In 1837, textile factories in Augusta, Maine built a dam across the Kennebec River for hydromechanical power. In 1913, it was converted to hydro electric power.
In 1997, the Federal Energy Regulatory Commission (FERC) compared the dam's 3.5 megawatt production to the blocked passage for spawning Atlantic salmon, shad, herring, bass, sturgeon, smelt, and eels. FERC ruled in favor of the fish, and on July 1, 1999, the state of Maine destroyed the 917-ft long, 25-ft high Edwards Dam.
Engineers are using software to simulate a fish's ride thorugh a spinning turbine.
Faced with an environmentally friendly government, hydropower companies across the country are looking for ways to allow migrating fish to move safely up and down rivers. One of these companies is using powerful computational fluid dynamic software to create computer models of the roller coaster ride a fish follows through a spinning turbine.
"We started our project to improve environmental compatibility of hydroelectric power to blunt criticism from environmentalists and federal licensing bureaus," says Richard K. Fisher Jr., VP of Technology at Voith Siemens Hydro Power Generation Inc. "Before we started, we were having a tough time understanding what really went on inside a turbine with regard to pressures and loads on fish."
The CFD models have provided a new way for the company to experiment with turbine designs, and provide a gentler ride for migrating fish.
Fish face several possible inuries as they swim through turbines.
For a typical, large-river Kaplan turbine, the survival rate for fish passing downstream through a dam is 88-92%, according to Voith Siemens Hydro. And an additional 3-5% may be injured, becoming vulnerable to predation. But with the results of their powerful computer models, Voith Siemens Hydro claims to have reduced fish injury at a test dam in Oregon, using a new turbine runner design called the Efishent! Kaplan Model to achieve a 40% reduction in both injury and mortality.
The project began when Voith Siemens Hydro won a grant from the U.S. Dept. of Energy, and began to work with computer scientists at Georgia Tech. Out of the relationship came the idea to create a program called the Virtual Fish. Now they can launch this virtual fish into the CFD model of water flow, and measure its motion and the forces acting on it.
"We simplify the fish shape to a 100- to 150-facet ellipsoid," he explains. "The program outputs data on the time history and loadings of the fish as it moves through the turbine components. Then we modify the design, seeking reduced loadings."
They quickly learned that the shear, or velocity gradient, can be even more dangerous than running into walls or getting sliced by turbine blades. "We look at the velocity change over the length of a fish, since loads can pop its gills back, tear membranes, or bend the fish to the point of injury," Fisher says.
Other dangers include swirling vortices in the water, which can cause a fish to get dizzy and disoriented, becoming easy prey for predators downstream.
When you add up all the variables, the Voith Siemens Hydro engineers have an enormous amount of computing work to do. In fact, computing time is a problem-the team allows 1-3 hours run time for the 250,000 nodes for the turbine blade analysis, and 10-20 hours for the 1 million nodes for the water intake. Then the draft tube, downstream, takes another 5-10 hours.
"The real situation is, you can analyze the entire turbine and use lots of nodes to get a good solution, and you can compute for months with a high power super computer to get a solution for just one point. But we are in the business of finding engineering solutions now, so we have to keep our computational models simple enough to keep the calculation time under 10 hours," Fisher says. It usually takes them 1-3 hours to calculate each operating point. "We're 80-90% correct for fish position, as long as the fish doesn't get an inclination to swim," he says.
"The virtual fish has no free will. It's like an inert, wooden ellipsoid, being carried by the water. But we can determine forces acting on it. One of our dreams is to program some virtual life into it, so it can move in response to those forces, pressures, and accelerations. But we have enough challenge reducing loads before we do that."
The team uses CFX-TASCflow CFD software from AEA Technology (Waterloo, ONT) to calculate time-averaged viscous 3D flow. Then they launch the Virtual Fish model into that grid. And finally, they import the computational data into EnSight, a program for analyzing and animating CFD problems, from Computational Engineering International (Morrisville, NC). The program makes the data viewable in animated, stereo 3D simulations, as users wear stereoscopic goggles. And EnSight produces a grid of where the fish were released, plotting their accelerations and impacts as they flow through the dam.
"We make post-processor visualization software, and animating can tell you things you didn't know about the model," says CEI President Kent Misegades. One of the challenges of the program was to correct for the fact that fish are heavier than water. "Just like a bug flies into your windshield, fish have a certain amount of momentum, so they don't follow the exact flow of water," he says.
What do you do with all this data? And does it really help the fish?
The Bonneville Dam on the Columbia River, 40 miles east of Portland, OR, may hold the answer. At 197 feet high and 2,690 feet long, it dwarfs the smaller dams in Maine, which use 48" diameter turbines. A typical Kaplan turbine in this dam is 280" in diameter, spinning its five blades at 75 rpm, and discharging 13,600 cfs. A discharge rate of 16,000 cfs creates 60,000 hp, which is about 44,760 mW.
Those velocity gradients are no joke when you see the power in a spinning turbine. The big Kaplan turbines there are made from CA6NM, a cast stainless steel with 13% chrome and 4% nickel. The blades also rotate within the hub, changing their angle from 15-30 deg. The Columbia River Kaplans deal with 80-120 feet of "head" in the altitude of the upstream water above the river, which creates a pressure range up to 3-5 atm.
How much can engineers change turbine design when they're dealing with these enormous pressures? Using advanced CFD, Voith Siemens Hydro engineers have created the Minimum Gap Kaplan Turbine Runner (MGR), which shrinks the gap between turbine blade and wall as small as possible, minimizing turbulence and velocity gradients. This is also known as the Efishent(TM)Kaplan Model.
"Passing through the turbine is like a revolving door," Fisher says. "You'd rather go through a slow, big one than a fast, small one. Especially if you're running in the dark!"
At Bonneville, one of the dam's 10 turbines has been replaced with an MGR, and early results are good. (see attached charts) But work still continues as engineers struggle to improve the computer model.
One of the next steps is to better calibrate the Virtual Fish to real animals. Biologists pass live fish through a turbine with balloons attached, so they can locate and inspect the animals afterwords. Hungry gulls often snatch the dazed, exhausted, floating fish before researchers can reach them, so they also have to scare off or shoot the gulls.
Another way to calibrate is a new "crash dummy fish" from Pacific Northwest National Laboratory (Richland, WA), a 6-inch rubber instrumented fish model with onboard pressure transducer, accelerometer, computer, and a battery, all with attached balloon tag.
In the meantime, politicians and environmentalists will continue to fight about the value of dams. Industry figures say that conventional hydroelectric power creates about 10% of US electricity, and about 20% of worldwide electricity. It's up to CFD modelers to balance the country's energy needs with the environmental impact of dams.
Other applications for this technology
Aside from the spinning blades themselves, another danger for fish around dams is that the water downstream can be oxygen-depleted, creating more predator danger even for fish who haven't been through the turbines. Certain reservoirs develop sharply stratified water layers in the summer, and dams often draw this bottom-level water into the turbine intakes.
So Voith Siemens Hydro and the Tennessee Valley Authority are using the pre-computed CFD flow field to apply a software code developed by Georgia Tech called "Virtual Bubbles," Fisher says. In a similar manner as Virtual Fish, the virtual bubbles are released into a simulated flow stream as a post processor, to calculate variables such as: bubble motion, gas transfer across the air-water membrane, bubble break-up, and pressure effects on bubble size.
The software is used to develop turbines to "aerate" the water to increase dissolved oxygen levels in the dam's discharge, so fish living downstream have a better habitat.
Lessons learned from the experience
"The development of the Virtual Fish code took place in an industry sponsor/university development partnership environment, so both students and professors worked with Voith Siemens Hydro to develop and refine the concepts and software," Fisher says. "If we were to do this again, I would require that the Industry-that's us-get more closely involved in the code development and in the verification of the code, and that we have an engineer as an active part of the team. He would provide continuity across the time of development and be responsible for the product's integrity."