Sensors - July 2004 - The Sensor Fish—Making Dams More Salmon-Friendly

An instrumented package traveling with real fish through hydroelectric dams collects data on the often hazardous conditions that migrating salmon smolt encounter.

Thomas J. Carlson, Joanne P. Duncan, and Theresa L. Gilbride
Pacific Northwest National Laboratory

Young endangered salmon on their migration from upriver spawning grounds on the Snake and Columbia rivers may pass through up to 12 federal hydroelectric dams as they journey to the Pacific Ocean. A better understanding of what the fish experience when they disappear into the dark, swirling waters of the dams’ turbine bays and spillways might help dam operators such as the U.S. Army Corps of Engineers design passage routes that are more friendly to salmon. Biologists wanting a "fish-eye" view of the trip haven’t figured out how to shrink themselves for the ride but they may have designed the next best thing. Researchers at the Pacific Northwest National Laboratory, with support from the U.S. Department of Energy's Hydropower Program, have built the "Sensor Fish," an autonomous package incorporating a pressure sensor and triaxial accelerometers. Triaxial rate gyros will be added later this year. The device travels through the dams, measuring the turbulence, impacts, and other forces experienced by real migrating smolts.

The Sensor Fish device currently in use, a 3-degree-of-freedom (3 DOF) sensor system, has modules that charge its internal battery, download data, and reprogram the sensors to perform specific data acquisition tasks. It also includes a rechargeable battery and analog and digital elements to filter, digitally sample, and store sensor output and communicate via a wireless IR link with an external IR modem for data transfer and for programming. When the rate gyros are added, the Sensor Fish will become a true 6 DOF device.

The Sensor Fish has taken many configurations since PNNL began developing the rubbery fish-shaped sensor package for DOE in 1997 (see Figure 1).

Figure 1. The first Sensor Fish contained a pressure transducer, accelerometers, strain gauges, and batteries, encased in a molded dielectric polymer body roughly the same size as a yearling Chinook salmon smolt (150 mm length). Unfortunately, its weight was ~160 g, more than 4 × that of an actual smolt and its negative buoyancy complicated retrieval.

The current version (see Figure 2), a 25 × 90 mm, 45 g polycarbonate cylinder rounded at both ends, has ~¹/₃ the volume and ¹/₄ the weight of the original.

Figure 2. In the current version of the Sensor Fish, the gray square on the end of the cylinder is the pressure sensor. The device also contains triaxial accelerometers and rate gyros. This sensor package will be sent through the turbines, spillways, and sluiceways of federal hydroelectric dams along the Columbia and Snake rivers of the Pacific Northwest to measure conditions experienced by migrating juvenile salmon.

It weighs roughly the same as a yearling salmon smolt and, like a fish, is nearly neutrally buoyant in water—it neither sinks to the bottom nor floats to the top unless pushed there by moving water. It is small enough to be implanted in or externally attached to an adult salmon if adult salmon passage data are of interest.

The Sensor Fish are typically deployed as an element of live fish studies to evaluate the biological performance of bypass alternatives at dams. After they are equipped with radio transmitters and balloon tags for retrieval in the tailrace (see Figure 3), Sensor Fish are injected along with live fish through a pipe into the bypass.

Figure 3. The Sensor Fish are tagged with balloons and a radio transmitter, sent through an operating hydroelectric dam via a release pipe, and retrieved in the tailrace. Pressure and acceleration data are downloaded to a database for analysis. Sensor Fish and live fish are usually tested simultaneously.

Their journey through the dam typically takes from a few seconds to a couple of minutes. The balloons are filled with a chemical that reacts with water and causes them to inflate and bring the package to the surface within a few minutes after its ride through the dam. The Sensor Fish has a data capacity of several minutes, depending upon selection of data acquisition parameters. In the 3 DOF Sensor Fish, the output of the multiple sensors is digitally sampled at 200 Hz. The 2 kHz sampling frequency of the 6 DOF Sensor Fish will more accurately capture the rapid accelerations and motions that occur during passage through the hydroelectric turbines.

Before the Sensor Fish was developed, biologists and dam operators relied primarily on physical models to characterize spillway and turbine passage environments. Using data collected by the Sensor Fish, researchers can better understand what conditions may be responsible for different types of injuries to migrating fish and where these conditions occur during passage. With its ability to measure all six degrees of freedom (see Figure 4), the 6 DOF device, like its 3 DOF precursor, will help identify injury mechanisms for fish such as strike, shear, and inertial effects, including nonlethal ones such as stunning or signs of vestibular disruption that expose fish to a higher risk of predation by birds and piscivorous fish downstream following passage.

Figure 4. The Sensor Fish’s rate gyros, accelerometers, and pressure transducer enable it to measure six directions of a fish’s motions in water: the three components of linear acceleration (up-down, forward-back, and side-to-side) plus pitch, roll, and yaw.

Simultaneous deployment with live fish allows the Sensor Fish to provide a physical history, exposure magnitude, and frequency/rate context which, when linked to injury data, researchers can use to identify injury mechanisms.

Sensor Fish studies at McNary Dam in Umatilla, OR, confirmed details about hydraulic conditions immediately above a turbine runner. In particular, it was discovered that the flow of the water approaching the runner was not uniform, but rather a “surging” flow with a number of distinctive acceleration and deceleration cycles (see Figure 5).

Figure 5. Simulated pressure distribution for the runner region of a Kaplan turbine shows a decrease in pressure from the leading edge to the trailing edge of the blade on the upper face of the runner blades and a low-pressure region below the blade. (Simulation and data by Sulzer Hydro, Ltd., Escher Wyss. Visualizations and images by ETH Zurich, SCSC.)

This surging is propagated downstream through the draft tube and into the powerhouse tailrace. Some of the injuries to fish attributed to shear might instead be associated with the differential between the velocity of a fish and of the surrounding fluid under these high “jerk” conditions. (Jerk is a measure of the rate of change of acceleration.) Sensor Fish have measured instances of jerk in fish passage environments as high as 15,000 m/s³ (50,000 fps³).

The Sensor Fish’s pressure transducer output is used in several ways, one of which is as a means of estimating the location of the sensor at specific times during passage. For example, during passage through a hydroturbine runner, a considerable change in pressure takes place within a very short period of time. This rapidly decreasing pressure produces a distinctive negative-going pressure spike that is used as a timing mark to aid processing of Sensor Fish data.

Sensor Fish are currently being used at Ice Harbor Dam (see Figure 6), a Corps of Engineers’ hydroelectric project on the Snake River just above where it enters the Columbia River in southeastern Washington state.

Figure 6. The Sensor Fish makes a 5 s trip down the Ice Harbor dam spillway (A), beginning with passage under the spill tainter gate (1), continuing with the trip down the spill chute (2), through an impact with/passage over the deflector (3), and finally passage through the spill basin (4). In (B), the pressure measurements are blue and the acceleration data are red. The increase in pressure at (3) is not a result of increase in sensor depth (the water was <1 ft. deep here), but rather results from stress in the fluid as it is forced to change direction when passing over the deflector.

To improve juvenile salmon survival during the spring outmigration, dam operators have been increasing the amount of water passed over the dam’s spillways instead of through the turbines. However, at high rates of spill (50%–100% of total flow), large volumes of air are entrained in the water at the bottom of the spillway, causing potentially lethal dissolved gas bubble disease (“the bends”) in the fish. To minimize the air entrainment, the Corps installed a deflector at the bottom of the spillway. Juvenile salmon mortality rates did not decrease so much as anticipa- ted, and Sensor Fish were sent over the spillway to find out why.

The test condition results shown in Figure 7 indicate that due to shallowness, the magnitude of exposure per unit time is higher over the deflector than it is either above the deflector in the chute or below the deflector in the stilling basin regions.

Figure 7. Twelve conditions were tested at the Ice Harbor dam. Pressure and acceleration time histories were separated into three segments: (1)“chute,” extending from the time the sensor passed under the spill gate until it encountered the spill deflector; (2) “deflector,” when the sensor passed over the deflector; and (3) “basin,” from the deflector out to the spill tailrace. Two flow conditions were tested: 50% and 100% of total spill. The Sensor Fish were released in shallow and deep water. It can be seen from exposure magnitude per unit time plotted against exposure duration that the spill deflector region is unique in that the per-unit-time magnitude of exposure is higher than for the chute and basin regions, which are quite similar. This is significant because studies of the biological performance of spillways have shown higher fish mortalities at spillways with deflectors than at spillways without them. Deflectored spill appears to be more hazardous because conditions over the deflector are more severe than elsewhere along the spill passage route. It is not clear at this time whether deflectors can be designed that will be less hazardous for fish while providing total gas supersaturation reduction benefits.

In other words, the extreme conditions around the deflector make that area the most hazardous part of the trip for juvenile salmon passing Ice Harbor Dam via the spillway.

Conclusions
The Sensor Fish has been successfully deployed in a number of fish passage environments at hydroelectric dams throughout the Columbia River Basin. The data have helped identify the location and duration of exposure conditions hazardous to fish. Continued improvements in Sensor Fish performance and in tools for processing and analyzing the acquired data promise a better understanding of these undesirable conditions. The Corps’ and private utilities’ experience with operating dams on the Columbia and Snake rivers has shown that changes in dam operations can often result in improved conditions for fish passage. Researchers are therefore hopeful that the insights they gain from Sensor Fish will lead to good news for real fish swimming through the dams.