How Robot Fish Led to Better Dams

Something was killing young salmon in the dams of the Pacific Northwest. Every spring, tens of millions of them would swim through the hydroelectric dams of the Columbia River on their way to the ocean, and every year as many as 10 percent emerged bloodied or suffocated. For years, the exact cause remained elusive.

Tom Carlson, a fisheries specialist at the Pacific Northwest National Laboratory, was so frustrated at how little was known about what happens inside the the dam turbine that he invented his own solution—a robotic swimmer called the Sensor Fish. After sending the aquatic machines on hundreds of journeys, he saw what was happening to the real young salmon on perilous trips through dams. Now these robot fish are getting smaller and more sophisticated and could lead to new turbine designs that allow many more salmon to survive the trip downstream.

Carlson first built a device designed to simulate the fish experience back in the early 1990s. He packed it with angular and linear accelerometers, a pressure gauge, and gyroscopes—the idea being to learn whether fish were going through spins or back rolls, or colliding with concrete abutments or turbines as they swim through dams. The robot fish was neutrally buoyant in fresh water, allowing it to sink to the depths at which salmon swim. The gadget was outfitted with a time-delay balloon that inflates at journey’s end to lift the bot up to the surface, where Carlson and his team could find it and download its data. (Carlson covered his invention with a squishy skin and called it "Flubber" until Disney complained.)

Handmade and costing about $4000 a pop, the Sensor Fish isn’t cheap. And the construction is far from perfect—one in 10 disappear into a dam and are never seen again. "Every time we lose one, I could see someone taking my beloved Toyota Tacoma truck and driving it off into the tailrace," Carlson said.

Still, Carlson endured long enough to send hundreds of Sensor Fish down the Columbia River. By 2000, he’d accumulated a mountain of data about what happens to young salmon traveling down a dammed river. Information from Sensor Fish shows that as a young Chinook salmon wiggles past the turbine blades, the pressure outside its body drops abruptly, which affects the fish’s internal organs. Its internal air bladder suddenly expands, damaging organs and even forcing bubbles into its blood—similar to what happens when a diver gets the bends. Scientists are still not certain why this affects only some of the fish that swim through dams, but the fact that most dams have spillways and outfalls that some fish will travel through helps to keep the mortality rate down.



The wealth of data gathered from this surrogate salmon is informing the design of the new, fish-friendlier Alden turbine. Shaped like a corkscrew to smooth out the pressure, Alden’s model also wrings out a bit more electricity than the turbines now in use (called Kaplan turbines), Alden says, and as much as the newest designs in the lab, with peak efficiencies around 95 percent. The Alden turbine is designed to fit more snugly into its channel, which means fewer fish get caught in its edges. It also means the turbine captures more water, which gives it the efficiency increase. Finally, the corkscrew design eliminates the sudden pressure gradients that harm the fish swimming through standard turbines—though Alden still must build a full-scale version to verify all these claims.

Meanwhile, Sensor Fish themselves are getting better. The original measured more than 7 inches long, about the size of an average smelt. But Carlson’s latest version is half that size, with further miniaturization on the way. Carlson is working with a team of scientists on a $400,000 project to shrink the Sensor Fish to a sphere an inch across. Within two years, he expects to have a design that could be handed off to private industry for mass production.

At such minuscule dimensions, the Sensor Fish could become an implant and impart even more realistic data. Or it could be used by itself to imitate the journey of any kind of fish, through any kind of turbine or pump, anywhere in the world.

The Sensor Fish will only become more important in the next decade, as underwater power plants begin to appear where they never have before. Banks of underwater turbines are expected in waterways as diverse as Cobscook Bay in Maine, Puget Sound in Washington state, and even the East River in New York City. Their purpose is to capture the power of free-flowing water and turn it into electricity without use of a dam, but the threat to fish from spinning blades remains. Turbine developers and federal wildlife officials are eager to use the Sensor Fish to develop turbine systems that pose as little threat as possible to marine life.