/V Atlantis, an oceanographic research vessel operated by the Woods Hole Oceanographic Institution, owned by the US Navy. Two undersea vehicles they carried with them would work in tandem to document the accident site and, with luck, locate the VDR. The automated underwater vehicle (AUV) was a tethered machine with broadband sonar that ran along a preprogrammed path, identifying anything that approximately fit the size and shape of the ship’s mast. Its twenty thousand feet of fiber-optic cable alone probably cost $1 million.
The SENTRY, another autonomous vehicle, was equipped with searchlights and high-resolution cameras. It followed behind the AUV and was programmed to investigate each target the AUV identified. Unfortunately, the AUV’s margin of error was thirty meters, so often it took a while, sometimes as much as forty-five minutes, for the SENTRY to find one target.
The AUV and the SENTRY could never be in the same area or they risked crashing into each other, so sometimes the tech team sent the SENTRY on a wild goose chase far away from the scene while the sonar worked closer to probable sites.
As before, the team aboard divvied up the watches, monitoring the computer screens in the nerve center to see if they could detect the mast. But they were stumbling in the dark.
The techs from Woods Hole Oceanographic Institution who ran the robotic equipment sat down with Eric as they were heading out to the accident site. They said, Remember: At fifteen thousand feet, there’s no guarantee you’re coming here tomorrow or the next day. Things go wrong down there. Every photograph, every dive, every view could be your last, so treat it as such.
During the first couple of days, Stolzenberg got a good taste of what they were talking about. Fittings on the vehicles’ casings leaked, small things went wrong, and Stolzenberg quickly learned that, Yeah, this could be the last time we see things. The autonomous vehicle took two hours to sink down fifteen thousand feet, and two hours to surface, which meant four hours of waiting each time they brought it up to upload and process its data. Time gnawed away at everyone’s patience. They only had six days over the wreck to find the VDR, and everything was an exercise in frustration.
The sonar picked up countless targets—automobiles, shipping boxes, and car batteries—resting quietly on the seafloor. It identified personal effects, standard household goods, bicycles, toiletry items, dolls, and roof flashing, all eerily still in the deep. The mast was big, about as big as a container. All the ship’s containers had fallen off as it went down and were now showing up on the sonar as nice, bright targets. There was an impossible amount of stuff to go through.
While the SENTRY investigated each of these targets, small, strange-looking deep-sea creatures wandered in front of its high-resolution lens. Three miles down, it was another world, now littered with the detritus of terra firma.
Wreckage distribution told Eric a lot about how El Faro fell through the water. From the wheelhouse north was mostly cargo; from the wheelhouse south to the hull was ship debris. The first mission identified where the house was, so, he concluded, the mast must be somewhere between it and the hull.
With four days left, hundreds of possible trails to explore, and two hundred identified targets to investigate, Stolzenberg threw a Hail Mary. They had to limit their search to the area between the hull and the house.
As soon as they committed to Stolzenberg’s plan, the trail got hot. They found railing from the wheelhouse, the ship’s stack, and a piece of steel with a porthole, all of which told them that they were in the right place.
They picked through the trail of debris, investigating everything they encountered. That night, when his shift was up, Stolzenberg refused to leave his post in the tiny control room. He’d been up for twenty-four hours but couldn’t pull himself away from the monitors. He could see maybe three or four meters in each direction, and a lot of “snow”—tiny particles that got caught in the spotlight—but he knew they were getting close.
A few minutes into the next watch, the SENTRY missed yet another target and had to swivel and go back. This time when it turned around, one of Stolzenberg’s team members saw a flash of light to its right. What the hell was that?
They steadied the robot and slowly approached the area from which they’d seen the flash. “Then,” according to Stolzenberg, “we were like, holy moly. That’s the reflective tape on the VDR!”
The superreflective tape wrapped around the VDR canister was no wider than two inches, but it bounced the high-intensity lights of the underwater search vehicle back at its camera. Just beyond the VDR, Stolzenberg could make out the three legs of the mast coming at right angles to the SENTRY; the rest was buried in mud.
“I guarantee you that on the first Apache mission, Tom went right by this thing,” Stolzenberg says. “You have to be so close to see anything down there with that much junk around. You almost have to be on top of the stuff.”
The tech team didn’t want to get the SENTRY’s tether tangled up in the mast, so they approached with caution and documented the scene. At 2:00 a.m., when they were certain that they’d found the VDR, they sent an email back to the NTSB.
The next day, they deployed the autonomous vehicle with its high-resolution lens to document the entire site. During that mission, the SENTRY took more than forty thousand photos, which the NTSB later patched together to create a map of the scene.
Meanwhile, they had to figure out how to extricate the VDR from the mast’s clutches. The steel canister was held to the mast with two quick-release bands. Although it sat proud, you couldn’t just scoop it up. You’d need a highly skilled robot to spring it free, drop it in a basket, and carry it three miles to the surface.
Stolzenberg didn’t have that kind of robot with him on the Atlantis. He could only stare at his prize, sitting undisturbed on the ocean floor in the watery deep.