GENERATION

Creating a Potential $5 Million Revenue Gain From Hydropower Turbines Using Computational Fluid dynamics Simulations

CFD simulations helped engineers at Hydro-Québec make design changes that will generate revenue gains of up to $5 million in twelve hydropower turbines. By shifting the maximum efficiency point to raise the power output by 7.8 megawatts and by raising the weighted efficiency of the turbine by 1.6 per cent, the changes will produce revenue gains of between $200,000 and $500,000 per year for each turbine. The exact amount depends upon the grid demand. These revenue gains have already been validated in one turbine and plans are being made to implement the design changes on the other eleven turbines. These gains were made by modifying the runner at the blade trailing edge to eliminate a large eddy in the draft tube elbow that was discovered to be reducing efficiency of the turbines. The eddy was discovered and the design changes were validated by using CFD in order to provide researchers with a clear understanding of fluid flows throughout the turbine.

Hydro-Québec is a world leader in generating green energy, with over 31,400 MW of installed capacity in 1998 and ranks among North America's largest distributors of energy. It serves 3.5 million residential, commercial, institutional and industrial customers in Québec. In addition, it supplies nine municipal systems, one regional cooperative, and some fifteen electric utilities in the Northeastern United States, Ontario, and New Brunswick. Since obtaining a marketer's license from the Federal Energy Regulatory Commission, it also makes direct sales, at market prices, to American power wholesalers, including public utilities, municipalities, resellers, and large industrial consumers in the United States. Its 1998 sales totaled 161.4 TWh, with Québec markets accounting for more than 88 per cent (142.8 TWh) and sales outside Québec for nearly 11.5 per cent. The company is publicly owned with a single shareholder, the Québec government.

Less efficiency than expected
A Hydro-Québec power plant, commissioned in the early 1980's, was operating with less hydraulic efficiency than was expected from the reduced scale model tests conducted by the manufacturer and confirmed by an independent test stand. When the power plant was originally built, there was no way to determine what went wrong in the design and fabrication of the 12 identical turbines used in the plant. Because of the tremendous advances since that time in numerical simulation tools, it was decided to take another look at this problem with the goal of finding the cause and, ultimately, a solution to raise the efficiency of the turbines. CFD, the technology used in this simulation, involves the solution of the governing equations for fluid flow at thousands of discrete points on a computational grid in the flow domain. When properly validated, a CFD analysis allows engineers to determine the direction and speed of flow at any point in the flow domain. Unlike a physical model, the geometry of the CFD model can be changed quickly on the computer and re-analyzed to explore different options in project design or operation conditions.

Fluid flow simulations were conducted in the whole turbine from the water intake through the penstock, the spiral casing, the distributor, the runner, and the draft tube. To compute the flow inside a complete turbine, iterations are required to link together the components, distributor, runner, and draft tube. Velocity profiles, turbulence parameters, and pressure distributions must be transferred from one component to the other in order to assure a coherent flow through the entire turbine. For a given operating condition, the mass flow and wicket gates angle are specified to compute the distributor flow field. This gives the velocity profiles and turbulence parameters to be used as runner inlet flow conditions. The runner flow is then computed and outlet profiles are used as draft tube inlet conditions. The pressure is then computed as the flow is solved in the draft tube and is used as runner outlet boundary conditions to recalculate the flow. The same is done with pressure at the distributor outlet.

Eliminating several areas from consideration
Fluid flow in the water intake and in the penstock is responsible for the flow profile entering the spiral casing. Questions were raised about the velocity profile at the entrance of the spiral casing due to the presence of an elbow just upstream and also because the flow at the water intake itself arrives at various angles. Experimental measurements were available on a scaled model of the water intake where the shape of the reservoir had been reproduced and the inlet angle could be changed. The CFD analysis correlated well with the model measurements and did not show any anomalies, leading analysts to believe that the efficiency problem was not related to the water intake or the penstock.

The flow field was simulated in the spiral casing to check for problems. The contour of radial velocities generated by the analysis indicated that the flow at the runner inlet was uniform along the circumference. In addition, no problematic flow pattern was seen in the spiral casing. The problem was then suspected to reside in the turbine itself. As the flow was seen to be uniform in the distributor, a section of the distributor was modeled for flow simulation and loss computations. In order to investigate the flow in the runner, two hydraulic passages were measured on site. A mechanical digitizing arm was used to measure the surface data of critical parts, such as leading and trailing edges and complex surfaces of the blade.

Selecting FEA-based CFD software
Hydro-Québec researchers selected the FIDAP CFD code from Fluent Incorporated, Lebanon, New Hampshire, as one of their modeling and analysis tools. This software package uses the finite element approach and has the advantage of using non-structured grids. Non-structured grids provide considerably greater flexibility in modeling the complex and irregular geometries involved in hydropower turbines. Non-structured grids also automate the otherwise impracticably tedious process of fitting elements to the complex geometries used in complex areas such as the draft tubes. Care was taken to ensure good mesh quality, especially near the walls, which are responsible in large part for the losses. Using the iterative approach described earlier, several turbine operating conditions were computed.

The simulation in the turbine showed a large eddy in the elbow. Further analyses led researchers to conclude that this phenomenon was the main cause of the efficiency problem. The eddy arises just before the peak operating point and up to the maximum load. It is related to inappropriate flow at the runner exit. The runner-draft tube interaction is responsible for most of the losses. However, another smaller eddy was detected in the runner, near the leading edge and the crown on the pressure side of the blade.

Reducing the losses
To reduce the losses, the water flow between the runner and the draft tube had to be improved. Researchers chose to modify the runner outlet and designed a new trailing edge. A parametric study was conducted on the draft tube flow to determine the influence of the inlet flow parameters on its performance.

Several blade cuts and extensions were tried to modify the critical parameters. Once the draft tube flow was optimized, the flow in the runner was examined to reduce, if possible, runner losses at the same time. The researchers arrived at a solution that reduced hydraulic losses in the runner as well as the draft tube at the maximum efficiency point.

The efficiency measurements on the prototype, measured by an independent team, showed a significant increase in turbine efficiency at all operating conditions. The gain is about 1.5 per cent at the peak and increases as the operation moves to the maximum power point. At the maximum efficiency point, the gain is due to the improved flow both in the runner and in the draft tube elbow. In the draft tube, analysis results showed that the eddy in the elbow was gone.

In the runner, the effect of the eddy is reduced.

The maximum efficiency point is shifted to the right and gives an additional 7.8 MW to the 195MW nominal unit with 1.5 per cent more efficiency.

Effects of new design on erosion
Researchers also looked at the effects of the new design on cavitation erosion. The CFD results indicated an increase of inlet cavitation with the modified runner. Measurements using a vibratory cavitation detection method before and after modification indicate a possible increase of 30 per cent in erosion at the maximum efficiency point and an increase of 85 per cent at maximum power. This increase was not considered to be a serious problem since the original runner cavitation level was low. The increase in efficiency far more than compensates for the expected small increase in maintenance costs.

The modifications were first applied to a single Francis turbine. The modified turbine provided more than the increase in efficiency that was predicted by the analysis. In less than a year of operation, this improved efficiency has already paid for the $200,000 cost of the modifications. Hydro-Québec management is currently making plans to implement the modifications on the other eleven turbines for this power plant. Planning is also in progress to use CFD to improve turbine efficiency at other power plants.

This application provides an excellent illustration of how CFD simulations can identify hydropower problems and help develop alternatives to improve machine performance.